What's new

What changed in Omaya, newest first. The same notes are in the platform: open About and choose What's new.

2026-10 · Trip speed

Date: 2026-10-23 Theme: Trips outside can show how fast they went. Administrator action: optional. Turn it on under System → Config → Outdoor & GPS → Speed; it is off by default.

Average and top speed on trips: the Sensor Trail, Outdoor Activity and GPS Replay show a trip's average speed (distance over time) and its top speed (the fastest two minutes). Speed is worked out from the GPS positions, so it suits telling driving from walking rather than speeding checks.

2026-10 · Fall detection: wear it right, and a report that stays readable

Date: 2026-10-23 Theme: Fall detection only works on a tag that is worn; the platform now says where to wear it, notices tags that are not being worn, and measures accuracy by position. Administrator action: optional. Check System → Config → Fall detection for tags that look not worn, and set where each tag is worn.

Wear it right: a short guide on the Fall detection settings and the Fall & Man-down report — on a lanyard at the chest or clipped at the waist, not on the wrist if it can be helped, never in a bag or pocket — with the accuracy measured for each position. The person's "Are you OK?" page carries a one-line tip.
Where each tag is worn: lanyard, waist clip, wrist, or carried in a bag or pocket — not worn, which switches fall detection off for that tag.
Looks not worn: Omaya notices a tag that lies still for hours at a stretch on most working days and says so, with the reason, on the settings, in the report, on the person's Entity Tracker panel and in the OAIA briefing.
Accuracy by position: the report compares your labelled events per position with the published figures, and leaves events from tags that looked not worn out of precision unless you count them in.
Events by day: the report's event list is grouped by day with each day's counts, opens on the latest three days, and filters instantly by would alert, not labelled, falls, man-down or a name.

2026-10 · Connector pages, platform administration and translations

Date: 2026-10-22 Theme: Connector pages and the platform administration pages get the same clear layout as the rest of the console, with more languages. Administrator action: none.

Connectors: every connector page shows whether it is getting through right now, with a one-click Test, and keeps its settings, recent activity and setup steps on separate tabs. Secrets stay hidden until you choose to show them. Matter bridge and S3 connectors can now be deleted from their page.
Languages: the Security Watch settings, Tag Power and Auto Discovery pages, the in-page dialogs and this month's outdoor features are available in Bahasa Melayu and Simplified Chinese.
Security: access checks tightened across the admin console.
Platform administration (platform operators only): a new layout for every page; the System Dashboard opens with Needs a look across all organisations, worst first, and an Outside & GPS tab (counts only); each organisation has its own page; a Superuser activity timeline records privileged actions; every page filters by organisation and opens it in one click; the Scheduler explains failed runs in plain words and runs any task on demand.

2026-10 · Clearer setup pages: Tag Power, Auto Discovery and three app settings

Date: 2026-10-21 Theme: Five setup pages that still read like engineering tools now explain themselves. Administrator action: none.

Tag Power (Device → Utilities) shows which tag models still need measuring, explains each correction in plain words ("reads 4 dB quieter than the reference — corrected"), and keeps the technical detail one click away. ?status=needs opens the list of models to measure.
Auto Discovery has summary tiles, the rules in plain words, a one-click way to turn discovery on, and an Add button on each found device.
Toilet Hygiene, Facility Management and Security Watch settings have a summary at the top, tabs you can link to (?tab=), and a plain explanation in each section. Toilet Hygiene puts toilets that still need a profile first; Security Watch keeps its on/off switch at the top, shows gateway coverage in words, and adds the missing switch for signals from other apps.
All five work on phones and in dark mode.

2026-10 · Ask OAIA follows role access; TV app widgets follow the board

Date: 2026-10-21 Theme: Two gaps closed: what Ask OAIA can tell a role, and what a place-limited TV board shows. Administrator action: none.

Ask OAIA respects each role's label access — in the web chat, the mobile app and Telegram. People, assets and tags outside a role's labels are left out of answers, lists and lookups, and asking about one gets the same answer as an unknown name. Totals are unchanged — counts per zone, and attendance and asset-utilisation totals, match the report pages; only the names listed follow the role.
TV boards for a site, floor or room now keep every app widget to that place: Energy (only that place's meters, netted among themselves), Data Centre, Warehouse flow and yard, Compliance and Roll Call show only that place's meters, halls, warehouse areas, checks and people. A place with nothing to show says so, instead of showing the whole organisation. The Energy headline now also shows when a single meter is counted.

2026-10 · Outside: privacy, TV boards and a scheduled report

Date: 2026-10-21 Theme: More control over what is shown about people outside, and outdoor time in your scheduled reports. Administrator action: optional. Review System → Config → Outdoor & GPS → Privacy, and Visitor's new Track badges outside the building option (on where you have GPS; if your organisation uses its own visitor terms, add the outdoor wording there).

Homes on maps and replays (System → Config → Outdoor & GPS → Privacy, hidden by default): the Entity Tracker, its Movement Replay and GPS Replay show a person's time at home as At home, with no marker, route or position near it. Back-to-back time at home reads as one stay. Assets and vehicles are always shown.
Outdoor Activity by email (System → Schedule Report, where phones or trackers report GPS): per person or tag and day, time outside, distance, trips, stops, geofences visited, the longest stop and first out / last back, with time at home counted separately and never located. The monthly site pack gets an Outside section.
Role entity access now covers OAIA (findings, counts, notifications and the briefing) and the Activity Log, including geofence entries and exits: a role limited to some labels sees only the people and assets carrying them.
TV boards: someone who steps outside leaves the live map and Who is where at once. In a multi-site organisation a board can cover one site (Where from → Some sites), and every widget then covers all of its floors. The Outside widget counts the people last inside the board's site, floor or zone, and can be limited to chosen labels — counts only.
Visitor: a Track badges outside the building option (Registration & privacy, offered with GPS). When on, a badge seen outside reads Outside, hosts can be told when a visitor leaves with the badge, and the default privacy notice says so. On by default where your organisation has GPS; switch it off to keep badges indoor-only.
Dashboard: Cancel in Edit Layout puts the dashboard back exactly as it was, including widgets added, removed, moved or resized.

2026-10 · Outside, across the platform

Date: 2026-10-21 Theme: People and assets outside on GPS now read the same way everywhere — dashboards, monitors, dispatch, visitors, rules and the AI assistant — with home locations kept private. Administrator action: optional. Add the Outside Now widget to a dashboard or TV board, set up the new Outside (GPS) rule conditions, and check Visitor's new Left the site with the badge notification switch.

Dashboard and TV. Live Devices shows what is in your buildings; a new Outside Now widget shows who is outside: on the move, stopped, at a geofence or not heard recently, with a short list and a street map. TV boards get an Outside now widget with counts only. Ctrl+K says where a person or device is now and finds geofences.
Rules. New conditions under Outside (GPS): time outside, stopped outside a geofence, not heard while outside (a lone-worker check) and speed. Each fires once per stretch outside and never counts time at home. GPS positions now record speed, and geofence entries and exits appear in the Activity Log.
Dispatch by distance. Security Watch's Assign and Facility's Nearest available list people on site first, then staff outside ordered by distance to the site, with people at home last.
Visitor and Warehouse. A visitor badge or warehouse tag seen outside reads Outside instead of not heard or no zone. Hosts and reception are told once when a visitor leaves the site with the badge, and the automatic check-out counts only your own gateways' sightings.
Monitors. Live Monitor shows a GPS tag's real last-seen time and At home for a person at home, with no position. On the Entity Tracker, time outside shows its real length, Today counts time on site indoors with time outside beside it, fixes taken while indoors no longer read as an outdoor stop (also in GPS Replay), and the floor-plan pin clears as soon as someone goes outside.
OAIA points out long stops away from your sites (compared with that person's usual stops), tags that went quiet outside, low batteries on field tags, and assets moving outside site hours — never a home location.
Privacy. Places that look like someone's home (most stays overnight) are left out of the frequent-stops lists on Location › Map and Outdoor Activity, with only a count; System → Config → Outdoor & GPS → Privacy can show them for fleets parked at depots. Data subject requests export and erase outdoor positions too. GPS Replay, Outdoor Activity, Monitor → Scout, Device Replay and OAIA's outdoor answers follow each role's entity access.
One word for it. Roll Call, Toilet Hygiene, Visitor and Live Monitor say Outside.

2026-10 · Watch gateways and sensors from Zabbix

Date: 2026-10-21 Theme: Your network monitoring system can watch every gateway and sensor next to the rest of your estate. Administrator action: to use it, create an API key with access Monitoring only and import the Zabbix template from the documentation (Integrate → Watching gateways and sensors from Zabbix).

Monitoring status API (GET /api/v1/monitoring/status): every gateway and sensor of your organisation, up or down right now, with a summary, minutes since each was last heard and sensor batteries. It uses the same line Omaya alerts on — a gateway is down after 15 minutes without a packet outside its quiet hours, a sensor is silent after its own silent threshold — so your monitoring system and Omaya never disagree.
Zabbix template: one request a minute discovers every gateway and sensor, adds new ones by itself, and raises Gateway … is down, … is silent and battery low, tagged by location and floor. People's tags are left out by default.
Monitoring-only API keys: My Profile → API Keys has a new Access choice. A Monitoring only key reads up/down status and nothing else, so the copy kept in a monitoring system's configuration cannot read or change anything else. Existing keys keep full access.

2026-10 · Monitor → Outside: who is outside right now

Date: 2026-10-21 Theme: One live page for everyone outside your buildings, from GPS trackers and Omaya Scout phones. Administrator action: none. The page appears in the Monitor menu for organisations with GPS positions in the last 30 days.

Outside (Monitor → Outside): everyone outside on one street map with your geofences, and tiles for Outside now, On the move, Stopped, At a geofence and Not heard 15+ min. Each person or tag says where they are, how long they have been out, today's distance and time outside (the same figures as Sensor Trail and Attendance), how old the last position is, and whether a tracker or a Scout phone placed it. Click someone for their route today; Details and Replay open the Entity Tracker and GPS Replay. Filter by label or by your field-staff label.
Home stays private. Someone stopped where they spend the night reads At home, with no position on the map, and their route today leaves home out.
Nobody just vanishes. People who came back in within the hour are listed under Back inside, and someone whose phone or tracker went silent outside stays listed as Not heard, rather than dropping off the page.
Right Now gets an Outside now tile that opens the page, and the Entity Tracker's Outside group links to it.

2026-10 · Field staff: working outside counts

Date: 2026-10-20 Theme: Attendance can count field staff's time outside — client visits, driving between sites — without a geofence for every client, and never their time at home. Administrator action: pick the label your field staff carry with the Field staff button on Reports → Attendance.

Working outside. For people with the field-staff label, time outside on their tag's GPS within working hours (your opening hours, else 08:00–18:00) on working days (your operating week, else Monday to Friday) counts as working outside, and with time on site makes their working time. A field day spent only outside counts as a day present. The page, the person view and both exports show it.
Home is never work. Home is found on its own — where the tag spends the night on 3 or more of the last 30 nights — and is never shown. Until it is known, the person's row says home not known yet.
Make geofence for places stopped at 3 times or more: on Location › Map (was twice), flagged when the place looks like someone's home, and on Reports → Outdoor Activity under Places they keep stopping at, opening the map ready to name it.

2026-10 · Reports that know a tag can be outside

Date: 2026-10-20 Theme: Now that people's tags report GPS through Omaya Scout as well as trackers, the reports that touch outdoor time say so plainly. Administrator action: none. Existing Device Trail schedules send the new layout from their next run.

Device Trail by email (System → Schedule Report): one row per day and person or tag — first and last seen and time on site, and where the tag reported GPS, time outside, distance, trips and stops, with a total. It used to send one line per trail record with a MAC and a zone.
Outdoor Activity speaks of people and tags, not trackers: Everyone with GPS, People and tags, Outside.
Toilet Hygiene › Live: a cleaner whose badge is outside reads Outside the building (or Last seen outside), instead of a room they left, or none.
Visitor movement: time a visitor's badge was picked up outside reads Outside in the path and the dwell list, instead of a blank place.
The same figures everywhere. Sensor Trail, its email, Attendance and Outdoor Activity now count alike: on site joins stays into visits (a gap of up to 30 minutes with no time outside still counts as on site), First in and Last out are indoors only, a stay ends at its last reading in the room, distance is trips only (not GPS wobble while stopped), and trips and stops count on the day they start. Outdoor time is called Outside on every screen and report.
Entity profile: its counts are Stays (each time the tag was placed in a zone), no longer called visits.

2026-10 · Sensor Trail: indoors, outside and the distance travelled

Date: 2026-10-20 Theme: Reports → Sensor Trail reads day by day, indoors first; when a tag also reports GPS, an Outside tab adds its route, time outside and kilometres. Administrator action: none.

Indoors first. Tiles for on site, stays, zones visited and average stay, a Per day table (first and last seen, time on site), and the Indoors tab with where the time went, how it was positioned, the zone timeline and every stay. Click a day to show only that day; Export days downloads the table.
Per day, by date. One row per date; with several people or tags, the date carries the day's totals and each of them sits beneath it.
Outside, only where there is GPS. When the tag reported GPS in the range (a tracker, or a phone running Omaya Scout), an Outside tile (time, km, trips, stops), outside columns in Per day, and an Outside tab: the route on a street map with Play it back, where the time outside went, and the Journey stretch by stretch with the sensor, distance and GPS fixes. With several tags outside, a chip per tag (with its km) narrows the map, the journey and the breakdown to it, each in the colour of its route. Without GPS the page is indoors only.
Fixed: outdoor stretches showed twice for a person's tag and as 0 seconds long, so no outside time or distance reached the totals; and the timeline stopped days short on a long range.

2026-10 · Activity Log: full timeline or what needs a look

Date: 2026-10-20 Theme: System → Activity Log keeps every event in order, and adds a view of only what deserves a second glance, how each problem ended, a record's own history where you work on it, and a weekly access review. Administrator action: none. To have the access review emailed, add an Access Review schedule under System → Schedule Report.

Full timeline or Needs a look. The timeline is every event in order, as before. Needs a look keeps deletions, failed sign-ins, password and role changes, administrator accounts added, removed or switched off, device commands that did not go through, and system problems.
Lanes. People (with Changes, Deletions, Sign-ins and In the apps), Devices, System, AI assistant and Page views; five tiles count the last 7 days and open their lane. A chart shows each day by lane; a bar opens that day.
How it ended. A gateway offline, a device silent, a connector failing or a zone over capacity says underneath when it came back and how long it was out, or that it is still out.
History where you work. The device panel has a History tab; gateway and administrator panels and the rule editor have History too: that record's own events from the last 90 days, with a link to the Activity Log narrowed to it.
Integrity you can see. The line under the title says when every record was last checked against its checksum and whether anything had changed. CSV and PDF exports use your organisation's clock and the list's own words, check each exported record again, and end with a fingerprint of the records exported.
What changed, on the row. An update shows the fields it changed in words, such as On: off → on · repeat at most every: 1 h → 5 min; the detail lists them all as Was and Now. Secrets show only that they changed.
One person's sessions. Showing only one person groups their rows by session: when they signed in and out, how many events, the IP address and the browser.
Access Review (System → Schedule Report): accounts added, removed or given another role, role changes, failed sign-ins, deletions, administrators with no sign-in for 30 days or without two-step sign-in, and the integrity check. For roles with View Audit Trail.
Plain words. Durations read as 2 h 43 min rather than minutes, times are on your organisation's clock, and a record in an app names the record and its app (updated company member #464 — Visitor). Changes Omaya staff make to your gateway-health thresholds are now in your log, under their name.

2026-10 · Scout: outdoor coverage per tag holder

Date: 2026-10-20 Theme: Monitor → Scout answers whether each person's tag keeps working outdoors, through their own phone running Omaya Scout, where no gateway hears it. Administrator action: ask anyone listed under No phone yet to install Omaya Scout.

One row per tag holder whose tag a Scout phone heard in the last 30 days: Outside now, Covered, Phone gone quiet or Seen at a site only. Holders no phone has heard are one count, so a large organisation stays a short list.
Their phone. Each holder's phone is worked out as the phone that places their tag away from your sites; a phone that only hears tags at a site carries nobody. A site is a geofence, or within 300 m of a location's pin.
Map and trail. Where covered holders were last outside; click a holder for their trail over 24 hours, 7 days or 30 days. The phones list says whose tag each phone carries and whether it is still reporting.
Person tags are covered where they are published to Scout for your organisation; the guides now say so.

2026-10 · Live Monitor: five tabs, each a summary

Date: 2026-10-20 Theme: Live Monitor is the "right now" page again: what each gateway and sensor is doing, with short summaries for rules and temperature and the longer history on its own reports. Administrator action: none. Bookmarks to the removed tabs open the report that has that content.

Five tabs. Gateways, Sensors, Floor Plan, Rules and Temperature. Predictive is the health strip on Gateways; Enter/Exit and History are the Enter/Exit and Sensor Trail reports.
Rules shows today by rule (how many times, when last, which actions, whether they went through) and anything that did not go through, with links to Rules, Central Alerts and Rule & Alert Effectiveness, instead of every run.
Temperature shows each sensor now with today's low and high, and one line across all of them, with a link to Environment for the hourly history.

2026-10 · Central Alerts: every module, repeats folded

Date: 2026-10-20 Theme: One place for everything that fired, opening on what needs a look, with the alerts that keep repeating folded into one row that says why. Administrator action: none.

Every module. Central Alerts now also carries connector failures, Data Centre and Warehouse alerts, rooms over capacity, OAIA findings, compliance findings, radar reflections and maintenance requests past their SLA, beside rules, emergencies, anomalies, battery, presses, gateways, coverage, Toilet Hygiene and Security Watch.
Needs a look. The page opens on what is still open or was raised in the last day; counts for Needs a look, Critical, Repeating and Done, and a bar per day, each open their part of the list.
Repeats folded. The same alert three times or more from the same place is one row: how often and since when, why it usually repeats, and a link to where to change it.
Plain words and links. "Has never reported" instead of minutes in the hundreds of thousands; every row links to the rule, device, app or page that handles it. Acknowledge one, a whole repeat, or everything shown.

2026-10 · A bigger entity view on Entity Tracker

Date: 2026-10-20 Theme: Clicking a person or asset on Entity Tracker opens a large view that answers where it is, how its day has gone, and whether that is usual. Administrator action: none.

The entity view. It opens over the list, with the floor plan on the left and the facts on the right, and closes with Esc or a click outside, leaving the list as it was. ‹ › or the ← → keys step to the previous or next entity; Open as page gives it its own link.
Today says when the entity turned up, how many rooms and how long tracked, and whether that was about its usual time, later or earlier.
Day strip: where it was across the day as coloured blocks, for today and the earlier days.
How it is placed names the gateway that placed it and the one its last packet came through, which can differ.
Outside folds to one line when the phones only placed it at one place and it made no trips.

2026-10 · Gateway timing, and scheduled tasks that say why they failed

Date: 2026-10-20 Theme: A Configuration card that did nothing now does one thing properly, and the scheduler's failures say what went wrong. Administrator action: none. Gateways are judged exactly as before until you change Gateway timing.

Configuration, tidied. Opening hours has its own card: the days and hours reports count, and whether they apply, moved out of Organisation. Industry is asked once, on Organisation. The AI card (was A.I Model) folds answer style, length and the connection check under Advanced. License says No license when there is no key instead of Active. The General card reads Kuala Lumpur · UTC+8 · °C. The presence interval card is for platform operators; organisation admins tune timing with Detection speed, whose estimate now uses the organisation's own interval.
Gateway timing. Configuration's IoT Defaults card is now Gateway timing: Stale after and Offline after set when a quiet Bluetooth gateway is shown stale, then offline, and when the gateway down and back-up alerts and the outage log count it down. LoRaWAN gateways keep their own longer windows. The signal windows and minimum stay the card used to show were never applied and are gone; a gateway's signal floor is set on the gateway itself.
Outside weather. A location Open-Meteo turns away for a moment is tried again within the same run, and a run counts as failed only when a location's weather is more than three hours old.
Scheduled tasks. A failed task now records its own last words as the reason, with the end of its output, instead of "Task failed.". A run cut short by a restart is closed as timed out instead of staying "running".

2026-10 · Panic buttons are rules

Date: 2026-10-19 Theme: A panic button and a rule were two ways to react to a press; now a panic button is a rule, so Rules lists everything a press can do. Administrator action: none. Panic buttons already set up were moved for you and keep paging the same people; check the list on Devices → Panic Buttons.

Panic Buttons makes rules. Each panic button is a rule with Button pressed and the new Page these people action, listed under Rules too; Open as a rule on a row opens it there to add a webhook, a work order or any other action. A panic button can now cover one device, every device under a label (or carried by someone with that label), or every button in a room, floor or site, and say which press: any press, a single, double, triple or long press, or a badge's SOS, Alert or man-down. Not again for the same button within sets how soon it may page again.
Page these people is a rule action anywhere: admins by email and in-app, app users by email, SMS, WhatsApp and push, with a message that says what was pressed, by whom and where.
Button pressed rules fire. A rule on Button pressed never fired before; it now does, and can match the kind of press.
Security Watch follows the press. A panic button set to one kind of press (a double press, say) makes only that press a panic on the guards' board; other presses on the same button are a warning, as for any button. Any press covers every kind, a worn button's man-down included. The Which button list puts devices with a button first.
What the button does. A device's When the button is pressed card lists the panic buttons that cover it with the rules, and which press each one reacts to.

2026-10 · Omaya Scout finds tags on iPhones

Date: 2026-10-19 Theme: Phones running Omaya Scout pass on what they hear, the way a gateway does, so iPhones now place your tags outside too. Administrator action: none. Give tags that share a factory ID their own ID, or iPhones cannot tell them apart.

iPhones place tags. iOS hides a tag's Bluetooth address, so Omaya now recognises a tag by its iBeacon, Eddystone or MOKO ID, learned from your gateways. A tag is found on iPhones once a gateway has heard it; Android phones find it straight away. See Phones as trackers.
For your own app. The Scout API takes every advertisement a phone hears and decides which are your tags (POST /api/scout/v1/adverts); see Pushing data in.

2026-10 · Button presses on Minew tags

Date: 2026-10-19 Theme: A Minew tag's button can be set up to raise an alarm from Omaya, without going to the tag with Minew's app. Administrator action: none. To use it, turn Button presses on for each tag.

Button presses. On a Minew MWC01 tag, Configure (Device → Sensor → Control) has a Button card. With Button presses on, a double press on the tag raises a Button Press alarm: Omaya sets up one empty broadcast slot through a gateway, broadcasting only for 20 seconds after a press, and reads it back to confirm. Off switches that slot off again.
When a slot broadcasts. Each Minew broadcast slot on the Configure page now also shows when it broadcasts: always, or only after a press or movement.

2026-10 · One look for everyone; Rules, Connectors, the Map Editor and System pages redone

Date: 2026-10-19 Theme: The pages you set Omaya up on lead with what needs your attention, the way Monitor does, and the new look is now the only look. Administrator action: none. Anyone who had chosen the classic look sees the new one the next time a page loads; bookmarks and links still open the same pages.

The classic look is retired. Back to the classic look is gone from the user menu and the page list; every browser shows the current design, including the sign-in page.
Rules. The list opens with counts (on, fired today, needs a look, off), says each rule in a sentence, and has an On switch and Run it now on the row. The editor opens on the Flow view, with the rule's priority, days, hours and repeat at most every under Rule settings when no box is selected; Card view shows the same rule as Watches, When, Then and Checked. Test now also checks each condition against the live reading. Floor and site rules on device readings (battery, status, signal, any sensor field) now fire.
Connectors. Each connector says whether anything is getting through: Working, Quiet, Failing or Off, grouped as data in, alerts out, business systems and storage, with the last message and a 24-hour chart for gateways. Test on a row runs read-only checks and sends nothing.
Map Editor. One floor list with rooms, gateways, sensors and walls; a three-step floor setup (picture, scale line, size and heights); Import for finding rooms in the picture or a CAD drawing; a coverage check; and the calibration walk from the top of the page. Zones and Zone types lead with zones not yet drawn on the plan and zones with no gateway.
System. Admins, Roles, Activity Log, SLA & Uptime, Usage, Labels, Data tables and Schedule Report share one layout. Roles set each area to None, View or Full. The Activity Log opens without page views, groups by day and folds repeats, and its export matches the list. Gateway uptime is now measured over the range you pick, from the outage log, and per gateway. Usage leads with anything that did not get through and compares each figure with the month before.

2026-10 · Pages that open each other, and calmer Sensor, Gateway, Location and Venue pages

Date: 2026-10-18 Theme: What a page names opens the page about it, already set to that thing, and the pages you manage devices and spaces on lead with what needs your attention, said in words. Administrator action: none. Bookmarks and links to these pages still open them.

Links between pages. A person, device, gateway, zone, floor, rule or day named on a page opens the related page set to it: Entity Tracker, Live Monitor, Sensor Trail, GPS Replay, Attendance, Asset Utilisation, Right Now, Room Usage, Space Utilisation, Heatmap, Environment, Battery, Distance, Button Press, Sensor Status, Gateway Availability, Central Alerts and Custom Reports all open on a given person, device, zone, floor, location, gateway, rule or report, and keep it in their address so the link can be bookmarked or shared. Incidents link to Security Watch (the guard app asks you to sign in to it), Environment's room cards name the sensors behind their figures, and the Activity Log links each entry to what it is about.
Sensor. Needs attention says what is wrong in a sentence, worst first, and opens the list on those devices; chips appear only when something is behind them. Tabs by kind give tags who carries them and where they are now, sensors their latest reading, and actuators an on/off switch in the row. The list puts what needs attention first, every row has a status dot, and Columns hides or shows any column per tab. A device's panel has Overview, Control (switches, Configure, its own settings), Data and Calibrate tabs, with Edit beside the name; selected rows get a bar to move, label, activate, deactivate or delete them.
Gateway. Three cards: up now, positioning per floor, and the last week's outages, each of which opens to the gateways involved. Where shows the building with the floor and zone under it, and Columns hides or shows columns.
Location. Sites shows each location with a row per floor (zones, gateways online, floor plan) and links to that floor's plan, Right Now and Heatmap; the full tree is folded underneath. Add Building is at the top of the page, and the map opens zoomed to your sites.
Venue. Floors shows each floor as a card with its plan as the picture, and whether it has the three gateways it needs to place tags on the map.
Sensor settings opens in a moment rather than several seconds.
The new look, everywhere. Report and system pages whose cards sat on a flat fill (Central Alerts, Enter/Exit, Loyalty, Asset Utilisation, Dwell Time, People Count, Heatmap Total, Button Press, Sensor Trail, Distance, Rule Effectiveness, Incident Response, Scout, GPS Replay, Fall & Man-down, Impact Events, Scheduled Reports, Usage, and the Sensor utility pages) now use the same glass panels as the rest, and dashboard widgets no longer show a white box behind them.

2026-10 · Minew tags and beacons can be configured from Omaya

Date: 2026-10-18 Theme: Minew's BeaconSET+ tags, beacons and buttons join MOKO in Device Configure: read their settings and change them through a gateway, without going to the device with a phone. Administrator action: none to start. To change the factory password on your Minew devices, open Device → Sensor settings → Passwords, choose Minew, and tick Also change it on the sensors.

Configure works for the Minew MWC01, MWL01, MBS02, S1, E7, E8, i7 and MBM01, through a gateway that can send commands (Dusun). Each of up to six broadcast slots shows what it broadcasts, how often and how strong; how often and how strong can be changed, and every change is read back from the device.
Passwords: Minew devices sign in with the organisation's Minew password, a password of their own, or the factory minew123. Omaya can change the password on the devices, one at a time; a Minew password is exactly 8 letters or digits.
Never sent to a Minew device: switching it off, factory reset, firmware updates, making it stop accepting connections, renaming it, and changing what a slot broadcasts or what a press triggers. Those stay in Minew's BeaconSET+ app.
Minew radars (MSR01-A, MSR01-B) use a different Minew kit and are not configurable from Omaya yet; the Settings library lists them separately.

Date: 2026-10-18 Theme: The Device menu leads with the lists you work in every day; the tools you open now and then sit together under Utilities, and pages that belong together are one page with tabs. Administrator action: none. Every page keeps its address, so bookmarks and links still open the right place.

Device now shows Gateway, Sensor and Panic Buttons, then a Utilities group.
New sensor types brings together what was Sensor Onboarding and Custom Decoders: Learn a sensor, Custom types and Test decode are its tabs.
Sensor settings brings together the Settings Library and Sensor Passwords (previously Device Passwords): Settings library and Passwords are its tabs.
Sensor Auto Discovery and Tag Power were buttons at the top of the Sensor page; they are now items under Utilities.
Bulk Import is a button at the top of the Sensor page rather than a menu item, shown to people who may create devices.
Beacon & Wi-Fi Anchors moves to Space, beside your locations, venues and zones.
Custom Decoders, Beacon & Wi-Fi Anchors and Device Anomalies use the same layout as Monitor: a summary card at the top and one card with tabs over the list.
The guides and the menu tour describe the new places.

2026-10 · The new Omaya look is now the default

Date: 2026-10-17 Theme: A calmer, clearer admin: a dark side rail, glass panels with depth, one colour for actions and one for live data, and readable status colours in light and dark. Administrator action: none. Anyone who prefers the previous design can pick Back to the classic look in the user menu; the choice is kept per browser, and Use the new Omaya look returns.

Side rail and page list. A carbon side rail holds the menu sections; choosing one opens its page list over the page, and the list closes as soon as you pick a page, so the page keeps its full width.
Colours with one meaning each. Violet marks what you can do and what is selected (buttons, links, the selected row or tab); cyan marks what is live. Status colours (ok, warning, alarm) are never reused for either, and status badges are drawn as tinted pills that stay readable in light and dark mode.
Glass panels with depth. Cards, tiles and the top bar are frosted panels over a soft glow; tables sit on a denser frost so figures stay easy to read; menus, windows and side panels float above both. Where the browser cannot blur, or the device asks for less transparency, panels are solid.
Type and wording. IBM Plex throughout, with even-width figures in tables and tiles, and sentence-case labels.
Your organisation in the top bar. The organisation you are working in and your role are shown as a chip in the top bar, in place of the separate band above it.
Phone width. The new look applies at every screen size. Below desktop width the menu slides in from the side, and tiles, filters and side panels rearrange to fit.
Gentle page motion. Blocks fade in in reading order, numbers count up and progress bars grow when a page opens. All of it is switched off when the device's reduce-motion setting is on.
Charts follow dark mode. Switching between light and dark redraws charts in the new mode; donut charts previously kept light outlines on dark cards, and their legends could be cut off.
Restyled screens. The sign-in page, Rules (the rule sentence leads, with its blanks highlighted, over a drafting-grid canvas), Settings, Devices and the OAIA page follow the new design. Page-header buttons such as New, Import and Refresh are smaller across Data Tables, Devices, Panic Buttons, Custom Reports, Rules, Zones and Tags. On short windows, the sign-in card no longer sits flush under the top bar.

2026-10 · Monitor: Right Now by floor, a clearer Entity Tracker, and Device Anomalies in the same layout

Date: 2026-10-17 Theme: The live pages say who is where in plain words, grouped the way a building is laid out. Administrator action: none. Bookmarks to the old Zone occupancy and Last 24 hours pages open their replacements.

Right Now & Live Alerts, grouped by floor. Zones are grouped by floor in building order, each floor header showing its short name (L4, B1, G), how many are there, the zones in use and a strip of where on the floor people are. Floors fold away (remembered per browser) and a chip bar jumps to a floor.
Zone occupancy is now part of Right Now. Each tile carries a capacity status (Empty, Normal, Busy, At or Over capacity). Clicking a zone opens a panel with who is in it, its last 24 hours, a 24-hour forecast and a comparison with another zone. The separate Monitor → Zone occupancy page has been removed.
Tiles that say who the count is. A zone counted by tags shows people when every tag is assigned to someone, otherwise the split ("1 person · 2 tags not assigned to anyone"). Under the bar: capacity in words ("3 of 20 capacity · 15%", or "No capacity set"), density where the zone has an area, the next expected peak, and a short sparkline of the recent window. The floor headers and headline figures use the same people/tags wording, and last-seen times are now correct in your time zone (a tag seen seconds ago could read "8h ago").
Entity Tracker. Status filters are number tiles beside on-site facts: people, assets, moving now and low battery. Each card leads with the room the person or asset is in, then two fixed slots, Here for and Last moved (or Movement while moving, Last heard when no motion is reported). Active cards carry a coloured edge in their status colour, so online and stale stand apart from offline at a glance. The tiles, toolbar and detail panel fit a phone.
Device Anomalies uses the Monitor layout: one card of totals (critical, warning, info, total) and one main card with status tabs Active, Acknowledged, Recovered and All beside the filters.
Monitor → Last 24 hours has been retired. It repeated the Activity Log and Central Alerts. The two things only it showed now appear in the Activity Log as system events: a new low or critical battery alert, and a zone going over capacity and back under.

2026-10 · Outside the building: Omaya Scout, GPS Replay and places with names

Date: 2026-10-17 Theme: Follow tags outside, from trackers' own GPS or from phones that hear them, and see where they went by name and on a map. Administrator action: Omaya Scout is switched on per organisation; ask your Omaya contact to enable it. Street names on Outdoor Activity are off until you turn them on under System → Config → Outdoor & GPS → Naming places.

Omaya Scout. Phones running the Omaya Scout app report the Bluetooth tags they hear, with the phone's own location, so a tag can be placed outside the building where no gateway reaches. When several phones hear the same tag, the most accurate fix is kept. Only tags your organisation publishes to Scout are accepted, and tags bound to a person are not published by default. Phone-relayed readings update a tag's last-seen time, battery and sensor values, but do not change its gateway or signal reading and do not fire rules or button-press alerts. Scout positions follow your organisation's data retention, like other GPS points.
Monitor → Scout (phones) shows where phones last placed each tag on a map (with an accuracy circle and colour by age), the tag list with how many phones heard each one and its battery, a tag's trail on click, and the phones doing the scouting, over the last 24 hours, 7 days or 30 days. On the phone, the scan interval can be set from 10 seconds to 5 minutes (30 seconds by default).
Reports → GPS Replay. Pick a person, asset or sensor and play its day on a street map. Beside the map, the journey lists each stop and the trip between ("Travelled 1.1 km in 8 min"), with the current one highlighted while it plays; click one to jump there. One summary line gives who, the day, distance, time still and time moving. The detail behind each position (which phones, accuracy) folds away under Fix details and Signal details. Without a pick, it lists who had a GPS fix in the range.
The entity panel shows the last day outside. Monitor → Entity Tracker's detail panel adds an Outside · last 24 h map with numbered stops, and the stops and trips newest first.
Outdoor Activity says where. A new Where they went map shows every stop and geofence visit as a numbered spot sized by time, beside a list of places, longest first, with visits, who, the last time there and a Google Maps link. Stops within 150 m are one place. A place outside your geofences is named by its distance and direction from your nearest site or geofence, worked out from your own places, so no coordinates are sent anywhere.
Street names, optional. With Street names on, stops also show the street, looked up on OpenStreetMap, which then receives those stop coordinates; answers are kept for 30 days. Off by default.
Between the two reports. Each tracker, day and place on Outdoor Activity has a Replay link that opens GPS Replay on the right person and day, and GPS Replay links back.

2026-10 · Reports: staff and visitors in Footfall, matching incident counts, and clearer pages

Date: 2026-10-17 Theme: Footfall tells staff from visitors, two reports now agree on incident pick-up, and a round of reports reads more plainly. Administrator action: on Reports → Footfall, open Staff label at the top and choose which entity label marks your staff (labels are set on Manage → Entity). Until then, people are not split into staff and visitors.

Footfall: staff, visitors and unassigned tags. People counted by tag are split into staff (holding the staff label), visitors (anyone else holding a tag) and unassigned (a registered tag bound to nobody); together they make the people total. The split is in the headline, the chart, the table, the places list, the CSV export, the AI Reporting Assistant and the public footfall API. Choosing or changing the label recounts the last 30 days. Visitors without a tag are only in the door counts.
Rule & Alert Effectiveness counts a Security Watch incident as picked up the same way Panic & Incident Response does: acknowledged, or resolved by a person, whichever came first. Before, an incident closed automatically as unattended also counted, so the figure ran higher than on Panic & Incident Response. The Security Watch row links to Panic & Incident Response for the detail.
Gateway Availability. A Bluetooth gateway only sends when it hears something, so an empty building at night looked like an outage. The SLA is now measured on opening hours by default where your organisation has them (still switchable), silence outside opening hours is drawn grey and hatched rather than red, and the default SLA target is 98 % (99 % is still available). One summary card gives the figure for gateways in service, the all-hours figure, and chips for gateways below target, outages in opening hours, site-wide silences and the longest outage. Rows below target are tagged instead of painted red, and removed gateways are faded and never flagged.
Sensor Status is built around fixed-sensor health: one card with the share of expected hours sensors were heard, split into parts you can click to filter; tabs for Fixed sensors, Tags & trackers (measured only while on site) and Actuators; one row per device with "heard 41 of 166 h", the last time heard, and the hour strip at full width underneath.
Battery. Devices that need a look (replace soon, no reading in the range, or no battery level for 3 days or more) are grouped by reason in one panel, one row each with the fact that put it there. Device cards glow amber up to 40 % and red at 20 % or below.
Space Utilisation room card reads as sentences ("Someone was in it 80% of its opening hours…"), with four labelled facts, a chart whose axis reaches the room's capacity, and the weekday strip as bars with their numbers.
People Count. The Hourly Readings by Venue table shows again; it stayed empty and Expand all had nothing to open.
Flow. The floor map draws one line per pair of rooms, thicker for more moves and arrowed when mostly one way, with one count per line, room names, Top 5 / Top 10 / All, and a click on a room to see where people came from and went to. The routes table is one row per pair.
Dwell Time venue cards count "stays", show Passby / Visit / Engaged as one bar that adds up, state the venue's engaged threshold, and label the daily chart with its dates and busiest day.
Fall & Man-down. The events table lines up again (values sat one column left of their headings). The page leads with the detection status, four figures (flagged, would alert, precision, response), accuracy beside the per-kind table with a link to label unlabelled events, and Real / False / Test as one toggle on each event.
Impact Analysis lists events grouped as Running, Upcoming, Ended and Paused, one row each with dates and Analyse; with no events yet it explains the three steps. An event whose floor was removed says so.
AI Reporting Assistant is one layout: the summary across the top, conversations and the chat side by side at the same height, and New chat shown once.

2026-10 · Dashboard: Gateways, Devices and Live Sensor Status add up again

Date: 2026-10-17 Theme: Correct counts on the dashboard tiles. Administrator action: none.

Gateways and Devices tiles show the current counts. They could show figures up to a day old that still included removed devices.
Live Sensor Status now adds up: stale means heard 5 to 15 minutes ago and not since, only registered devices are counted, and online + stale + offline equals the total.

2026-10 · Visitor app settings: tabs, cards that save in place, visit and ID types

Date: 2026-10-17 Theme: The Visitor app's settings page is easier to work through, and only the right people can change it. Administrator action: check Types of visit offered and ID documents accepted under Visits & check-in. Only an administrator, or an app user with configure access to the Visitor app, can now change these settings.

Four tabs: Host companies, Visits & check-in, Registration & privacy and Notifications & integrations, under tiles for companies, visits today, on site now and awaiting approval, each opening the desk.
Host companies is a table at the top, with Create, Edit and Delete on its toolbar. Create and edit open one side panel with every company field, and the company's members are added and removed there. Changes appear in the table without a page reload.
Cards that save in place. Each settings card saves on its own, shows errors beside the field in plain words ("walk-in length", "check-in window"), and keeps Save off until something changes. Switches are real on/off switches, and a ticked option is clearly ticked.
Types of visit offered and ID documents accepted are now set here.
Public pages can be copied, opened or downloaded as a QR code, and the kiosk link is shown.
Fixes. Any app user of the organisation could change these settings. Telegram showed "Not connected" for a bot connected from OAIA. Members of a removed company still saw its visits. Changing a member's role reset their joined date.

2026-10 · Saved views on the device list: no error on opening, and in your language

Date: 2026-10-14 Theme: Opening the device list straight from a link or a refresh showed an error message; the Saved views menu now works however you arrive, and speaks Spanish, Indonesian and Chinese. Administrator action: none.

No more error on opening. Opening Device → Sensor directly (a bookmark, a link, or refreshing the page) showed an error message in the corner. It no longer appears, and your default saved view is applied as it should be.
Tidier menu. Pin, share, default and delete appear on a saved view when you point at it, instead of on every row at once, and the view you are using is shown in bold.
In your language. The Saved views menu, the Save current view window and its messages ("View saved", "View applied" and the rest) are now translated into Spanish, Indonesian and Chinese, as they already were into Malay. They were shown in English before.

2026-10 · Configure devices from Omaya: a settings library, a Configure page and device passwords

Date: 2026-10-14 Theme: Change a sensor's settings from Omaya, only what its model supports, through the gateway, network server or Home Assistant that already hears it. Administrator action: under Device → Device Passwords, save your MOKO password (or change the factory one on every sensor from there). To change LoRaWAN sensors, set your network server's downlink topic as the guide shows.

Device → Settings Library lists every model Omaya supports — including ones you do not have yet — with what can be changed, how the change reaches it, and what Omaya never sends.
Configure on each device (Device → Sensor → View → Configure): only the settings its model has, what the device holds now, and one Apply. Bluetooth changes are read back from the device before they count as done; LoRaWAN changes are confirmed from the sensor's next reports; Home Assistant changes are read back from Home Assistant. Everything is in the device's history and the Activity Log.
MOKO S01: every setting — motion sensitivity, how long it stays occupied, broadcast rate, signal strength, name and beacon IDs, the button, its clock and its password. MOKO BeaconX Pro tags and buttons: broadcast rate and signal strength per slot or press, double-press window, name. Milesight LoRaWAN: how often it reports, and restart. Home Assistant and Aqara: whatever settings Home Assistant offers for the device.
Device → Device Passwords: one encrypted password per maker for the organisation, a device's own where needed, and changing the password on the sensors themselves. Passwords are never shown again.
Never sent from Omaya: switch off, factory reset, firmware update, "stop accepting connections", rewriting a device's identity, and a LoRaWAN sensor's network settings. These are blocked where Omaya talks to the gateway, not only hidden on the page. A panic button's press is never switched off.
Minew devices are listed but cannot be configured yet: Minew keeps its command format private.
The View panel of every device has a Configure card that says what can be changed, or why not.

2026-10 · Space figures that hold up: workplace seats, rented area, and a venue form that saves its cost

Date: 2026-10-14 Theme: Corrections after checking the new space figures against a real office. Administrator action: on each venue, enter its Rented area if your lease charges for less than the floor plan shows, and re-enter its Cost of space and Capacity if you typed them before (they were not being saved). Check each zone's Kind of space.

The venue form now saves Cost of space and Capacity. Both fields were shown but never sent, so whatever was typed there was lost.
Rented area on a venue: the area your lease charges for. Rent on Space Utilisation is this area times the cost of space; without it Omaya uses the floor plan's size, a rectangle round the whole floor that is often larger.
Seats are workplace seats. Zones have two more kinds — Meeting / training room and Toilet, pantry or service room — and only work rooms, open areas and desks count as seats in Seat planning and the Buildings card. Rooms the Meeting Room app books and toilets the Toilet Hygiene app watches were classified for you.
Buildings card says up front when only tagged people are counted, and names the meeting rooms on a floor it suggests giving back.
Data health judges sensors on opening hours only, so a door counter that is quiet at night is not marked as missing data.
Walk-ins reach the room calendar once they have lasted 10 minutes, not for every person passing through.
Space Utilisation has a new layout. Four headline figures (occupied, seats needed, idle space, data health), each building as a verdict with one bar per floor, rooms grouped by floor with unmeasured and service rooms folded away, and the selected room's day beside the list.
Attendance has a new layout. The team is a calendar — one tile per person per day, shaded by time on site, with late arrivals and early leaves marked on the tile's edge and a tag left on site overnight shown as such; people present each day as bars; a person's days and stays open in a side drawer from their name.
Footfall has a new layout. Headline figures with the change against the period you compare with, missing-data and closed-day notes as small chips, the day-by-day chart, places ranked with bars beside the typical week, and the day table with one row per day that opens to its places.
Room Usage Map has a new layout. The room from above fills its card, with the coverage line under it and the replay beside a timeline of the day you can click to jump; busiest hours and every day of the range (quiet days included) sit alongside, each day one click from its replay. Spots used now reads against the size of the counting area, and time in use is now real: the radar reads every few seconds, so each sighting counts for the gap to its next reading instead of one second (a full hour used to show as a few minutes, here and in OAIA answers).
Outdoor Activity has a new layout. Headline figures, hours outside per day stacked by tracker, trackers ranked by time outside — open one to see each day as a ribbon of geofence visits, trips and stops — and geofences ranked by time spent inside. The default range now follows your time zone, not the browser clock in UTC.

2026-10 · What your space costs, office days by team, Slack and Teams, and dashboards in Power BI

Date: 2026-10-13 Theme: Turning occupancy into the decisions a real-estate, HR or facilities team makes. Administrator action: for the Buildings scorecard, set a cost of space per m² on each venue, scale its floor plan and set seat capacities on the zones.

Reports → Space Utilisation → Buildings. Each building gets a class — Opportunity (savings without building works), Alert (the space does not fit how it is used) or Resilient — with the sentence that explains it. It shows the cost per square metre actually used, seats needed against seats provided, and which floors could be given back because their people fit in the spare seats elsewhere. Where rooms count only tagged people, it says so.
Data health on reports. Space Utilisation and Attendance show how much of the sensor and gateway data behind the dates on screen arrived, name the worst devices, and link to Device Status. Gaps are unknown, not empty.
Reports → Attendance → Office days by team. For a hybrid or return-to-office policy: each team's (label's) average office days a week, how many meet the policy you pick, and how many are in on each weekday.
Slack and Microsoft Teams are rule actions. Paste the channel's incoming webhook URL into the action; the message lands in the channel with the title in bold.
Rules know about bookings. The new condition Minutes Until Next Booking (meeting room) lets a switch-off rule leave a room alone when its next meeting is about to start. OAIA suggests it for meeting rooms.
Walk-ins on the room calendar. A meeting the sensors noticed without a booking now appears on the room's Google or Microsoft 365 calendar, grows while people stay and ends when the room empties.
Power BI, Excel and Grafana. Two new read endpoints give occupancy as tables, in JSON or CSV: one row per zone and hour, and the Space Utilisation report per zone, floor and building. Step-by-step setup is under Docs → Integrate → Power BI, Excel and Grafana.
Square feet. Configuration → General → Area Unit shows areas and costs of space per ft² instead of per m²: the venue's cost of space is typed per ft², and the Buildings scorecard, idle cost and the API's area columns follow. Nothing stored changes.
ROI estimate on the About page (m² or ft²): put in your people, desks, area and rent, and see the seats you need, the space you could give back and the yearly saving, with every assumption editable.

2026-10 · Security, honest delivery status, LoRaWAN energy, and a working connector hub

Date: 2026-10-13 Theme: Fixes for the problems found while writing the setup guides. Administrator action: none. If you relied on a Manager editing roles above its own level, an Admin now has to do it.

Roles. Nobody can give a role or a permission they do not hold themselves, change their own role, or manage an administrator whose role can do more than theirs. Adding or removing permission names is for Admins only. Re-seeding roles only adds what is new and never undoes your changes.
Passwords. Administrator passwords are no longer kept in a form that can be turned back into the password; the stored copies were cleared.
Rule actions say what happened. Monitor → Live Monitor → Rules shows each action as sent, skipped with the reason (for example No active Telegram connector, No recipient set) or failed with the provider's error. It used to say sent in all of these cases.
Email from a rule is sent through Omaya's own mail service when you have no SMTP connector, instead of being skipped. Gateway status rules fire even in an organisation with no active administrator.
Configuration → Schedule no longer shows Notification Throttle and Device Monitor Interval, which were saved but used by nothing; each rule's Repeat at most every is the control.
Milesight CT clamps and WS5xx smart sockets over LoRaWAN now feed the Energy app and the power and energy rules. A socket's mains voltage is no longer shown as a battery level, and a meter's own temperature no longer appears in the Environment report.
Connectors. Telegram Test and Find chat IDs, the S3 buttons and the Identity Provider form work inside the connectors page. S3 Auto-export scheduled reports now uploads each scheduled report, and Test writes a small file. External CMMS offers the one contract it really speaks (Generic REST), shows the HTTP status on Test, and retries a failed push. Tiles now appear only to roles that can open them. SMTP encryption choices now differ; WhatsApp numbers accept a leading + or spaces; Home Assistant's Skip SSL verification applies everywhere; a TCP or UDP port can be used by one connector only.

2026-10 · Energy: a meter whose kWh counter is stuck is counted from its power

Date: 2026-10-13 Theme: Some meters report power correctly while their own kWh counter does not move, or send no counter at all. Their energy used was shown as 0 or a dash; it is now worked out from their power readings and marked as an estimate. Administrator action: none.

Today kWh, the total, the cost, reports, rules, the TV energy widget and OAIA all use the estimate for such a meter, marked ≈ and estimated from power on the Energy page.
A counter that did move is always trusted over the estimate, and gaps longer than 15 minutes between power readings add nothing.

2026-10 · Docs: step-by-step setup guides, and the fixes they turned up

Date: 2026-10-13 Theme: Every setting, connector and sensor kind now has a click-by-click guide with diagrams; writing them found several things that did not work, now fixed. Administrator action: none. A leak rule now uses the field leak_detected and a door rule door_open, whatever the brand.

New guides: Connectors (every connector, including the steps outside Omaya: a Telegram bot and chat id, an app password for Gmail or Microsoft 365, an S3 bucket and key, a Home Assistant token), Organisation settings, card by card, Start from an industry pack and Registering many devices at once in Setting up a site, Roles and permissions, Who did what: Activity Log and Usage, Being told when a gateway drops, Why did I get 40 emails (or none)?, and hardware pages for presence and door sensors, environment sensors, energy meters, and radars.
Fixed: Bulk Import. Commit was refused on every file; it now creates the devices, and a refused file says why instead of "Commit failed". The sample file always uses your own venues and zones.
Fixed: Configuration → Schedule. Save answered "The PUT method is not supported" and kept nothing; it saves now.
Fixed: starter packs. Two schedules sent an empty "Unknown report type" file and others sent a report unrelated to their name; each now sends the report its name says (People Count, Footfall, Space Utilisation, Dwell Time or Environment). Apply and Dry-run now show what was created or would be.
Fixed: Data Retention. The hints showed old defaults (90 / 365 / 30); they show the real ones (60 / 30 / 7), and the audit log takes 7 to 120 days, the range the nightly clean-up keeps.
Fixed: leak and door sensors. Milesight EM300-SLD, EM300-ZLD, EM300-MLD and WS303 leak states, the EM300-MCS door contact, and Home Assistant door, window and moisture sensors were received but never reached a rule. They now arrive as leak_detected and door_open.

2026-10 · Docs: Recipes, answers to business questions end to end

Date: 2026-10-13 Theme: The guides explain each part of Omaya; the new Recipes tab puts the parts together to answer a question someone in the building is asking. Administrator action: none. Open /docs/recipes/.

Toilet visits by gender from one door counter on the shared corridor and one PIR in one toilet: the data table, both rules, the report with only where and a calculated column, and why a PIR needs a few minutes' Repeat at most every or it counts one visit several times.
Space and occupancy (over capacity, rooms nobody uses, an unusually busy or quiet day, stuffy meeting rooms), Energy (switch a room off when it empties, power left on at night, a floor using more than usual), Safety and security (lone worker, panic button, an asset leaving its floor, a vehicle leaving the site, after hours, a door left open) and Keeping the building running (too warm, a leak that becomes a job, a gateway down, a noise log, anomalies).
Tips and tricks: short answers, each linked to the guide that explains it.
Fixed: the door-counter examples in Data tables and Rules on any sensor reading named the field entry_delta, which no counter sends; it is entry. A rule built from those examples never fired. OAIA's own instructions had the same name and now say entry, and the rule builder labels entry and exit as Entries / Exits since last report. The door example now uses door_open.

2026-10 · Toilet Hygiene: a cleaner can close a job early, where the toilet allows it

Date: 2026-10-13 Theme: A toilet that needs less than a full clean no longer leaves its job open until an admin closes it. Administrator action: none. To use it, open a toilet, press Edit and turn on Cleaners can close a job early in the Cleaning card.

Off by default, per toilet. With it off nothing changes: a job closes when the badge has stayed the clean time, or when an admin closes it.
On: once the cleaner's badge has been seen at the toilet, the job on My work shows You stayed 3 of 10 min and a Close job button. They pick a reason (Already clean, Light clean only, Called away, Other); photo and note stay optional.
Recorded as what it is. The job is saved as Closed early with who closed it, minutes stayed against the minimum, and the reason. The schedule board shows a Closed early tag and a line in the job's story, and the admins told about the job get "Closed by Siti after 3 of 10 min. Reason: Already clean."
Called away does not reset the toilet's dirty score or visitor count, so the toilet can raise the next job.
The "not finished yet" notification adds "If the toilet is already clean, open the app and close the job" at toilets where this is on.
A cleaner can no longer close a job that is nobody's yet without having been at the toilet (the app never offered this, but the server accepted it).

2026-10 · Warehouse: see what is stuck, do the next task, feel every scan

Date: 2026-10-13 Theme: The Warehouse app now opens on what is stuck and why, puts each person's next task first, groups alerts by their cause, and answers every scan at once. Administrator action: none.

Today (was Dashboard)

One sentence at the top: how many orders are late, whether the floor has gone quiet, the oldest late order and which stage holds the most work.
The flow: every inbound and outbound stage with how much is waiting and the oldest wait; the stage holding the most late work pulses. Click a stage to list its orders, oldest first.
Docks today by the hour, and silent storage sensors grouped into one line with what to check.

Floor

Your next task: the single most urgent job you may do, one button away.
Only the tasks with work waiting are shown, with how late the oldest is; the rest are folded under Other tasks.
Every scan answers at once: a green ring and a tick when it is right, a shake and a sentence saying why when it is not. Phones buzz on a wrong scan.
Picking shows how many lines are done and where to go next in large letters.

Orders

Filter chips with counts, including Late; 10 days late or due in 3 days on every order; search by partner name.
Click an order to open its steps and lines in place.

Alerts

Grouped by cause: sensors that went silent together, jobs stuck on the floor, an overdue gate pass. I'm on it takes the whole group.
Times in hours and days instead of minutes.

Motion

Cards arrive one after another, numbers count up, an order glides to its new stage and new alerts slide in. Nothing moves for anyone whose device asks for reduced motion.

2026-10 · Data Centre: the problem first, every rack at a glance, a wall screen in 2D or 3D

Date: 2026-10-12 Theme: The Data Centre app now opens on what needs someone, colours every rack by what you choose, shows the power path from mains to rack, groups repeating alerts into their cause, and puts the whole site on a control-room TV. Administrator action: in Setup → Devices, give each UPS its Side (A or B feed) so the Power page can say what happens if one side fails.

Now (was Overview)

One sentence at the top: a UPS on battery and its minutes left, the open alert with who is on it and how often it happened this week, a unit that stopped reporting — or Everything is normal and what was checked. I'm on it takes the alert from there.
Four key numbers: PUE against its target, IT load against capacity, power (minutes on battery) and cooling.
Every rack, row by row as they stand, coloured by Inlet air, Power or % of rated. Click a rack for its last 24 hours of inlet air and what to do about it.
Works on a phone: the numbers first, every rack across the width, a tapped rack's details just below.

Floor

The same Inlet air / Power / % of rated choice in 2D and 3D, a panel for whatever you click, and Find a rack or unit.

Wall screen (new)

A full-screen, always-dark page for the control-room TV: the sentence, PUE, IT load, power and every rack's inlet air, readable from the back of the room.
Shows the room in 2D or 3D (the 3D room turns slowly by itself), or switches between the two. It keeps the TV awake and says when it is not updating.

Power (was UPS)

The power path from mains through the transfer switch and each UPS, grouped by side, to the racks, with the generator and when it last ran.
If one side fails: what the other side would carry against its rating, whether the site stays up, and how much load you can still add.
Every time it went on battery: one row per mains dip, whether the generator took over, and the pattern when dips keep coming at the same time.

Alerts

What keeps happening: alerts grouped into their cause — A08 runs warm 4 times a day, mains dips every 4 days — with the hours it happens, whether the neighbours stayed cool, and Make a work order when the Facility app is on.
Every alert filters by critical, not acknowledged or one pattern.
A red line when critical alerts would reach nobody by email.

Energy

Trends is now Energy, with Report and Scheduled reports under it. Old links still work.

2026-10 · TV Dashboard: connect screens with a code, build boards as the TV shows them

Date: 2026-10-12 Theme: TV Dashboard now knows its screens — which TV shows which board, and whether it is on — and boards are built on a live, full-size picture of the TV. Administrator action: none. Administrators (or the people under Organisation → Notifications) now get an email when a connected screen goes off during its hours, and when it is back.

Screens

Connect a screen: open /tv on the TV, type the code it shows, name the screen and pick its board. No long link to type.
The page lists every screen as On, Off for …, Asleep (outside its hours) or Waiting. Shows changes the board from your desk, Reload it reloads the TV, and Rename and Remove do what they say.
A screen that goes off during its board's hours is reported once by email, and again when it comes back.

Boards

New board starts from a template for the place: Lobby, Operations room, By the meeting rooms, Cleaners' station, Warehouse floor, or blank.
The editor shows the real board at TV size with live data. Drag widgets on, drag to move and resize, click one to change its title and where its data comes from. A badge says whether everything Fits the TV; screens update by themselves when you save.
Board cards show a small picture of the layout and which screens show it.

On the TV

Clearer, bigger text and short titles (Meeting rooms, Washrooms, Busiest right now), a new Night look for new boards, and the board's message as a band along the bottom.
The Compliance widget names each standard instead of its code.

2026-10 · Warehouse, Visitor, Toilet Hygiene and Data Centre move like the other apps

Date: 2026-10-12 Theme: The four apps that still appeared all at once now open the way Facility, Compliance, Meeting Room and Energy do. Administrator action: none.

Cards and tiles fade in one after another when a page opens, headline numbers count up, and table rows follow.
On the Warehouse and Data Centre live pages and the Visitor reception desk, a number that changes on the next refresh counts to its new value, and a new order, alert or visitor slides in with a short glow — the rest of the page stays still.
The cleaner's My work page fills its progress bar.
Nothing moves for anyone whose device asks for reduced motion, and a page shows normally if the animation cannot load.

2026-10 · About pages: book a walkthrough, see the real product, and a capability sheet to forward

Date: 2026-10-12 Theme: The public About page now asks for a walkthrough properly, shows real screens and our own measured accuracy, explains privacy, and answers the usual questions; the capability reference can be downloaded as a PDF. Every number on both pages now comes from the platform itself, and claims the product did not back up have been corrected. Administrator action: none required. Walkthrough requests go to the sales mailbox.

Book a walkthrough

Book a 20-minute walkthrough now opens a short form — name, work email, organisation, kind of site and what you would like to see. Each request is kept and sent to the sales team, who reply to arrange a time. (It previously opened the sign-in page.)
A bar along the top of the page, once you scroll, jumps to each section and keeps the booking button in reach.

New on the About page

See it running: real screens from our demonstration site — Security Watch, Roll Call, meeting-room insights, a data hall and a warehouse map.
Measured, not promised: how often Omaya places tags on the right floor and in the right room on our own floors, with the dates measured.
Trust & privacy: anonymous by default, each organisation in its own database, retention you set, roles and single sign-on, the audit trail, and an AI assistant that only reads your organisation.
A comparison against running separate systems and spreadsheets, and answers to common questions — hardware, apps on phones, accuracy, privacy, where data is kept, what happens when the network drops, integrations and time to go live.
The nine worked examples are now tabs, the industries show six with Show all, and every section links to its full detail in the capability reference.

Capability reference

Download as PDF: the whole reference as a capability sheet to forward to procurement or IT; it always matches the page.
Book a walkthrough at the top, in a bar while you read and at the end.

Corrected

Counts of apps, reports, rule triggers, rule actions, industries and modules are now taken from the platform, so they change when it does.
Removed claims for things the product does not do: Slack, Teams and LINE alerts, lone-worker check-in, forced-door and denied-badge events, gateways that keep working offline, signed audit records and multi-region hosting. The supported gateway makes are now named.
Headings on the About pages were dark grey on the dark background, and the main button's text was hard to read; both are fixed.

2026-10 · Indoor positioning: see what your building needs, and try a setting before you switch it on

Date: 2026-10-12 Theme: The Indoor Positioning settings now read your own building and say what to switch on and why, point out where the gateways themselves cannot tell floors apart, and replay your last day with any switch changed so you can see the effect before it reaches the live trail. Administrator action: none required. Open System › Configuration › Indoor Positioning; the Overview tab has the advice and the Try buttons.

What this site needs

The advice now also covers Weigh the floor over time, Change floor sooner while a tag is moving and Room estimation, each with the evidence from your own last day: how many gateway contests were between floors, how many of your tags report motion, how many rooms have no gateway of their own.
New Where coverage is thin: a floor of a multi-storey building with no gateway or only one, and the pairs of gateways on different floors that hear the same tags about equally — how often, and how close their readings were — with what to move or add. No setting can fix these.

Try it before switching it on

Each suggestion has Try on / Try off on the last day. Omaya replays the last 24 hours of up to ten of your tags through the live positioning code twice — as you run today, and with the change — and shows side by side: floor changes, short visits to another floor, room changes, time on each floor, tags with a known floor shown on it, and how each version did on any walks you recorded.
When AI is on for your organisation, it explains what the difference means for your building. Nothing changes until you press Switch it on or Switch it off. A trial takes a few minutes; you can close the panel and come back.

Floors tab

Weigh the floor over time is now a switch: each floor keeps a running likelihood and a tag moves only when another floor clearly and steadily wins, so one loud reading through the slab cannot move it and it cannot get stuck on the wrong floor.
Change floor sooner while a tag is moving: for tags with a motion sensor, a short trip to another floor shows up instead of being smoothed away; a tag lying still keeps the cautious setting. Tags without a motion sensor are not affected.

2026-10 · Roll Call: who is still to find, floor by floor

Date: 2026-10-12 Theme: Roll Call redesigned around the question that matters in an alarm — who is still to find, and where — with floor wardens, "Are you safe?" on people's phones, a screen at the assembly point and lobby TVs that switch to the roll call. Administrator action:
Open Roll Call › Floors & wardens and choose a warden for each floor.
Set up the assembly-point screen link on the same page and open it on a tablet or TV at the assembly point.
To let lobby TVs show a roll call, turn on Take over this board during a roll call on their TV Dashboard boards (off by default).
"Are you safe?" reaches people through their app account: link app users to their person record and give them an email or phone.

The board

Still to find first: the people not accounted for, grouped by the floor their badge was last on, with that floor's warden. Mark safe, Excuse… (with a reason that goes in the report) and Track on each.
Wardens press Mark floor swept when their floor is clear.
The time against the drill's target, and the curve of people accounted for. Everyone shows how each person was counted.

"Are you safe?"

One press sends everyone still to find a personal link by email, the Omaya app, SMS or WhatsApp. I'm safe accounts for them; I need help puts them at the top of the board in red and alerts the person running the roll call and their floor warden.

Starting and readiness

One press starts an evacuation roll call or a fire drill at the usual gathering point. Drills can have a target time.
Ready for the next alarm? checks the assembly point's gateway, badges gone quiet, floors without a warden, the next drill and who can be asked "Are you safe?".

Screens

The assembly-point screen shows the clock until a roll call starts, then the count and the names still to find; numbers only if you prefer.
TV Dashboard boards can be taken over during a roll call, numbers only.

After the roll call

Everyone out in (against the target), 90 % out by (against the last drill), floors swept, the arrival curve and the last people found. The report's timeline names each person marked and gives the reason.

2026-10 · Meeting rooms at a glance, and room calendars both ways

Date: 2026-10-12 Theme: Meeting Room redesigned around the question people ask — which room is free now? — with each room's day one tap away, booking in two taps, extend and end early, door displays people can use, and Google Workspace / Microsoft 365 room calendars linked both ways. Administrator action:
Calendar sync moved to the admin console: Your Apps › Meeting Room › Open Calendar sync. It is no longer a tab in the app. To link Google Workspace or Microsoft 365, follow the new guide Integrate › Linking room calendars.
One redirect address for every organisation: register https://<your Omaya address>/admin/apps/meeting-room/calendar/callback once on the Google OAuth client or Microsoft app used for room calendars.
Door displays need their link: open Meeting Room › Rooms & displays, press Set up for each room with a tablet and open that link on the tablet. A tablet on the old address keeps showing the room but cannot book.

Rooms now

One tile per room, grouped by floor, with what it is doing now: Free until …, In use, Booked, not checked in, Booked at … or In use, not booked (people in a room nobody booked). Counts across the top are also filters, and a Seats filter finds a room big enough.
Press a room to open its day: a time line with a now-line, the list of meetings, and 15 min / 30 min / 1 hour to book it on the spot. All rooms' day shows every room on one time line; press a free spot to book it.

Booking

+15 / +30 min and End now on a booking. Ending early frees the room straight away.
The booking form offers a smaller free room when the one you picked seats far more people than are coming.
Report a problem in a room (screen, video, too hot or cold, lights, cleaning, furniture) becomes a Facility job on that room, once per problem while it is open.

Door display

A new full-screen layout for the tablet by the door. With the room's link people can check in, extend, end, book 15 / 30 / 60 minutes and report a problem there; without it the display only shows the room.

Google Workspace and Microsoft 365

Each meeting room is linked to its Google room resource or Microsoft 365 room mailbox. Rooms booked in Google Calendar, Outlook or Teams show in Omaya; rooms booked in Omaya show busy there and invite the people on the booking.
When nobody turns up and Omaya releases a room, the room declines the meeting in Google or Outlook too, and the organiser is told. Extending or ending a meeting in Omaya moves it in the calendar.

Insights

Trends is now Insights: booked time actually used, bookings nobody came to, people per meeting against seats, meetings that got stuffy (CO₂ over 1000 ppm), when rooms are booked by weekday and hour, figures by room, and plain suggestions.

2026-10 · Badges that find their room, rules that watch events

Date: 2026-10-11 Theme: A badge with its own SIM, placed in a room by the beacons it hears; rules that watch people arriving, gateways, incidents and work orders; labels with fields of their own, and roles limited to them; an assistant that checks whether its changes worked; and automatic work orders you can switch off by source. Administrator action:
Look at Admin › Your Apps › Facility › Work orders opened automatically. All four sources are on, as before; untick any you do not want jobs for.
Two pages moved in the menu: Beacon & Wi-Fi Anchors is now under Device, Data tables under System. Their addresses did not change, so bookmarks still work.
If you use MOKO PS101-M badges: keep them on JSON, and add anchors for the rooms you want them found in.
Every gateway accepts commands, as before. To keep a gateway read-only, edit it under Device › Gateway and turn off Accept commands.

Change a sensor's settings from Omaya

On a MOKO S01 sensor's Sensor settings card, Change on the sensor now sets its PIR delay and sensitivity through a Dusun gateway that hears it. Omaya connects, gives the sensor's password (kept encrypted), writes, reads the values back and disconnects, then checks the sensor's own broadcast. No trip with a phone.
Each change is listed on the card as running, done (✓ once the broadcast agrees), failed with the reason, or refused. Only these two settings can be sent; nothing can switch the sensor off, reset it or update its firmware.

Read-only gateways

New Accept commands switch on each gateway. With it off, Omaya still uses the gateway's readings but never sends a command through it. A command that only a read-only gateway could carry is refused and logged, including one queued before the switch was turned off.
Every command sent to a device is now in System › Activity Log: when it was sent or refused, how it ended, who or which rule sent it, and through which gateway. A new Device commands chip shows only these, and the export follows whichever chip is selected.
A System events chip in the Activity Log shows what happened without anyone doing it: a gateway going offline and coming back, a device going silent and reporting again, a scheduled report that could not be sent or was paused, a webhook that gave up after every retry, and a connector that started failing or works again. Each is logged once, when it changes.

MOKO PS101-M badge

A badge with its own SIM talks to Omaya directly over LTE-M or NB-IoT, with no gateway: where it is, its battery, and its SOS, Alert and man-down alarms. An alarm reaches Panic Buttons and Security Watch as one episode, not a page for every message, and readings the badge stored while out of coverage arrive with their own time.
Outdoors its GPS position becomes geofence visits, trips and stops, as for other trackers. Indoors it lists the beacons or Wi-Fi access points it hears, and Omaya places it in a room (next section).
Badge settings on its device page: read and change how it finds its position, how often it reports, its alarms and more. One-click Locate now, Find (it beeps), trigger or exit an alarm, and Restart; Power off and Factory reset ask you to type a confirmation. Each command is tracked until the badge answers or it times out, and who sent it is recorded.
Rule actions Locate badge now, Sound badge and Exit badge alarm. An alarm Omaya set off itself, to find a badge, pages nobody.
Built from MOKO's protocol documents; the first badge in use is being checked against them this week.

Beacon and Wi-Fi anchors

New page Device › Beacon & Wi-Fi Anchors: register a beacon by its Bluetooth MAC address or an access point by its BSSID, and give it a room. One at a time, as a pasted list, or picked from what badges have already heard. Gateways, and beacons or sensors with a zone, count as anchors without any typing.
A device that scans instead of broadcasting is placed in the room whose anchors it heard loudest. From then on it is on the live map, in Roll Call, rules, the Entity Tracker and reports like any tag, and for a while after a scan the room wins over GPS.

Work orders

Admin › Your Apps › Facility has Work orders opened automatically: a critical battery, a device that stops reporting or behaves oddly, a gateway offline too long, and a meeting room's people counter gone silent. Untick one and the alert still appears; only the job is not raised.
A rechargeable device (the PS101-M badge, the Minew MWC01) gets a Charge job instead of a battery job, and no "replace in N days" forecast.
One open job per device and anomaly type: a door counter that kept going silent had 21 open jobs. Repeats are now counted on the one job, and a device that reports again closes its untouched "silent" job.

Rules

Scopes: Tag is now called Label, and there are two new scopes, Person / asset (follows whichever tag the person or asset carries) and Gateway. Scheduled Label and Person / asset rules now check each device or person; before, they saw no values.
New triggers: someone entering or leaving a zone, venue or location; a gateway going down or coming back; an anomaly; a work order opened or resolved; a Security Watch incident; a date in a label field, as days from today; and a fall or man-down (fall detection, below). An event rule keeps its own cooldown, and its own work order, per device or gateway.
Builder: choose what the rule watches first and search for it (a label shows how many members it has), read the rule back as one sentence with its AND / OR brackets, set the repeat interval in seconds, minutes or hours, or start from one of 13 presets.
Test replays the rule you are writing, before you save it, over 1 to 30 days of real stays.
A label rule also matches the labels on the person or asset a tag is bound to. Before, such rules saved and never fired.

Labels, people and roles

Label fields: give a label fields of its own (the Label Fields button on the Entity page), such as a licence expiry or a department, and every person or asset carrying the label shows them on its form. Search them in the grid, and read or write them over the API (custom_fields, ?custom[key]=, GET /entity/fields).
A label can apply to several kinds of record: venues, zones, gateways, devices, people and assets, set with Used for on System › Labels.
Entity access on a role limits it to the people and assets carrying chosen labels: on the Entity pages, the Entity Monitor, Right Now, the dashboard and the pickers. Reports and the public API are not limited yet.
Label filter on Heatmap (zone presence), Flow, Button Press, Outdoor Activity and Dwell Time.
The Entity page has Label, Type and Status quick filters. A search in any grid matches anywhere in the text, and the all-fields box also searches label names, sensors, descriptions and label-field values.
Top-bar search also finds devices, gateways, people, assets, zones and venues by label name; a person or asset opens straight in the Entity Tracker.

OAIA

The page is redesigned: the briefing first, then what needs acting on, then Ask. The tiles and the list now count the same open findings.
Apply from the page: a suggestion that carries a ready change, such as a room's capacity, is applied from its card, with the current value beside the proposed one.
Did it work? After a change is applied, by a person or by Autopilot, OAIA looks again later and says whether it fixed what it was for.
In the chat: suggested follow-up questions after each answer (buttons on Telegram); an alert rule you ask for, created once you approve it; make a room a toilet in Toilet Hygiene; and apply a gateway's receive offset measured on the Calibration Bench, then check the room bounces a day later.
Once a day it rechecks older suggestions and closes the ones whose problem has gone. Silence or a flat reading counts as unsure, never as resolved.
The likely cause written on a finding's first day is asked for again after three days; a battery at 30 % or more is never blamed; and an energy counter that stopped rising while power is drawn is flagged.

Floor plans

Import a CAD drawing (DXF): walls, rooms with their names, and doors are taken from the drawing. A DWG, IFC or Revit file is exported to DXF first, and the upload says how.
Auto-extract measured the scale wrongly on some floors, by up to half. Room areas and door widths are now right, and narrow real doors are found.

Energy

A Home Assistant plug's kWh stood still for weeks: when power and energy arrived in the same minute, the energy reading was dropped. Every reading is now kept, and the plug's current and voltage are read too. A plug Home Assistant reports as unavailable no longer looks alive.

Fall and man-down detection (preview)

Falls and man-down detected from the accelerometer in a worn tag, with no calibration. It learns how each tag is usually put down, so a tag left on a desk the same way is not a fall. It can ask the person "Are you OK?" and, without an answer, open a critical Possible fall incident in Security Watch for the guards; rules can watch it too. Rest places (a sick bay, a rest room) wait before asking; a toilet never can be one.
In preview: switched on per organisation under System › Configuration › Fall detection, and it only records until alerts are turned on there. Reports › Fall & Man-down shows what it flagged and, once events are labelled, how much was real.

Data tables

The list is a grid like the other pages, with a View panel (rows, the rules that fill the table, its columns), New and Edit beside the list, and a reason when a table will not save, such as a reserved column name like count.
A zone column holds the zone; a rule that wrote the zone's name is still grouped correctly.

Trails

Split long stays at midnight (System › Configuration › Indoor Positioning) works again: a trolley parked for three days is three rows, and a stay that began before the chosen day shows on it. Device Trail lists the parts in order, and "still here" is the stay that ends last.

Toilet Hygiene and the phone

Who gets notified has See every notification: each message a job can send, who gets it, and the words on the phone. People told by a step now also get the "done" follow-up.
Profile › Notifications fits a phone; on an iPhone it says to add the app to the Home Screen first, and how.

2026-10 · Choose who is told about a toilet job

Date: 2026-10-08 Theme: Each toilet can say which cleaners and admins hear about a job, and who next when no one responds; the edit page lists everything that raises a job; every date in the mini app follows the user's Date format. Administrator action: Ask cleaners and supervisors to allow notifications in the mini app (Profile). A cleaner who has not done so receives nothing.

Toilet Hygiene

Who gets notified (toilet › Edit): Every cleaner (as before) or Choose people, step by step — Step 1 when a job is raised, then any number of steps after a wait of your choice, each telling chosen cleaners and admins, then Missed. A job stops moving once a cleaner responds.
What raises a cleaning job (toilet › Edit): cleaning rules from Schedules, busy, dirty score and bad review in one list, with links to change them.
The daily board's "What happened" shows each step, who it went to and how many phones took it.
Notifications in each person's language. Every Toilet Hygiene push (new job, still waiting, step alerts, missed, reopened, closed, already handled, overdue) is sent in the language the recipient uses the app in: English, Bahasa Melayu or 简体中文. The language is now kept on the person's account, from their Profile.

Mini app

Date format on the Profile: DD/MM/YYYY (default), Day Month Year or YYYY-MM-DD, for every date in the app — date boxes too, whatever language the phone uses.

2026-10 · Reports you can trust about the past

Date: 2026-10-06 Theme: Omaya now remembers who had which tag, when gateways went down and when assets moved; panic buttons stop crying wolf; and the AI can answer from the new reports. Administrator action:
Give the rooms that are not offices a compliance profile under Space › Zones (a toilet, a server room, a cold store) — every room is checked as Office comfort until you do.
Panic buttons: a gateway that keeps producing corrupted frames needs a firmware check; Minew tags that still broadcast their factory-default beacon and should be given another major / minor in BeaconSet+; an H8 button lying on a desk raises man-down every few minutes — switch its inactivity alarm off or wear it.

Remembering the past

Tag binding history. Omaya now remembers which tag was bound to which person or asset, and when. Reports about past days use the binding at the time: Panic & Incident Response names who carried a button when it was pressed; Footfall counts a tag moved from a person to a trolley as a person before and an asset after; Attendance and Asset Utilisation count the tags bound during the range. History before 6 October was rebuilt from the Entity Tracker trail.
Gateway outages are logged as they happen, from the gateway's last packet to the first one after, whether or not anyone was alerted. Gateway Availability uses the exact times from 6 October (downtime to the minute, a "partly down" cell, exact start and end on hover); earlier periods stay to the hour. It already shows an 80-minute LoRaWAN outage on 5 October that the hourly figure missed.
Asset motion. For tags bound to assets, Omaya keeps the minutes each tag said it was moving. An asset pushed about inside one room now counts as in use, not idle, and each asset says whether its figure comes from motion or from location. People's motion is not stored.

Panic buttons

Of 173 panic incidents on our test site between 14 September and 5 October, 162 were detection errors, all now fixed: a factory-default Minew beacon read as a double-click (152), corrupted MOKO frames (6), tag motion paging (2) and, fixed today, a gateway attaching a neighbouring tag's beacon to a temperature sensor (1). A device without a button can no longer report a press, and moving a tag never pages anyone unless a routing asks for motion. The other 11 were real: button presses, and a man-down timer on a button nobody was wearing.

Environment

Compliance checks each room against its own profile instead of one range for the whole site that defaulted to a 2–8 °C fridge, which made every office breach all day. Profiles: Office comfort (23–26 °C, 40–60 %RH, CO₂ up to 1000 ppm, noise up to 55 dBA, opening hours), Cold chain, Frozen, Controlled room, Server room and Custom, set per zone, in bulk from the tab, or as a venue or organisation default. Any date range; hours with no data are shown and never counted as breaches; mean kinetic temperature only for cold stores. The scheduled Compliance report and the Monthly site pack use the same profiles.

AI and the API

AI Reporting and OAIA answer from Attendance, Asset Utilisation, Panic & Incident Response and Gateway Availability — "who was late this week", "which trolley is missing", "how fast do guards answer panic presses", "which gateway was down most last month" — with the reports' own figures, and only for roles that can open those reports.
Public API: GET /api/v1/assets/utilisation, /api/v1/incidents/response and /api/v1/gateways/availability.

Scheduled reports

An email that cannot carry its report (too large for the mailer, or a mail route that cannot attach files) now has a download link valid for 7 days instead.
The Customer Monthly Summary is replaced by the Monthly site pack; existing schedules were converted with their recipients, time and format, and new organisations get a pack.
The Data Centre report can be scheduled for one hall; a hall report keeps site-wide UPS units, generators and alerts in their own section.

Phone

Mini apps with scheduled reports have a Scheduled reports screen for people who can change that app's settings: see, pause, resume, test or delete. Security Watch reads its outcomes from the server, so Test and Duplicate appear without an app update. (In the next app update.)

Languages

The Reports menu, Days counted and Schedule Report are in Bahasa Melayu and Chinese.

2026-10 · A report for every part of the platform

Date: 2026-10-06 Theme: reports for assets, guards, gateways, the warehouse and the data centre; operating days everywhere; and scheduled reports that the people running each app can set up themselves. Administrator action:
The new reports have their own permissions (View Asset Utilisation Report, View Panic & Incident Response Report), already given to the roles that see the comparable reports. Check System › Roles if someone should not see them.
Scheduled report emails now carry the report. They never did before, although they said so. Through a mail relay that cannot carry files, the email says the file is missing.

New reports

Asset Utilisation (Reports › People & movement): per asset and per group, hours seen, moves, time in use away from its base, utilisation of opening hours and where it spent its time; idle assets, and missing ones (not seen for 1, 3 or 7 days, with where and when last seen). Use is read from location changes, not motion.
Panic & Incident Response (Reports › Analysis, and a Reports tab in the Security Watch app with the same figures): incidents by type, place and hour; how fast a person picked them up and resolved them; still open, unattended and reopened; outcomes and the false-alarm rate; each guard's workload; each panic button and who carries it, with the presses nobody picked up.
Gateway Availability (Reports › Sensors): each gateway's availability against an SLA target you set, of all hours and of operating hours, with outages, downtime, an hour-by-hour timeline and totals by location and floor. Read in whole hours.
Warehouse › Reports: orders received and shipped, on-time %, order cycle time and backlog; lines picked, picked in full and picks per hour per picker; stock by storage class, slow movers, aging, stock-outs and bin use; count accuracy. What the app does not record reads not recorded, never 0.
Data Centre › Report: rack inlet temperature and humidity time in the recommended, allowable and out bands, with worst racks and excursions; PUE and DCiE from metered energy; UPS time on battery and events; alert response; rack and hall power against rating.

Security Watch

Guards can resolve an incident as Test (a test press or drill) or Duplicate (already handled as another incident), besides Genuine, False alarm and Expected. Neither counts in the false-alarm rate. The phone app shows them after its next update.
Rule & Alert Effectiveness keeps counting an incident the sweep closed as unattended as acknowledged. Panic & Incident Response counts only a person picking it up, so it reads lower by those, and both pages say so.

Days counted

Space Utilisation, Zone Compare and Impact Analysis have the same Days counted choice as Footfall and Attendance: Operating leaves closed days out of the per-day figures and says what they held; All days counts every calendar day. In Impact Analysis, a weekend inside a "before" or "after" window no longer looks like a drop. Its Export button works again.
Dashboard: Visitors This Week (per open day), Visitors Today (against the last open day), Visitor Trend and Top Locations / Venues follow your organisation's operating week when it has one.

Monthly site pack

System › Schedule Report has a new Monthly site pack: one PDF for the last calendar month across the whole platform. A cover (data coverage, days counted, contents), then one short block per area — footfall, door entries, attendance totals (no names), assets, space, environment, gateway availability, rules, panic and incident response, and each switched-on app's summary — each against the month before where it can. Blocks follow the permissions of the administrator who saved the schedule; those they cannot see are counted in a closing note. Best sent monthly on the 1st. The Customer Monthly Summary is unchanged.
Asset Utilisation, Gateway Availability (against a 99 % target) and Panic & Incident Response can be scheduled too, by roles that can open those reports.

Scheduled reports in the apps

Facility Management (work orders, planned maintenance compliance, vendor activity), Toilet Hygiene (cleaning and feedback), Visitor Management (visits), Energy (per meter), Security Watch (incidents and response, following the creator's own scope), Warehouse (KPIs; stock figures as of the send) and Data Centre (period report, the whole site) can be emailed daily, weekly or monthly as PDF or spreadsheet.
People who can change an app's settings set them up in that app, under Reports › Scheduled reports, and can send them to the app's users or any address. A schedule made in an app holds only what its creator can see there, checked at every send; it pauses itself, saying why, if the creator loses that right, and skips recipients who can no longer open the app.
Administrators set them up under System › Schedule Report, where app reports appear grouped by app, and see every schedule made in an app as "App — made by name", to pause, resume or delete.

2026-10 · People, not trolleys

Date: 2026-10-06 Theme: one "people today" figure everywhere, Footfall for your own systems, and rules that watch how busy a place is. Administrator action: none. To keep a tag on a trolley or a wheelchair out of the people figures, bind it to an asset on the Entity page.

Reports

People heard, asset tags apart. Footfall used to count every registered tag, so a tag on a trolley was a person. It now counts the tags people carry and shows tags bound to assets beside them as "+ N asset tags". The rule is the TV Dashboard's, so the office dashboard's Visitors Today and the lobby TV's People seen today now show the same number. The dashboard's visitor tiles and chart, the scheduled Footfall attachment, the monthly summary, Energy per person and AI Reporting follow. The last 30 days were recounted.
TV Dashboard: People seen today reads Footfall when the widget covers the whole organisation, one venue or one zone. For several places or chosen people it still counts from the trail, because places cannot be added up.
Custom Reports can pick people and asset tags from the Footfall table.
Attendance counts operating days. Attendance has the same Days counted choice as Entry / Exit: Operating counts only your open days, All days every calendar day. Each person shows days present out of the days counted with an attendance rate, and a new By day table gives people present, attendance, average time on site, first in and last out for each day. Visits on closed days are kept and marked, never dropped silently. The CSV says whether each day was an operating day.
Footfall counts operating days too. The same Days counted choice: Operating leaves closed days out of the chart, tables, people a day, visits, busiest day, per 100 m², typical week and comparison, and a note says how many people-days and door entries the closed days held. All days counts every calendar day, as before. A venue in the filter uses its own week.
Late, left early and not in today. With opening hours set, Attendance marks a day late when the first in is more than a margin you choose (15 minutes by default) after opening, and left early when the last out is more than that before closing, per person and in the By day table. It never marks a stretch the gateways were not sending, so an outage at 9 am does not make everyone late. Not in today lists who has not been seen yet, and who has no tag bound and so never can be.

Scheduled reports

Attendance by email. System › Schedule Report can send Attendance (spreadsheet or PDF) for the last full days: one row per person and day on site, then each person's days present out of days counted and their rate. Days counted follow the operating week (the venue's, for a schedule limited to one venue). Only a role with View Attendance Report is offered it, and it goes only to chosen administrators whose role has that permission; anyone left out is noted on the schedule's card.

Integrate

Footfall and Attendance on the API. GET /api/v1/footfall gives people, asset tags, entries, exits and the most inside per place and day or hour; GET /api/v1/attendance gives each person's first in, last out and time on site. Same key and headers as the other endpoints. See Integrate › Reading data out. /attendance takes days=operating|all (default all) and returns each person's rate against the days counted.

Rules

Rules on how busy a place is. Three new metrics for a location, venue or zone, on a scheduled rule: Footfall Today (people so far), Footfall vs Usual (today so far against the same hours on the last four same weekdays, as a percent) and Most Inside Today (% of capacity) (from the door counters). A day the gateways sent nothing is never read as an empty building. See Your first rule › Rules on footfall and capacity.
"Not checked" in the rule history. When a scheduled rule cannot be checked because its figure is missing (no gateway data today, before the first full hour, nothing to compare against, no door counter, no capacity), Monitor › Live Monitor › Rules shows a muted "Not checked — reason" line, once a day per reason. It is not a firing: nothing is sent and it is not counted as one anywhere. Rule & Alert Effectiveness lists it under the rule as "Not checked on N days", and the Rules list marks such rules next to Last Fired.

AI

OAIA's Footfall view gains people and assets, and OAIA is told to answer with people.

Toilet Hygiene

Reports in tabs. The long Reports page is now one tab per topic: Overview, Visitors, Cleaning, Cleaners, Guest reviews and Downloads. The period you pick stays when you switch tab, and each headline number on Overview has See details.
Visitors trend. The new Visitors tab shows people going into the toilets over time, by Hour, Day, Week or Month, as bars or a line, all toilets together or each toilet. It adds the busiest day, time and toilet, the average for each hour and weekday, each toilet's share, and the numbers as a table with Download CSV. Pick one toilet to see only its trend. It counts entries only, and only the toilet door sensors; the Overview Visitors figure now counts the same way, so a sensor elsewhere in the building no longer adds to it.
Service level. A new Service level tab in Reports shows whether a cleaner arrived within each toilet's Job is late after time once a job was raised, how long they took to arrive and how long they cleaned, per toilet, per cleaner, per day and per job, with Download CSV. Arriving means the cleaner's badge stayed inside at least 2 minutes; shorter visits show as First seen. Jobs it cannot judge (still within time, cancelled, closed with no cleaner stay recorded) are listed, not counted. The Cleaners table's On time is renamed Finished on time, which is what it always measured.
Automatic cleans are switched on per toilet. Two new switches in a toilet's Edit → Advanced settings, both off by default: Send a cleaner when the dirty score passes its limit (uses the toilet's Dirty score limit, checked every minute, with the busy-hour clean's safeguards) and Send a cleaner after a 1–2 star guest review. A bad review no longer raises a job unless that switch is on; reviews still show in Reports. When it is on, the job now reaches the cleaners' phones straight away instead of about 30 minutes later. No admin rule is needed for either.
Deleted people stop being cleaners. Deleting a person under Manage › Entity now also takes them off the Toilet Hygiene cleaner list. Before, they stayed on the Staff page and their app login kept receiving every cleaning job.
Menu order. Toilets, Reports, Work orders, Live, Schedules, Staff: set-up screens last. In Reports: Overview, Service level, Visitors, Cleaners, Cleaning, Guest reviews, Downloads. Reports open on today, with Yesterday, 7, 14, 30 and 90 days as quick picks under the dates.
Home page. The Manage › My profile row is gone; the profile is in the header and the bottom bar.
Staff app opens in Light. The staff app now opens in the light theme unless the person picked one under Me › Appearance. Follow device is still there.

2026-10 · Know when data is missing

Date: 2026-10-06 Theme: reports that say when the site sent nothing, a fair way to compare places, and scheduled reports that actually arrive. Administrator action:
Scheduled reports with no recipients chosen now go to your first administrator, so check System › Schedule report if that person should not get them.
An administrator account with no email address cannot receive them, so give each one an email.

Reports

No data is not nobody. When a place's gateways send nothing for a day, usually because of power or network at the site, Footfall and Attendance now say No data received instead of showing nobody there. A day with data for less than half its hours carries a "data for N of 24 h" mark, those days are named above the table, and the dashboard chart leaves a gap.
Comparing places of different sizes. Footfall's totals now show devices a day per 100 m² of floor beside "most inside" as a share of capacity, so a large floor and a small one can be compared fairly. A location's capacity is its venues' capacities added up.
Footfall counts registered tags. Devices you have not registered are dropped when they are heard, so they were never counted, although the page said otherwise. Visitors without a tag appear only in the door counts.

Scheduled reports

No recipients chosen now means your first administrator. Previously such schedules never sent, including the monthly summary and alert digest created for new organisations. The form says who it will go to.
Venue: a schedule can be limited to one venue.
Fixes: Pause / Resume works (it was refused), Test now works (it failed to start), and editing a schedule saves (the time came back in a form the save refused).

AI

OAIA can read Footfall directly (oaia_report_footfall), once a database administrator has run the one-off grant.

2026-10 · Reports that answer more

Date: 2026-10-05 Theme: two new reports, Footfall that compares and explains, and the reports that showed wrong numbers put right. Administrator action: optional. Give each venue a capacity on its form (Space › Venues) if the sum of its zones' capacities is not the real figure. To have Footfall, Entry / Exit, People Count or Dwell emailed, add them under System › Schedule report.

New reports

Attendance (Reports › People & movement): per person and day, first in, last out and time on site, with room stays joined up and the gaps you choose bridged. Also time on each floor, and who is here now. Click a name for their days and stays. CSV export.
Rule & alert effectiveness (Reports › Analysis): per rule, how often it fired, on how many devices, and whether its actions went out or failed, with the error. Rules that never fired. For anomalies, battery alerts and Security Watch incidents, how many were acknowledged, the median time to acknowledge, and which have been waiting longest. Rule firings are kept from 4 October 2026, for 90 days.

Roles

Attendance and Rule & alert effectiveness have permissions of their own: View Attendance Report and View Rule & Alert Effectiveness Report. Every role that could see Loyalty or Alerts was given them, so nobody gained or lost a page. Attendance shows named people's times, so it can now be given or withheld separately.
The Roles page names every permission:
- Footfall's permission is called View Footfall Report.
- Movement Flow and Distance are listed under Reports.
- A new AI & Assistant group holds OAIA, AI Reporting and AI Reporting may read … for each app.
- Data Tables and Platform Usage have their place.
- None of these sit in an unnamed list any more.

Footfall

Compare with the period before or the same dates last year: change on the headline cards, the earlier period on the chart, and its row in the totals.
A typical week: average devices heard and people through the doors, each weekday and hour.
Time here: how long a tag stayed in each place, on average, its stays joined up.
Against capacity: the most inside at once as a share of what the venue or zone holds. Venues now have a capacity of their own; if it is empty, their zones' capacities are added up.
Per device heard: door crossings for every device heard, a rough sign of how many visitors carry nothing.

Other reports

Zone Compare now also compares one place over two periods, such as this week against last or this month against last year. Its figures are now right:
- entries come from door counters (they were hourly head counts added up)
- devices are a daily average (they were added up across days)
- people present never mixes in occupied/empty flags
- busiest hour and weekday are on your clock (10:00 showed as 02:00)
- time at capacity reads each hour's peak
Movement Flow follows people between floors: pick All floors to see trips floor to floor or room to room, the typical time between floors, and the most common routes through a visit. A slider ignores positioning bounces shorter than a few minutes. The single-floor view's device times are now local (they were 8 hours off).
Loyalty filtered to a venue or zone counts a device on every day it was there. It used to credit each day to the first room only: one server room read 14 device-days instead of 39.
Custom Reports:
- offer Footfall as a table
- People Count has a true average and the hour's peak, and leaves out occupied/empty flags
- averages over two months are exact
AI Reporting no longer reads occupied/empty flags as head counts.
Energy: energy per person (from people counted in the rooms, else visitor-days), and how much was used while open but empty.

Scheduled reports

Footfall, Entry / Exit, People Count and Dwell can be scheduled by email as CSV or PDF, for the last full days.
The monthly summary emailed to administrators was mostly empty or wrong:
- rules fired and top zones read tables that do not exist
- detections covered a single day on the 1st, and added tags, people and occupied flags together
It now reports, for the last 30 days: visits, door entries, most inside, people and tags per room, rules fired, and the top zones.
The alert digest counts only your organisation's rows.

2026-10 · Footfall

Date: 2026-10-05 Theme: how many were in a place each day, counted once per place, and the visitor numbers everywhere else brought into line with it. Administrator action: none. The last 30 days are already filled in. Bookmarks to Devices Detected open Footfall.

Reports

Footfall (Reports › People, replacing Devices Detected): for the whole organisation, a location, a venue or a zone, day by day or hour by hour, three figures side by side:
- devices: your registered tags, each counted once per place however many rooms it went through. A device you have not registered is never stored, so visitors without a tag show only in the door counts
- the door counters' entries
- the most inside at once
- Rows can break the place down by location, venue or zone, with CSV export.
The table reads one day at a time. Each day opens with a row for the whole place you filtered on, with its venues (or zones) indented beneath it. Each column shows one number with its detail underneath: devices heard (tagged · untagged, with a bar), came in (and how many left), and most inside (and at what time). A place without a door counter says so instead of showing 0, and a day with nobody on one quiet line. A short legend above the table says what each column counts.
Totals for the range open the table: for the whole place and for each venue (or zone), the visits (each day's devices added up, and the average a day), how many came in and left, and the most inside on any day, with when.
Per hour reads the same way: each day opens with its whole-day totals, then every hour has a row for the whole place with its venues beneath it. Hours run morning to night, and empty hours are left out.
Why it replaced Devices Detected: that report added up its zones, so someone who walked through four rooms counted four times. On one floor it read 16 for 10 devices.
Older days: the once-per-place count needs the day's device list, which is kept for 30 days. Before that, a zone's own figure and any hour are still exact, but a venue, location or whole-organisation day shows — rather than a sum.
Detection Heatmap by venue or location reads the same once-per-place counts.
Impact Analysis rates a venue, location or whole-organisation event on them. Days older than 30 days are left out of the per-day rate.
Custom Reports no longer add up Devices Detected's day rows across zones: group by zone or by hour, or use Footfall.

Dashboard

Visitors Today, Visitors This Week, the Visitors chart and Top Locations / Top Venues were several times too high. They added up every zone's row and every hour's row for the day (about 12 times on one site). They now show Footfall's figures, and today's refreshes every 15 minutes instead of showing yesterday's.

AI

AI Reporting's device tool answers with the Footfall figure, and gives door entries beside it for "how many visitors". OAIA is told not to add up zones for a venue or a site, and points to Footfall.

Also

Panic buttons: SMS and WhatsApp now arrive. They never sent before. They go out through the organisation's SMS and WhatsApp connectors, and a panic routing can name app users as well as administrators. App users get email, SMS, WhatsApp and push; administrators get email and in-app.
TV Dashboard: each board's share link has a Link expiry in the app (days until it stops working, empty for never). The separate admin settings page is gone.
Apps menu: Module settings is gone, because the gear on each app's Your Apps card opens its settings. Meeting Room settings no longer send tenant administrators to the App Store, which they cannot open.
A Facility photo whose files are gone shows as no image instead of an error.

2026-10 · Warehouse, and the world outside

Date: 2026-10-05 Theme: a warehouse app run on scans; GPS trackers followed outdoors as well as indoors; Energy totals you can bill against; a TV widget for every app; rules that log every firing; and every mini app on one layout. Administrator action: Warehouse is off until switched on in Your Apps (enterprise plan). For Energy cost, enter a rate per kWh under Your Apps → Energy (no rate, no cost shown), then set each sub-meter's parent and mark any solar or generator meter as Generates power under Meter hierarchy. GPS behaviour (stop time and radius, indoor grace, fix quality) is on the new Outdoor & GPS card under Configuration; the defaults suit most sites. AI Reporting's per-app permissions are given to the Admin role; grant them to other roles under Roles. Creating or changing a TV board now needs the TV Dashboard configure grant. Admin sign-in no longer offers Remember me.

Warehouse (new app, `Your Apps → Warehouse`)

A warehouse run on scans. Receiving with QC sampling, putaway with bin suggestions and storage-class guards, picking with FEFO / FIFO reservation and a guided next unit, outbound QC, packing and shipping. Also counts with approved variances, transfers, quarantine, work orders with material issue and returns, and gate passes. Every box is a labelled unit with a QR code, and every change is one line of a ledger that is only ever added to.
Returns checked against what was shipped. A customer can return only what that order shipped, less what has already come back. Return to supplier goes out on its own gate pass. Goods received notes, delivery orders and gate passes print on A4 with signature boxes.
A tablet floor, a desk and the phone. Each floor task appears in the mobile app, and every action goes through the same checks and ledger as the tablet.
ERP in step. Items, partners and orders come in through the API or CSV, and are pulled again each night through the new ERP connector. Every movement goes back out as a webhook or a movement feed the ERP can read from where it left off.
What Omaya adds: tagged pallets and forklifts on the live map, flagged misplaced when the tag disagrees with the bin. Also cold-chain exposure per unit from the area's sensors, dock occupancy and dwell, and goods overdue on a gate pass or moving after hours raised in Security Watch. OAIA hears the app and can propose QC sampling rates or open a count.

GPS and outdoors

One trail, indoors and out. A GPS tracker outdoors is recorded as geofence visits, trips and stops instead of a row every few metres. Indoors the Bluetooth room still wins. A tracker that goes outside ends its indoor stay at the last Bluetooth sighting, so nobody reads as still in the pantry while driving.
Geofences on the map. Draw, rename, reshape, switch off and delete them under Space → Location → Map, beside a searchable list. Frequent stops lists the places trackers keep stopping at that have no geofence yet; one click makes one. Stops already recorded there over the last 90 days become visits.
Where they are now. Entity Tracker has an Outside group with a street map. Live Monitor shows Travelling · 1.4 km or the stop instead of a home zone. Movement Replay draws the day's route. Facility's live staff and asset locate say Outdoors (GPS).
Rules and webhooks on a geofence. A rule can be scoped to a geofence and fire on entered or left, once per tracker. Subscribers get geofence.entered and geofence.exited.
Panic and man-down say where: the room indoors, or the GPS position and a map link outside.
Roll Call: a person outside on GPS is off-site and accounted for, not missing, with a map link on the board and in the phone app. A muster can also gather at a geofence, such as a car-park assembly area, beside or instead of an indoor zone: anyone whose GPS tag reports from inside it counts as arrived. It is offered only where people carry GPS tags. A rule's Start Roll Call action can name it too, and if it can't be used when the rule fires, the roll call starts without it rather than failing.
Outdoor Activity report (Reports → Sensors): time outside, distance, trips, stops and geofence visits per tracker and day, and per geofence. CSV export.
Cellular trackers that report their own position over MQTT (MOKO PS101 and similar) are accepted, with a setup guide.
OAIA and AI Reporting answer where is X and how long was the van out from the same trail. The public API adds whereabouts, GPS points, outdoor activity and geofences.

Energy

Meter hierarchy. A meter can name the meter that feeds it, so a main incomer and its sub-boards are counted once, at the top.
Generation offsets the total. Mark solar or a generator as Generates power: every total is net (used X · generated Y), and below zero reads as exporting.
Cost today and this month, from the rate per kWh, on the Overview, the Meters page, the CSV and the TV headline. Energy rules also measure the net figure.
A real Overview (meters reporting, energy today, power now), a styled Meters grid, and kWh added (24h) from counter rises. It used to sum each meter's lifetime counter.

TV Dashboard

A widget for every app: meeting rooms, facility, security, roll call, warehouse, data centre, energy and compliance. The picker offers only the apps that are on. Public boards still show zones and counts, never a person.

Rules

Every firing is logged, with the reading, each condition and what each action did. Live Monitor → Rules and Right Now's live alerts now show them.
A Tag rule's cooldown is per device, so one tag firing no longer silences every other tag. It raises one work order per device and names the device in its messages.
Rules with a held for time on a tag or zone are re-checked every minute against each device's own readings.
The builder says how it joins conditions: AND / OR join to the conditions above, evaluated left to right, with the minimum cooldown of 30 seconds shown.

Apps

One layout for every app: the same page, section bar and header across Meeting Room, Visitor, Facility, Compliance, Security Watch, Roll Call, Energy, Data Centre, Warehouse and TV Dashboard. Most apps now also come in Malay and Simplified Chinese.
Toilet Hygiene:
- a plain-language toilet page with a to-do list, forecast and AI summary
- a schedule board for cleaners
- per-toilet escalation settings
- main and backup cleaners per toilet
- reports rewritten with In short and What to do next, and their wrong figures fixed
- a job that escalation gives up on becomes missed instead of staying open for ever and blocking new triggers
Security Watch: a restricted zone's Allowed tags now admit the people who carry them. Before, every tagged person was raised as an incident. Incident evidence reads in plain words and local time, on the web and the phone. Other apps can raise incidents, starting with Warehouse.
Roll Call: Badge reliability on the after-action report, worked out when you ask for it.
Meeting Room, Visitor, Facility, Compliance and Data Centre: colleague search, approval and scanner fixes, request photos, and schedules that start this month. Times are local throughout. Every query is kept to its own organisation.

Positioning

Floor model. The floor a device is on is now worked out from the evidence over time, not from one yes / no test at a time. A per-organisation switch, off by default.
A gateway that goes quiet for a minute no longer hands its tags to a faint neighbour at once. The same challenger must win two sweeps in a row.
A corner gate stops a tag beside a gateway on a three-room corner from swapping rooms, and hold the room while a tag lies still is a per-organisation switch. Both are under Indoor Positioning → Rooms.
Why here? explains the newer decisions in plain English. The calibration walk shows its report on the desktop page, and the Map Editor's reference-line tab is now called Scale.

Reports and AI

Entry / Exit: people inside, counted as entries minus exits that resets each day, now and at peak. A missed exit no longer carries forward for a week.
AI Reporting reads your apps' data: meeting rooms, visitors, work orders, toilets, security, compliance, roll call, energy, data centre and warehouse. It is read-only, and covers only apps that are switched on and that the asker holds the permission for.
People Count shows head counts only: yes/no occupancy sensors no longer appear in it as a count. Loyalty counts only the days the site is open, so each device's figures agree with the report's totals.
Custom Reports offer Noise, Air Quality and Button Presses. Rule Write Records, which was always empty, is retired.
OAIA Autopilot: a change applied from the chat keeps its before-state, so Undo restores it, and it follows the same limits as the heartbeat.

Single sign-on works end to end, and an account owing two-factor setup or a password change can no longer open admin pages by URL. Remember me is gone from admin sign-in.
App sign-in locks out one person, not one office: five wrong passwords lock that person, and the shared office Wi-Fi stays open. Forgot password is always shown. App Users can be sent a sign-in link by email.
Dashboards: each administrator picks their own default. Changing the default for everyone, or the System dashboard, needs Manage General Config.
An expired session says Session expired instead of a bare token error.
An unused, unauthenticated location-report endpoint was removed.

2026-10 · The calibration bench

Date: 2026-10-01 Theme: Measure how loudly each tag model transmits, guided and checked, so every kind of tag is ranged as accurately as the one your floor was calibrated with. Administrator action: once per tag model, measure it on the calibration bench (about twenty minutes, see the guide). Nothing changes until you do.

Devices

Calibration bench. Devices → Sensor → Tag Power → Measure. Lay the tags beside the reference tag near one gateway and run a few one-minute rounds, turning the tags between them. The bench shows live readings while a round runs, then a verdict for each tag: how many decibels it reads against the reference, how well the rounds agree, and how wrong its distances were before. It saves only a result that agrees with itself, and counts only frames the gateway really heard again. Tags and gateways are listed with their MAC address and the room each gateway is assigned to, and the bench ticks the gateway nearest the reference tag for you. Guide: /docs/guides/calibration-bench.
One way to measure. The old Measure a model panel on Tag Power is gone; Measure opens the bench.
Tag Power card on a tag collapses to one line showing the offset in force, with a warning icon when tags of its model advertise different power. More details on the same panel is now a card of the same shape.

Platform operators

SYSOP → Calibration Bench measures tags for any organisation, and also gateways: stand them together on one table with one tag in front, and the bench finds how much louder or quieter each one hears than the group. It refuses to save when a ticked gateway reads too quietly to be on the table. Saved gateway offsets take effect only where the organisation's Gateway receive offsets switch is on.

Fixes

A presence decision is no longer lost when its "percent stronger" figure is very large; such rows failed to save about once a day.
The per-gateway count of re-sent frames is now a fixed one-hour window instead of a total that never reset.

2026-09 · Every sensor at a glance

Date: 2026-09-26 Theme: See what each sensor last sent and how well it is heard; a Compliance check that reads as a plain question; charts that draw the hours in order; cleaning jobs judged against a fair deadline. Administrator action: none required. In Toilet Hygiene, a toilet's grace minutes (on its edit page) now sets how long cleaners have to respond to that toilet's auto-raised jobs, so check it where the default does not fit.

Devices

Signal on the sensor list. Device → Sensor has a Signal column showing how well each device is heard: Strong, Good, Fair or Weak, with signal bars and the gateway that heard it. Hover for the exact dBm. LoRaWAN sensors are judged on their own scale, because they normally receive much weaker signals than Bluetooth. The column is on by default; hide it from Columns.
Latest reading on the View panel. Open a device to see the last values it sent (temperature, humidity, CO₂, people, entries and exits, distance, power and more), with when they arrived, through which gateway and at what signal. Other fields are under All fields. Temperatures follow the organisation's unit.
LoRaWAN packets in Monitor now show their values (temperature, humidity, battery) as chips in the packet history, as Bluetooth packets already did.

OAIA

Links in answers are clickable and open in a new tab, so the conversation stays open. Bulleted answers show as lists.
"What is sensor X reading now?" is answered in one step, with the device's latest values and when they arrived. It used to guess at tables that do not exist first.

Environment

Compliance, redesigned. The tab asks one question: is each room staying in its range? Range, period and hours are set in one card, and a sentence underneath says exactly what was checked. The headline figures are rooms that stayed in range, days in range, days out of range and days with no data. Room by room, you see a result, an in-range bar, days too warm and too cold, hours out of range (with the longest stretch) and the lowest and highest reading. Click a room to see its days. MKT is kept for medicines and explained, and can be ignored for comfort ranges.
Hours checked: All day or Operating. Operating checks each room only during its venue's opening days and hours, so a closed day is neither counted nor treated as missing. Use All day for fridges, stores and server rooms, and Operating for comfort in offices. The page and the CSV both say which was used, and the page warns when no opening hours are set.
The trend chart draws the hours in order. A day's chart ran 18:00 → 02:00 → 21:00 → 04:00, drawing the early-morning hours after the evening ones. The readings were right; only their order on the chart was wrong. The Distance report's per-sensor sparkline had the same fault and is fixed too.

Toilet Hygiene

Schedule board job cards. Each job in an hour is its own card showing the date, when it was raised and closed, how long it took (2h 16m, not 136 min), and who was in the zone at the time. One stay that covered two jobs says Same visit as … instead of repeating itself, and separate visits are matched to their own jobs by time. Tooltips explain what each duration is measured between.
Late jobs are marked on the job card (13m late, with the due-by time), on the 24-hour grid and in the hour list.
Fair deadlines. An auto-raised job's due time now allows for the clean itself: response time plus the toilet's expected clean time. Before, a quick response followed by a normal-length clean could never count as on time. The response time follows the toilet's own grace minutes. Jobs raised before today keep the score they were given.
The auto-trigger badge shows the numbers behind it: visits since the last clean against the toilet's threshold.
Open jobs no longer disappear at midnight from My Work and the home tile. Only completed jobs reset each day.
Recent activity reads in plain words: no internal IDs, "Guest rated 4 of 5 stars ★★★★☆", icons that match each item's state, and a note explaining what SLA means. AI insight wording is fixed: "Hour 10:00 alone…" instead of "10:00–10:00", a realistic number of cleaners, and "no baseline yet" instead of "+100 %" for a new sensor.
A visit already closed by its own job is no longer recorded a second time.
The high-traffic cooldown can't be set below 5 minutes, because the scan only runs every 5 minutes.
For support: toilet-hygiene:export-high-traffic-wos, clear-work-order-data and reimport-high-traffic-wos export, clear and reimport one organisation's high-traffic jobs. They are dry runs unless --force is given, and they touch only Toilet Hygiene jobs.

Mobile

Security Watch: the mobile API now lists open incidents as New (nobody has looked yet) or In hand (acknowledged or someone attending). The app can then move an acknowledged incident off the to-look-at list without calling it resolved.

2026-09 · Assistant everywhere

Date: 2026-09-25 Theme: OAIA answers faster, shows its work, learns from ratings and reaches the phone; LoRaWAN is proven on real Milesight hardware. Administrator action: to ask OAIA from Telegram, connect the organisation's bot once from the OAIA page (Telegram → Connect the bot to Omaya, needs manage-oaia), then each administrator links their own Telegram with the one-time code shown there. The mobile app's OAIA endpoint is for administrators with OAIA access; app users are refused. php artisan oaia:eval <organisation> runs the standing set of real questions and compares with the previous run — it spends the organisation's AI allowance like any other question.

OAIA — the assistant, rebuilt around the questions people ask

Live progress. While OAIA works, the chat shows what it is doing — Looking at what data exists, Checking the data, Reading the Omaya documentation — and the answer appears as it is written, instead of a spinner until the end.
Ready-made lookups for gateway status, finding a device (by name, MAC, model, zone, or "not heard for 24 hours"), who is in a zone now and open issues, plus a site snapshot, so a simple count or status question is answered in one step and about two seconds.
Links in answers to the device page or the report that holds the data.
👍 / 👎 on every answer; 👎 asks what was wrong. SYSOP › OAIA Usage shows satisfaction and the answers marked wrong beside the most-asked questions.
Long conversations are summarised automatically, keeping the last few exchanges word for word, so a long chat stays quick.
Ask from anywhere: an OAIA endpoint for the mobile app, and OAIA over Telegram in a private chat with the organisation's bot. A change OAIA proposes from the app or Telegram is never applied there — it links back to Omaya to Apply.
Measured, not guessed: on the standing set of 25 real questions OAIA now answers 24 correctly, median 4.2 seconds and 2 steps. A faster model was measured for the easy questions and rejected: it answered half of them wrong.
A rule OAIA drafts for the whole organisation (a battery or silence rule) now saves; it was refused by the rule editor since the stricter scope check.

Security

AI keys never cross organisations. In background work (OAIA's heartbeat and daily review, Telegram answers) one organisation's work could have run on the previous organisation's AI key, or picked it up as the platform key. Each run now uses its own organisation's key, and a missing key is cleared rather than inherited.

LoRaWAN and devices

Every Milesight LoRaWAN model — 138 of them in 14 series (AM, AT, CT, CTH, DS, EM, FT, GS, TS, UC, VS, WS, WT, WTS) — can be picked on Device → Sensor, grouped by series as in Milesight's own decoder library, each with what it reports.
A LoRaWAN EUI can be entered. The gateway and sensor forms cut a 16-character EUI to a 12-character MAC (a Milesight gateway EUI was saved shortened, so no uplink would match it). More than 12 characters is now kept as a full EUI.
The Milesight UG56 guide is proven on hardware: a real UG56 with an EM-series temperature sensor, set up as the guide says (including its payload codec), delivered its first reading to the Environment report within a minute — the same 25.1 °C and 54 %RH the Milesight app showed.

Toilet Hygiene

A people-counter chart with AI insights and recommendations, redesigned for clarity, with its threshold explained and the 12/24-hour preference honoured.
"Needs cleaner" uses a threshold that means something and says why; visits since the last clean is an exact live counter, estimated for the current hour rather than frozen at zero, with a five-minute grace after a clean.
The high-traffic trigger is skipped when a cleaner is already on site, and its cooldown is per toilet.
Cleaners tracked by tag are assigned and closed by how long they stayed; a live cleaner the entry event missed still gets their stay; an unclaimed job's card shows who is already there; separate visits are no longer merged into one.
Ad-hoc jobs show on the schedule board even with no schedule set up.

Positioning

The room estimate blends in trilateration alongside the every-gateway zone likelihood (a switch per organisation), with its history kept by zone.

2026-09 · Operations release

Date: 2026-09-24 Theme: two new apps, LoRaWAN proven end to end, positioning that knows the room, and alerts that reach a person. Administrator action: Data Centre and Visitor Management are off until switched on in Your Apps (Data Centre needs the enterprise plan). Visitor self check-in is now off by default. LoRaWAN gateways stay online for 30 minutes between uplinks — see below. OAIA's read-only users need SELECT on the new Data Centre views: php artisan oaia:provision --print-grants, run once as the database root user. Omaya's shared AI service now runs on MiniMax (it was Groq): every organisation without its own AI key moves to it automatically; an organisation with its own key is unchanged.

OAIA — faster, better-grounded answers

An audit of a slow, unsure answer (two minutes for "how do I report on my door sensor?") found the assistant searching for the wrong device, not knowing the report already existed, and losing rounds to failed database queries. Fixed:

Reads Omaya's own documentation. For how-do-I, where-is and set-up questions OAIA searches the guides at /docs/ and answers with the link, instead of improvising.
Points to the report that already answers it — Enter / Exit for counters, People Count, Space Utilisation, Environment, Distance and the rest — before offering to build a data table, rule and custom report.
Knows what a distance sensor is set to (occupancy, level or counter), so "my sensor at the door set as a counter" finds the right device.
Fixes a failed query in one go: when the database refuses a query, OAIA is told the views and columns that do exist, rather than guessing again.
The same question now takes about 20 seconds and 3 steps instead of two minutes, and answers it correctly.
Shared AI moved to MiniMax. The previous shared provider rejected longer questions outright ("request too large") and throttled a second question within a minute; organisations on the shared service were getting failures.
SYSOP › OAIA Usage shows the most-asked questions (the same question counted however it was typed), and every recent question with its answer, how long it took and how many steps.
Switching AI off for an organisation now also stops OAIA's scheduled daily review, which had ignored the switch.

Data Centre (new app, `Your Apps → Data Centre`)

A data hall or comms room watched like the rest of the building: PUE and the facility / IT / cooling / other split, from the meters the site actually has — the card says which basis is in use and warns when a figure is estimated.
The plant: UPS (on battery, battery low, load), cooling units and fan walls (stopped, warm supply, failed fans against the wall's size), and the generator that has not started while a UPS is on battery. A unit can be marked standby or in maintenance so it raises nothing it should not.
Alerts that hold their nerve: one per hall, named for the worst rack; escalate past the critical limit or while still climbing, never step down while open, clear only once settled — and a unit that goes quiet keeps its alerts and gets a not reporting one of its own. A critical nobody acknowledges is sent again; one that clears sends an all-clear.
2D and 3D floor plan: racks coloured by inlet temperature, every plant unit at its real size and angle, recoloured live by the alert it carries.
Operating practice: 17 checks against recognised standards (thermal envelope, N+1 on cooling and UPS, fan-wall headroom, the 80 % rack-load rule, monthly generator test, PUE category, sensor coverage, time to acknowledge), on the site's own readings — not measured where the site lacks what a check needs. Ask OAIA for a review picks the two or three to do first.
Trends: daily energy by IT, cooling and other, PUE against baseline and target, cost at the tariff, and the UPS / generator log. Dark theme for the wall screen.
Readings arrive as already-decoded telemetry from an edge gateway (Modbus, OPC UA or SNMP on the gateway side) over the generic MQTT contract. OAIA hears the app and can propose a hall's inlet limits and the site's PUE target or tariff.

Visitor Management (new app, `Your Apps → Visitor`)

Registration: ID-first public registration, a searchable directory of host companies with their logos, terms on or off, a returning visitor's details filled in; direct registration link, no app needed.
Approvals: approve or deny from the admin, by e-mail, WhatsApp, Telegram or any messaging connector; the approval e-mail is the pass — QR code, building, company and time.
Reception: one screen for guard and receptionist — Board, Today, Awaiting, Past — readable with 300 on site; the desk camera scans the visitor's QR or the badge's own QR to check in and out; a visit opens in a slide-over.
Self check-in is off by default and needs the entrance check-in code; scanning never checks anyone in by itself.
Where they went: the badge trail clipped to the stay; a Movement table in Reports.
Gates and lifts opened by HTTP or MQTT, plus the MicroEngine xPortalNet access-control connector; ID verification through the new Identity Provider connector (Singpass, MyDigital ID, any OIDC).
Privacy: a retention period with nightly erasure, a privacy notice as the default terms, PDPA erasure covers the visit's audit records too; no public IC lookup.
A legacy-VMS importer, a demo-data command, and the visitor app's own pre-registration. OAIA and Security Watch both hear visits.

LoRaWAN, gateways and sensors

Milesight UG56 / UG65 supported through their built-in network server, with a box-to-readings guide (Docs → Hardware) — including the small codec wrapper the gateway needs, and how several UGs share one network server (Gateway Fleet), so no separate server is needed.
ChirpStack v4 and v3, over its HTTP integration (recommended) or its MQTT integration, tested end to end against a live ChirpStack with simulated gateways and devices. Fixed on the way: v3 uplinks were dropped (gatewayID), device-status battery was ignored, ChirpStack's built-in Cayenne LPP codec reported nothing, and a device that re-joined within six minutes lost its first uplinks as "duplicates" — now keyed on ChirpStack's own uplink id.
Milesight VS133 / VS135 people counters: line crossings become Enter / Exit, regions become the room's People count.
A LoRaWAN uplink's motion and air quality now reach presence and the Environment report even when they arrive beside temperature and battery (they were decoded and dropped). Milesight's pir words and illumination are understood; their 0–5 light_level is no longer shown as lux.
LoRaWAN gateways (registered by their EUI) are only heard when a sensor sends, so they stay online for 30 minutes after an uplink and go offline, with an alert, after 60 — Bluetooth gateways keep 5 and 15.
MQTT over TLS on port 8883, and the full contract for gateway manufacturers (Docs → Integrate → Connecting a gateway over MQTT). MOKO MKGW gateways read natively.
New decoders and models: Eddystone (MOKO M2), MOKO B2, H8 and S01P, BeaconX in Eddystone mode, Minew MBM01 / MWL01 / MBS02, MOKO M3 mini; MOKO SO2R time-of-flight with its own calibration panel and a Distance report under Sensors.
Ingestion ignores the frames a gateway re-sends unchanged after losing a tag, enforces each gateway's configured RSSI floor, and a listener restart no longer costs ten seconds of packets.
Docs → Hardware is regrouped (Plan / Connect a gateway / Sensors and tags / On site), with step-by-step guides for Dusun, MOKO and Milesight gateways — each ending with the steps in Omaya.
Distance sensors: pick a mode for your scenario (Docs → Hardware): the SO2R's occupancy, level and counter modes taught by scenario — parking bay, bed, desk, loading dock, waste bin, tank, supplies, trolley lane or vehicle gate — with where to mount it, the settings, what appears in Omaya and the rule to add.

Positioning

The room is decided from every gateway, including rooms that have none of their own, instead of the one that heard best.
Mounting height, per-gateway receive offsets, per-venue tag offsets and tag transmit power read from what the tags advertise — so a tag other than the calibration tag ranges correctly.
A Health tab on the floor plan says why a floor positions the way it does; Indoor Positioning and IoT Defaults panels redesigned; Tag Power moves under Sensors.

Rules, alerts and notifications

Rules on any sensor reading: a condition on any decoded field, with edges — becomes X fires once when the reading changes to X, changes on any change; a condition can be required to hold for a while; label-scoped rules actually run; the interval throttles the check, not the firing.
Pick who is told: a recipient picker that says who can actually be reached; Mobile app push and Mini app notification channels; the in-app alert now reaches somebody; Organisation → Notifications holds the default recipients; test where an alert lands before arming it.
Rules can raise work orders and fill a tenant's own data tables, which Custom Reports read.
One branded layout for every e-mail Omaya sends, with the platform mailer as the fallback when a tenant has no SMTP connector; Usage shows which sender was used.
Fewer notifications: coverage alerts stop flapping, identical repeats collapse, battery alerts are one per episode.

Reports

Environment: the weather outside beside the rooms; temperature read as over-cooled / comfortable / warm / hot, humidity with the dew point and mould risk, noise as quiet to very loud, a card per pollutant, light and pressure where measured.
Sensor Status rebuilt (availability per sensor, hour by hour); Battery shows the fleet now with a 30-day forecast; Alerts gathers gateway outages, coverage, toilet and Security Watch incidents, rule triggers and emergencies; Button Press lists the events and which raised an alarm.
People Count, Enter / Exit, Dwell, Flow, Loyalty, Heatmap, Impact, Space Utilisation and Room Usage each say what they count and where the rows come from, measure against hours that have happened, and use the organisation's own calendar day.
Enter / Exit, by sensor: expanding a sensor shows one row per day — entries, exits, active hours and the busiest hour — and a click opens that day's hours; open days stay open across refreshes.
The Reports menu is a two-column flyout in four sections; OAIA can build the report the platform does not have (a table, the rule that fills it and the custom report, in one proposal).

Apps, mobile and admin

Mobile app API for Roll Call, Security Watch, Toilet Hygiene My work, visitors, bookings, the restroom queue, tasks and reports with photos — gated by what the organisation is entitled to; documented in the API reference.
Toilet Hygiene: dispatch goes to every cleaner and the first to walk in claims it; the schedule board with photos and remarks; high-traffic cleaning judged only on traffic since the last real clean; settings redesigned.
Facility: work-order list pages, due times in the building's own clock.
Dashboard: six building-status widgets and Auto Arrange; Ctrl+K finds pages.
Devices show reporting / late / silent / never; Gateways show a 30-day outage history and quiet hours that silence expected outages; Devices → Panic Buttons routes each button.
An activity log of what app users do; light and dark themes on every tenant-app page; OAIA records what people ask it and what it could not answer.
Audit fixes across every app (2026-09-24): Data Centre, Meeting Room, Energy, Security Watch, Roll Call, Compliance, Visitor, TV Dashboard, Toilet Hygiene and Facility.

2026-09 · Intelligence release

Date: 2026-09-13 Theme: the platform explains itself, judges instead of listing, and can be asked. Administrator action: OAIA is off per tenant until switched on from the A.I Model card; Security Watch is off until its master switch is on. Superusers may withhold OAIA per tenant on AI Keys & Tenant Limits.

OAIA — Omaya AI Assistant (`/admin/oaia`)

Heartbeat on the tenant's schedule (2, 4, 6, 8, 12 or 24 h): gateways offline or weak signal, ingestion gone silent, batteries, devices off their reporting pattern, environment off baseline. Detectors detect; the model narrates only what changed (new, escalated, cleared) and every admin is told.
Daily review (every 1, 2, 3 or 7 days): proposes rules, safety and comfort improvements and configuration fixes — conditioned on sense (a comfort alert for a meeting room only while someone is in it). Accept opens the rule pre-filled; dismiss and done are remembered.
Findings can be acknowledged, snoozed, reopened or marked won't-fix; cleared findings are hidden but retrievable (filters: open / muted / cleared 7d / won't fix).
Chat knows the tenant, the local clock and the asker's role. /compact folds a long conversation; /new starts over. A working indicator shows while the model runs.
Actions with approval: the assistant drafts one of eight catalogued actions (venue hours, zone capacity, rule on/off, device zone, finding and suggestion states); a person with the permission applies it, once. Nothing is changed by the model itself.
Guardrails: one tenant per assistant through read-only oaia_* views with the tenant baked in; a read-only database account per tenant (200 rows, 5 s per query); credentials redacted; no shell, no file system, no other tenant. Red-teamed with prompt injection, jailbreak, raw-table and cross-tenant requests.
Off means off: with the tenant switch off, the menu entry is hidden, the page shows an off-state, and every OAIA endpoint refuses. Superusers can withhold OAIA per tenant (oaia_allowed), which locks the tenant's switch.
Console redesign: light and dark, chat on the left, decluttered header, scan-eye emblem.

Security Watch (app-store module, `Your Apps → Security Watch`)

A guard-facing app (app users, /{tenant}/apps/security-watch) with a Live board in two modes — Traditional (every zone and incident) and Adaptive (one status: green / orange / red, then only the zones that earned it, then incidents) — remembered per device.
Detectors (every 5 minutes): panic button, after-hours presence, restricted-zone entry (allowed by entity tag), unknown device lingering, coverage lost, environmental danger (heat level or hour-over-hour rise). Assets never count as people.
Out of habit, judged by AI: a person somewhere they have never been in 30 days, or a room in use at an hour it never is with no booking, is a candidate; the model weighs the context (usual zones, the zone's regulars, who else is present, the booking, the hour) and only "abnormal" becomes an incident. Every verdict is recorded and shown on the settings page. No per-person zone lists to maintain.
Incidents: new → acknowledged → attending → resolved (genuine / false alarm / expected), with an AI assessment and a suggested first step; push to guards; escalation to named people (app users and admins) by email, SMS or WhatsApp, plus a Telegram chat.
Admin settings: master switch (off by default), watched zones with tag pills, armed hours (Mon–Sun toggles), detectors, whitelist windows, recipients, AI judgements log.

OAIA hears every app — and manages what a person approves

One assistant for the whole organisation. Each app you switch on plugs into OAIA through one adapter (app/Services/Oaia/Modules, one class and one registry line per app): its read-only views, its section of the daily review with its own gaps, its heartbeat detectors, its catalogued actions, the records another page can put in front of the chat, and what the model is taught. All of it keyed on whether the app is switched on for the tenant in the app store — an app you have not enabled is never mentioned, so the assistant cannot be led to speculate about it. Every app keeps working on its own without OAIA; every app's settings page gains Ask OAIA where OAIA is on.

Security Watch — incidents and out-of-habit judgements (guard ids and raw evidence excluded). security_7d: incidents by type and zone, how guards closed them, median minutes to acknowledge, what is open, the same alarm closed as false three times. Heartbeat: an incident unacknowledged 30 min (critical after an hour) or open 4 h. Actions: whitelist a person or device, set armed hours. Never acknowledges, resolves or re-judges an incident. Your Apps → Security Watch gains an Incidents — last 7 days card with Ask OAIA per row.
Meeting Room — rooms, bookings, calendar events and sync (booker ids, organiser emails and tokens excluded). meetings_7d per room: bookings, no-shows, walk-ins, booked hours beside the hours the room's own sensor counted it occupied. Heartbeat: a calendar that stopped syncing, bookings stuck open. Actions: release a no-show booking, set automatic release, make a zone bookable. Never books, cancels or moves a meeting.
Facility Management — work orders (with an assigned flag; assignee, vendor, CMMS ids and metadata excluded), assets, service records, schedules. facility_7d: opened by source, closed with median hours and on-time rate, the backlog's overdue, stale and unassigned orders, assets past service. Heartbeat names the five most overdue urgent orders and rolls the rest into one line. Actions: open, re-prioritise, cancel a stale order. Never marks work done.
Toilet Hygiene — toilets, incidents, cleaning schedules (rosters, shifts, absences excluded). toilets_7d per toilet: score and band, target, last clean, orders and on-time rate, SLA breaches, dedicated cleaner and schedule, fixtures out of order. Heartbeat: a toilet in the orange or red with no cleaning order open. Actions: dispatch a clean now (to the nearest available cleaner, as the high-traffic trigger does), set the target score, set the high-traffic threshold. Never marks a clean done or touches a shift.
Energy — readings (raw payload excluded). energy_7d per meter: the week's kWh, open-hours versus closed-hours draw in the venue's own hours, kWh per occupied hour. Heartbeat: power left on after hours — 70 % of the daytime draw an hour after closing. No action of its own: the saving is a rule on the building controls, which the review now proposes with the watts and the weekly kWh.
Connectors (a platform feature, on for every tenant) — health and counts only; the credential column, webhook URLs and secrets, payloads and response bodies are excluded, and any URL inside an error message is replaced with [endpoint]. Heartbeat: an enabled connector whose latest error is newer than its latest success (critical once a working one has gone a day without success), saying what stopped — alerts, automations, gateway data; webhook deliveries that died today. Actions: re-send a webhook's dead deliveries, pause or resume a webhook. Never changes an endpoint, a key or a mapping.
Visitor Management — visits, hosts and companies; on-site now, awaiting approval, the week by status and channel. Joins the same review, heartbeat and chat.
Compliance — standards, rules, profiles, assessments (payload, hash and author excluded), findings, calibrations (certificate references and document paths excluded). compliance_7d: each profile's latest verdict with the findings behind it, calibrations overdue and due. Heartbeat: a profile without a current assessment, lapsed calibrations. Actions: run an assessment (7, 30 or 90 days), switch a profile on or off. Never edits a standard, a rule or a finding.
Roll Call — events, attendances, badge reliability (actor ids and the IP-bearing audit trail excluded). rollcall_90d: drill cadence and days since the last drill, the last drill's counts and who was missing with the zone their badge was last heard in, events left running, badges too unreliable to muster. Heartbeat: drill overdue, a roll call still running after six hours. No action — a roll call is a live safety operation run by a person on the spot.
Approve one Autopilot action at a time. The Autopilot card is now a ledger like the findings: one row per pending action (not one per heartbeat), opening to the reasoning and the before-snapshot, with Approve, Undo, Open proposal and Ask about this. In Shadow mode, Approve is how you try Autopilot one action at a time before switching to Act — it goes through the same path, is marked done by you, and undoing your own approval never pauses Autopilot. Held-by-limit rows no longer appear in Shadow, where nothing is spent.
The OAIA page puts the chat on the whole left column, the numbers strip folds away (counts stay in its header), and Recent heartbeats sits under Autopilot.
Autopilot stays the only automation. No app action can ever be an Autopilot kind (a test pins it): whitelist, release, dispatch, cancel, retry and the rest wait for a person to click Apply whatever the mode, and with Autopilot off OAIA changes nothing on its own. Switching Autopilot to Act, and making it available to a tenant, now confirm in words that it is beta and can get one wrong.
Fix on the way: Roll Call stamped every event, attendance and audit row with tenant_id = 'default' on the admin path and read presence from the wrong tenant's cache when recomputing attendance; it now uses the active tenant, and existing rows were restamped inside each tenant's own database.
Audit pass: an app the tenant has not enabled is now invisible to the model end to end — its views are dropped from DescribeSchemaTool and refused by the SQL guard, not just left out of the prompt; the app-store gate is remembered per tenant for a minute (48 landlord queries per chat prompt → 2); and the OAIA page shows every time in the tenant's timezone rather than the browser's, like every other admin page.
The About page and Capabilities page describe the assistant across every app and Autopilot (English and Malay).
Administrator action: the read-only database users need SELECT on the new views — php artisan oaia:provision --print-grants prints the script; run it once as the database root user. Views for an app enabled later are added at the next heartbeat.

OAIA Autopilot (beta, off by default)

The reversible subset of OAIA's proposals applied without waiting for a person, behind three switches: the platform grants it per tenant (AI Keys & Tenant Limits), the tenant picks Off / Shadow / Act on the A.I Model card, and turns on tiers — Draft (create proposed rules switched off), Configure (zone capacity, venue hours, assign an unassigned device), Operate (cooling on when a room is hot or humid with people in it, an equipment room passes 30 °C, or a booked room is hot 15 minutes before the meeting; fans on for stale air; lights on for people in a dark room; everything off when the radar says a room has been empty ten minutes or the venue has been closed half an hour with nobody counted). Configure also learns a capacity for a zone that has none from 14 days of counter data.
Guardrails in code: 3 actions per heartbeat, 20 per day, an hour per target, never flipping a device back within 30 minutes, setpoints inside 20–28 °C, no locks/sirens/scenes, no enabling rules, no deletes. Every action is listed on the OAIA page with Undo; an undo pauses that kind for a day, three in a week drop Act to Shadow. Administrators are told of each action.

Home Assistant / building control

Rules could already call any Home Assistant service on any entity (light, switch, cover, climate, fan, scene, …) through the Smart Home Bridge connector, with extra service data such as brightness or temperature.
New metric radar_vacant — "Area Empty For (radar)", the mirror of radar_dwell — so a rule can switch lights, air-conditioning and blinds off ten minutes after the last person leaves, not the instant the radar reads zero. Available in the rule builder and to the AI rule author.
OAIA proposes rules that act. The daily review now sees the tenant's mapped building controls and drafts automations on them — lights and air-con off after the room has been empty ten minutes, air-con on when it is hot and someone is in, blinds closed in the afternoon, everything off at closing. Only entities mapped on the Mappings page can be drafted (the same check the rule editor applies), and a draft is saved inactive for a person to switch on. Arbitrary endpoints (HTTP, MQTT) and record writes stay form-only.
A contract test now guards generated rules: every word the assistant is taught exists, and every draft the validator accepts is accepted by the rule save endpoint.

AI Reporting, keys and limits

Ask for a report in plain language (/admin/report/ai); table and chart come from the same data as every other report. Credential chain: tenant's own key (BYOK) → platform key → environment.
Token caps: tenants on Omaya's shared AI service have a monthly cap (per-tenant value, else the platform default, else 500,000 tokens); a tenant's own key is never capped. Superusers set caps and the platform default on AI Keys & Tenant Limits.
Omaya Platform Usage (/admin/usage): what the tenant consumed this month — AI tokens per provider (shared vs own key, against the cap), SMS, WhatsApp, email (alerts, OTP, reports, portal, module mail), webhooks, in-app alerts — with recent failures. Backed by a message log written inside every sender.

Sensing, positioning and reports

Radar people counting (MSR01-A): positions persisted and used; coverage calibration panel and gate; Right Now fuses radar and sensor counts over tags; Room Usage Map report; radar ground truth pairs a lone tag with the radar dot every minute.
Space Utilisation report (all zones against capacity); anomaly detection revived (episode dedup, tags get a 3-day threshold).
Cross-floor routing (floor links: lifts and stairs as real links between plans) for the visitor app's indoor navigation; 3D floor plan widget on TV dashboards sharing the admin 3D geometry.
MOKO B2 button decoder; per-tenant detection speed presets.

Integrations and access

Calendar integration (Google and Microsoft 365) per tenant, feeding Meeting Room bookings.
Admin SSO (OIDC per tenant): per-tenant link, email-domain routing, SSO-only option, group→role mapping; MFA skipped after SSO.
Public API docs at /documentations/api.

What's new page

/admin/about/whats-new renders these notes; linked from the About page and the capability reference.

Housekeeping

Anomalies page shows who acknowledged, by name (or "by system", with the note).
Auth pages use the white wordmark. /admin/config#ai opens the A.I Model card directly.
/admin/about and /admin/about/capabilities rewritten with an Intelligence section, nine worked scenarios, seven drawn illustrations; the full page in Malay.

What's new ​

2026-10 · Trip speed ​

2026-10 · Fall detection: wear it right, and a report that stays readable ​

2026-10 · Connector pages, platform administration and translations ​

2026-10 · Clearer setup pages: Tag Power, Auto Discovery and three app settings ​

2026-10 · Ask OAIA follows role access; TV app widgets follow the board ​

2026-10 · Outside: privacy, TV boards and a scheduled report ​

2026-10 · Outside, across the platform ​

2026-10 · Watch gateways and sensors from Zabbix ​

2026-10 · Monitor → Outside: who is outside right now ​

2026-10 · Field staff: working outside counts ​

2026-10 · Reports that know a tag can be outside ​

2026-10 · Sensor Trail: indoors, outside and the distance travelled ​

2026-10 · Activity Log: full timeline or what needs a look ​

2026-10 · Scout: outdoor coverage per tag holder ​

2026-10 · Live Monitor: five tabs, each a summary ​

2026-10 · Central Alerts: every module, repeats folded ​

2026-10 · A bigger entity view on Entity Tracker ​

2026-10 · Gateway timing, and scheduled tasks that say why they failed ​

2026-10 · Panic buttons are rules ​

2026-10 · Omaya Scout finds tags on iPhones ​

2026-10 · Button presses on Minew tags ​

2026-10 · One look for everyone; Rules, Connectors, the Map Editor and System pages redone ​

2026-10 · Pages that open each other, and calmer Sensor, Gateway, Location and Venue pages ​

2026-10 · Minew tags and beacons can be configured from Omaya ​

2026-10 · A shorter Device menu, with its tools under Utilities ​

2026-10 · The new Omaya look is now the default ​

2026-10 · Monitor: Right Now by floor, a clearer Entity Tracker, and Device Anomalies in the same layout ​

2026-10 · Outside the building: Omaya Scout, GPS Replay and places with names ​

2026-10 · Reports: staff and visitors in Footfall, matching incident counts, and clearer pages ​

2026-10 · Dashboard: Gateways, Devices and Live Sensor Status add up again ​

2026-10 · Visitor app settings: tabs, cards that save in place, visit and ID types ​

2026-10 · Saved views on the device list: no error on opening, and in your language ​

2026-10 · Configure devices from Omaya: a settings library, a Configure page and device passwords ​

2026-10 · Space figures that hold up: workplace seats, rented area, and a venue form that saves its cost ​

2026-10 · What your space costs, office days by team, Slack and Teams, and dashboards in Power BI ​

2026-10 · Security, honest delivery status, LoRaWAN energy, and a working connector hub ​

2026-10 · Energy: a meter whose kWh counter is stuck is counted from its power ​

2026-10 · Docs: step-by-step setup guides, and the fixes they turned up ​

2026-10 · Docs: Recipes, answers to business questions end to end ​

2026-10 · Toilet Hygiene: a cleaner can close a job early, where the toilet allows it ​

2026-10 · Warehouse: see what is stuck, do the next task, feel every scan ​

Today (was Dashboard) ​

Floor ​

Orders ​

Alerts ​

Motion ​

2026-10 · Data Centre: the problem first, every rack at a glance, a wall screen in 2D or 3D ​

Now (was Overview) ​

Floor ​

Wall screen (new) ​

Power (was UPS) ​

Alerts ​

Energy ​

2026-10 · TV Dashboard: connect screens with a code, build boards as the TV shows them ​

Screens ​

Boards ​

On the TV ​

2026-10 · Warehouse, Visitor, Toilet Hygiene and Data Centre move like the other apps ​

2026-10 · About pages: book a walkthrough, see the real product, and a capability sheet to forward ​

Book a walkthrough ​

New on the About page ​

Capability reference ​

Corrected ​

2026-10 · Indoor positioning: see what your building needs, and try a setting before you switch it on ​

What this site needs ​

Try it before switching it on ​

Floors tab ​

2026-10 · Roll Call: who is still to find, floor by floor ​

The board ​

"Are you safe?" ​

Starting and readiness ​

Screens ​

After the roll call ​

2026-10 · Meeting rooms at a glance, and room calendars both ways ​

Rooms now ​

Booking ​

Door display ​

Google Workspace and Microsoft 365 ​

Insights ​

What's new

2026-10 · Trip speed

2026-10 · Fall detection: wear it right, and a report that stays readable

2026-10 · Connector pages, platform administration and translations

2026-10 · Clearer setup pages: Tag Power, Auto Discovery and three app settings

2026-10 · Ask OAIA follows role access; TV app widgets follow the board

2026-10 · Outside: privacy, TV boards and a scheduled report

2026-10 · Outside, across the platform

2026-10 · Watch gateways and sensors from Zabbix

2026-10 · Monitor → Outside: who is outside right now

2026-10 · Field staff: working outside counts

2026-10 · Reports that know a tag can be outside

2026-10 · Sensor Trail: indoors, outside and the distance travelled

2026-10 · Activity Log: full timeline or what needs a look

2026-10 · Scout: outdoor coverage per tag holder

2026-10 · Live Monitor: five tabs, each a summary

2026-10 · Central Alerts: every module, repeats folded

2026-10 · A bigger entity view on Entity Tracker

2026-10 · Gateway timing, and scheduled tasks that say why they failed

2026-10 · Panic buttons are rules

2026-10 · Omaya Scout finds tags on iPhones

2026-10 · Button presses on Minew tags

2026-10 · One look for everyone; Rules, Connectors, the Map Editor and System pages redone

2026-10 · Pages that open each other, and calmer Sensor, Gateway, Location and Venue pages

2026-10 · Minew tags and beacons can be configured from Omaya

2026-10 · A shorter Device menu, with its tools under Utilities

2026-10 · The new Omaya look is now the default

2026-10 · Monitor: Right Now by floor, a clearer Entity Tracker, and Device Anomalies in the same layout

2026-10 · Outside the building: Omaya Scout, GPS Replay and places with names

2026-10 · Reports: staff and visitors in Footfall, matching incident counts, and clearer pages

2026-10 · Dashboard: Gateways, Devices and Live Sensor Status add up again

2026-10 · Visitor app settings: tabs, cards that save in place, visit and ID types

2026-10 · Saved views on the device list: no error on opening, and in your language

2026-10 · Configure devices from Omaya: a settings library, a Configure page and device passwords

2026-10 · Space figures that hold up: workplace seats, rented area, and a venue form that saves its cost

2026-10 · What your space costs, office days by team, Slack and Teams, and dashboards in Power BI

2026-10 · Security, honest delivery status, LoRaWAN energy, and a working connector hub

2026-10 · Energy: a meter whose kWh counter is stuck is counted from its power

2026-10 · Docs: step-by-step setup guides, and the fixes they turned up

2026-10 · Docs: Recipes, answers to business questions end to end

2026-10 · Toilet Hygiene: a cleaner can close a job early, where the toilet allows it

2026-10 · Warehouse: see what is stuck, do the next task, feel every scan

Today (was Dashboard)

Floor

Orders

Alerts

Motion

2026-10 · Data Centre: the problem first, every rack at a glance, a wall screen in 2D or 3D

Now (was Overview)

Floor

Wall screen (new)

Power (was UPS)

Alerts

Energy

2026-10 · TV Dashboard: connect screens with a code, build boards as the TV shows them

Screens

Boards

On the TV

2026-10 · Warehouse, Visitor, Toilet Hygiene and Data Centre move like the other apps

2026-10 · About pages: book a walkthrough, see the real product, and a capability sheet to forward

Book a walkthrough

New on the About page

Capability reference

Corrected

2026-10 · Indoor positioning: see what your building needs, and try a setting before you switch it on

What this site needs

Try it before switching it on

Floors tab

2026-10 · Roll Call: who is still to find, floor by floor

The board

"Are you safe?"

Starting and readiness

Screens

After the roll call

2026-10 · Meeting rooms at a glance, and room calendars both ways

Rooms now

Booking

Door display

Google Workspace and Microsoft 365

Insights