Verify required equipment
Confirm tools, PPE, assets or components are present before work begins.
Tiliter Vision Agent
Visual Reference Agent
Verify tools, equipment, inventory and operational requirements from images automatically. Detect missing items, validate setup requirements and automate visual checks across inspections, logistics, healthcare and industrial operations.
Visual Reference Agent
Stop verifying setups by hand.
Tiliter compares every photo against your expected setup – flagging missing items, confirming what's present, and catching errors before they reach the customer.
Submitted photo
Property turnover · Lounge
Verifying
Comparing against expected setup
Tiliter is matching every required item in the room against your expected configuration.
Required itemschecking…
- Layout matches reference
- All inventory items present
- Cleanliness standard met
- No damage detected
- Ready for guest
What this means for your operation
Catch missing items before they become write-offs.
Most missing items across turnovers, inspections and field operations are discovered too late to recover or escalate effectively.
10s
per verification
vs. several minutes manually
100%
required items checked
before sign-off or completion
Early detection
while there's still time to act
before disputes, losses or reimbursement windows close
Reduced losses
across the operation
in inventory, equipment and operational assets
Built for operational verification
Use the Visual Reference Agent to validate required items, equipment, setup conditions and operational readiness from images automatically.
Validate expected room conditions
Compare rooms, workspaces and environments against expected setup or cleanliness standards.
Detect missing or incorrect items
Identify missing products, incomplete kits or incorrect configurations automatically.
Automate visual inspections
Reduce manual checking across logistics, healthcare, facilities and industrial workflows.
Built for real operational verification
The Visual Reference Agent automatically checks images against expected items, setup conditions or operational requirements and returns structured verification results.
Capture operational evidence
Teams upload images from mobile devices, inspections, warehouse workflows, healthcare environments or existing operational systems.
AI validates requirements
The Visual Reference Agent checks whether expected items, equipment, configurations or setup conditions are present and correctly matched.
Exceptions are flagged instantly
Missing items, incorrect setups or failed checks are identified automatically and returned as structured operational results.
Use cases powered by this Agent
Explore real-world examples using the Visual Reference Agent Agent to automate checks, improve safety, and reduce errors.
Hospitality Room Verification
Verify hospitality room readiness using operational workflows before guest check-in.
Read use case →Footwear Quality Assurance
Detect defects and verify returns across manufacturing, fulfilment and customer returns.
Read use case →Bag Usage Analytics
Measure plastic, reusable and paper bag usage automatically from checkout footage.
Read use case →Part Verification for Industrial Operations
Verify incoming parts against drawings or expected items to catch errors early, reduce delays and keep projects on track.
Read use case →Surgical Tray Verification
Verify tray contents accurately to improve safety, reduce delays and support hospital workflows.
Read use case →Shoe Verification for Retail
Detect fraudulent returns and verify shoes at checkout using image recognition.
Read use case →Visual Reference Agent questions, answered
What is it actually doing when it checks a photo?
Comparing what is in the image against what should be there. You define the expected set, whether that is a list of items, a reference photo or a spec, and each submission comes back as a pass or a fail with the exceptions called out. It is the difference between asking whether something is present and asking whether everything that should be present is.
Can it match a part against a reference?
Yes, and this is where it earns its keep on industrial work. Two part numbers that differ by a few millimetres, a fitting that looks right until you hold it against the correct one, a kit assembled from the wrong variant. The industrial part verification use case shows it working on site.
How is this different from product recognition?
Product recognition answers which item this is. This answers whether the right items are all here, against a reference. A checkout needs the first. A kit, a tray, a delivery or a shelf needs the second, because the interesting failure is the thing that is missing rather than the thing in front of the camera.
And how is it different from counting?
Counting returns a number. This returns a verdict against a specification. Six items when six were expected passes; six items where one of them is the wrong variant does not, and a count would never have noticed.
Can it be used for inventory and stock checks?
Yes. Checking a bay, a shelf or a store room against what the system says should be there is the same operation as checking a tray against its list. What you get back is the discrepancy, which is the part somebody would otherwise have to work out by comparing a photo with a spreadsheet.
What do we compare against, in practice?
Whatever you already have. A written list of required items, a reference image of a correct setup, or a specification. Teams usually start from the checklist somebody is already carrying on a clipboard, because that is the thing being replaced.
Where is it used today?
Surgical tray verification, proof of delivery, construction site standards, room readiness and goods coming back as returns. The common thread is a required state that somebody currently confirms by eye.
What happens when something is missing or wrong?
The exception is flagged rather than buried in a pass. The submission fails with the specific item called out, the photo attached as evidence, and the next step raised in the workflow, so the person who has to act on it is told rather than left to notice.
Do we have to define the expected set for every check?
Once per type of check, not once per submission. A tray configuration, a delivery manifest or a room standard is defined once and then applied every time that check runs, which is what makes the hundredth verification as consistent as the first.
Ready to see the Visual Reference Agent in action?
Build visual verification workflows and deploy them through mobile, web, cameras or your existing systems. Deployed in thousands of locations worldwide and ready to scale with your operations.