Damp & water damage
Staining, discoloration and spread rate tracked between inspections.
Our Technology
Computer vision that respects how buildings are really inspected.
No new hardware, no sensors in every room. Habyta works from the photography your site teams already capture, and tells you when it is not sure.
The pipeline
Capture, detect, decide.
01 Capture on any phone during a normal inspection.
→02 Classify and segment condition issues per room.
→03 Compare against the room's history to detect change.
→04 Raise work orders and update room status.
Low-confidence detections never raise an automatic work order. They route to a human inspector, and their decision feeds back into the next training cycle.
What the models read
Condition signals with consistent visual signatures.
Damage & wear
Cracked panels, broken fittings, worn flooring and door damage.
Missing inventory
Furniture, mattresses, alarms and appliances absent from where they should be.
Cleanliness failure
Rooms marked ready that would not pass a supervisor's check.
Safety-relevant items
Blocked egress, missing alarm units, and tampered safety equipment.
Occupancy signals
Visual indicators that a bed is in use, held, or genuinely empty.
Data & privacy
Tenant imagery is handled carefully, by design.
Images are processed in AWS's Africa (Cape Town) region, personal information is removed before any training use, and your data is never pooled with another operator's without a written agreement. Full deletion on request.
Honest limits
What Habyta does not do.
We do not detect anything hidden behind walls, we do not replace statutory gas, electrical or fire safety inspections, and we do not claim accuracy figures we cannot evidence from live deployments. Current condition-detection agreement with partner inspectors sits at 91% across pilot portfolios, measured on flagged rooms.