Describe it. Save it.
Write the situation in your own words. Auto Enhancement prepares visual instructions, examples and a screening question, with independent reviews. Inspect the setup or use Operator override for exact wording.
CAMERA-AGNOSTIC VISUAL AI
Any camera.
Your situation.
In your words.
Tell MATA what matters. It watches for the situations you describe and brings you alerts with evidence, using the cameras you already have.
Any industry · Plain-language rules · Your choice of camera
Ordinexis Sdn Bhd Official OpenAI Partner
MATA / SITUATION MONITOR
Illustrative example
YOUR DESCRIPTION
“Tell me when something blocks the walkway.”
Walkway obstruction
MATCH
Camera 01 · Loading area
Evidence ready for review
01 camera
01 situation
Human review
YOUR CAMERAS. MORE CONTEXT.
From watching footage
to understanding situations.
MATA connects visual AI to your camera feeds. A situation can be a visible condition, something that lasts, or a sequence of steps. You describe it. MATA prepares the rule, checks sampled footage and keeps the evidence for your team to review.
Any industry
The description defines the use case
2–8 feeds
In an optional multi-camera view
Image + event
Still checks and short event windows
Live + recorded
Monitor now or review later
HOW IT WORKS
Connect a camera.
Describe a situation.
Let MATA watch.
No industry-specific hardware. No model to train for each new rule. Start with what you want to see.
Find your connection →-
01
Bring your camera
Connect a supported live feed, browser or USB camera. You can also upload images and video for offline review.
-
02
Say what matters
Describe the situation. MATA builds and reviews visual instructions. Choose a region, timing or evidence type when you need more control.
-
03
Review what happened
Receive policy-qualified alerts with captured evidence. Review the incident, archive it, or share it with your operational workflow.
DEFINED BY YOU
One platform.
Whatever you need to watch.
Start with a visible situation. These are example descriptions you can adapt to your own camera and environment.
01 / KEEP ACCESS CLEAR
“Tell me when something blocks the walkway.”
Define the walkway as a region and describe what should stay clear. Review the captured evidence before deciding what action to take.
THE RULE YOU DEFINE
- Watch for
- An object obstructing the marked walkway
- Where
- A region in your camera view
- Then
- Alert your team with evidence
THE PLATFORM TODAY
From a simple rule
to a complete review.
Set up in plain language. Fine-tune when needed. Keep the camera, context and evidence connected.
A moment. Or a sequence.
Choose a still-image rule, a short Event check, or a reviewed sequence of steps. Event checks collect 1–10 seconds of footage and use chronological sampled images to evaluate what happened.
Several cameras. One workspace.
MultiInput brings 2–8 supported feeds into a tiled view. Bind rules to camera regions, keep evidence tied to its original camera, and review simultaneous matches together.
Alerts at the right time.
Choose immediate, repeated, persistent or continuous conditions. Set cooldowns and timing per rule. Review live alerts, archive incidents and send notifications through Telegram or webhooks.
Evidence you can revisit.
Review captured evidence, camera regions and timestamps alongside the rule that fired. Generate observation digests, or upload images and videos for offline analysis and reports.
Keep an eye on running costs.
See estimated inference cost for each run, usage breakdowns and projections. Choose screening and confirmation routes, sampling intervals and monthly budget controls to suit your setup.
Local examples. More context.
Experimental local CPU image matching compares still images with descriptions and reviewed positive or negative examples. Live matching stays local; AI-assisted example preparation is optional. Events, sequences and MultiInput are not supported in this mode.
A workspace for the operator.
Configure cameras, search rules and review event sequences separately from live monitoring. Saved settings survive restarts, and failed rule saves retain what you typed.
BRING THE CAMERAS YOU HAVE
Camera agnostic.
Situation specific.
MATA works across camera brands and industries. What matters is access to a supported feed, then a clear description of what you want to watch.
Compatibility depends on the camera’s available stream or integration. Closed systems may need additional setup.
Check my camera setup ↗
01
Browser & USB cameras
Use a connected device or a supported browser camera.
02
RTSP camera feeds
Connect accessible IP camera and recorder streams.
03
HTTP, MJPEG & push frames
Bring snapshots, streaming sources or integrated frame uploads.
04
Uploaded images & video
Review recorded material through supported offline workflows.
RECENT PRODUCT UPDATES
More ways to watch.
More control over the details.
Product highlights from August–November 2026. Updated 2 November 2026.
Simpler rule authoring
Auto Enhancement prepares the visual setup on Save, with independent checks and a repair attempt. Original descriptions stay intact; Operator override protects manual wording.
Optional Workforce pilot
Local employee enrollment, sightings and observed presence in a separate workspace, with named roles and encrypted biometric templates. An optional pilot, not certified attendance.
Understand the pilot →Flexible detection & spending
Expanded screening providers, local reference matching, four-image batching and run cost accounting. Select compatible routes and tune sampling against your coverage needs.
Multi-camera events & review
Camera-bound mosaics, separate image and Event rules, configurable operating profiles and a dedicated configuration workspace. Full camera views remain intact.
PRIVACY, ACCESS & CONTROL
Your environment.
Your operating choices.
Choose where MATA runs, which models process images, and what your team monitors. Make those choices explicit before connecting production cameras.
Read the AI Use Policy ↗Local hosting, configurable models
Run on your server or LAN with a supported vision backend. Cloud routes send sampled images to their providers; local-only processing depends on your selected features and backend.
Regions, masks and protected access
Scope rules to camera regions and apply privacy masks. The monitoring console uses login protection; the optional Workforce workspace has separate accounts and roles.
Evidence before assumptions
Missing, skipped or obscured observations are not proof that nothing happened. Sampling and budget limits affect coverage. Validate rules on representative footage and keep people in the review process.
OPTIONAL PILOT
Workforce recognition,
Workforce recognition,
with a separate set of controls.
Enroll authorized employees, review local sightings and export observed presence. Recognition runs in a dedicated local CPU worker; biometric templates are encrypted and access is role-based.
Observed presence is not worked hours or certified attendance. Commercial licensing and liveness validation remain open release gates. This pilot is separate from core situation monitoring.
Does MATA work with any camera?
MATA is camera-brand agnostic. Connect a camera through a supported browser, USB, RTSP, HTTP/MJPEG or push-frame source. A closed camera system may need an accessible stream or integration. We check your actual feed during setup.
Do I need to train a model for every situation?
No per-rule model training is required. Describe the visible situation and save the rule. MATA prepares and reviews the visual setup. Test the rule with representative footage, lighting and camera angles before relying on it.
Is it only for shops or restaurants?
No. The situation defines the use case: a blocked walkway, an unattended counter, an object in a region, or a sequence of visible steps. The same platform can be configured across industries.
Does it analyse every frame?
No. Live checks default to a five-second interval, with per-rule sampling controls. Source speed, model latency, budgets and queue capacity affect coverage. Brief events can fall between samples.
Can MATA run locally?
The application can run on your own server or LAN, with a supported OpenAI-compatible vision backend. Cloud detection routes send sampled images to their selected providers. Local-only operation depends on the backend and enabled features; the optional Workforce recognition worker runs locally.
Can I review recorded footage?
Yes. Upload images or videos for supported offline workflows and receive a report based on policy-qualified decisions. Offline analysis and some reference or sequence features require compatible vision modes; Strict Filter blocks separate unscreened image workflows.
Does MATA identify employees?
Core situation monitoring does not require employee identity. The optional Workforce pilot adds local enrollment, sightings and observed presence, with separate named accounts and encrypted biometric templates. It is not certified attendance or a worked-hours system, and production licensing and liveness validation remain open.
How accurate is detection, and what does it cost?
Both depend on your footage, rules, selected models and sampling. Scores are not calibrated accuracy percentages. Test representative cases and review misses before operational use. The console shows inference estimates; hosting, subscriptions and provider invoices can differ.
START WITH YOUR SITUATION
What would you like
your camera to notice?
Tell us about your camera, environment and the situation you want to detect. We’ll show you how MATA fits.