Book a consultationBook a call

Turn what happens in the physical world into data and action.

AVA, Agentic Visual Applications, gives JotVision visual understanding of physical environments. Describe what matters in natural language, and AVA can observe the scene, identify relevant objects and events, create structured data, and trigger the workflow that should happen next.

See→Understand→Record→Act

See AVA in action.

A short demonstration of AVA observing a scene and turning it into structured context.

Watch on YouTube →

Computer vision

Detects what is visible

Finds objects and motion in an image.

AVA

Understands what matters

Interprets the scene in the context of your stated intent.

JotVision

Turns it into action

Combines it with enterprise context and carries the workflow to completion.

Some of your most important data doesn't exist in a system yet.

It exists in rooms, equipment, procedures, interactions, objects, and events.

1

Objects

What is present, missing, moved, or used?

2

People

Who is present, interacting, or performing an activity?

3

Events

What happened, in what order, and at what time?

4

State

What is the current condition of the environment or process?

5

Relationships

How are people, objects, and events connected?

AVA turns these observations into structured context that JotVision can record, analyze, or act on.

Observe, understand, decide, act.

Camera movement and video capture are things AVA can do. They are not what defines it.

1
Observe

Watch the environment

Continuously interpret what is happening in a physical environment.

2
Understand

Identify what is there

Identify relevant objects, people, states, events, and relationships.

3
Decide

Work out what matters

Use the stated intent and operational context to determine what matters.

4
Act

Carry it forward

Create data, capture evidence, notify someone, trigger a workflow, control a device, or request human review.

Describe what matters.

Instead of configuring every workflow around a fixed taxonomy of predefined events, describe the observation or outcome you care about in natural language.

“Tell me when this tray is complete.”
“Track when the patient enters the room and when positioning is complete.”
“Inspect this instrument for visible damage.”
“Record when this object is removed and when it returns.”
“Keep the camera on whoever is demonstrating the procedure.”
“Alert me if this step is skipped.”

Intent can change without rebuilding the entire visual application.

The physical world, connected to action.

AVA is not an isolated computer-vision product. It is a source of real-world context for a system of action.

Physical world

Where it happens

PeopleObjectsEquipmentRoomsProcessesEvents
AVA

Sees and structures

  • Observe
  • Understand
  • Structure context
JotVision system of action

Decides and executes

  • Apply rules
  • Combine with enterprise data
  • Determine next action
  • Escalate if needed
  • Execute workflow
Outcome

What gets done

RecordNotifyUpdate systemTrigger processControl deviceCreate evidenceRequest approval

Visual context is more useful when combined with business context.

AVA does not operate only on pixels. JotVision can combine what AVA observes with enterprise context.

AVA sees

Instrument placed on table

Your ERP and operational systems know
  • Which procedure
  • Which tray
  • Which instrument should be present
  • Which staff member owns the workflow
JotVision can determine

Whether something is missing, unexpected, or requires action

System of action
RecordEscalateUpdateNotifyRequest review

An illustrative example. The specifics adapt to each validated use case.

AVA in practice

Example applications across use cases. Each turns what the camera sees into data, evidence, or action.

Instrument tray with AVA overlays labeling each instrument by manufacturer and part number
Observations to data

Surgical instrument tray identification

Point AVA at a tray and it identifies each instrument by manufacturer and part number, matched to catalog reference images. What was a manual count and check becomes a record.

  • Each instrument tagged on sight
  • Matched to the catalog entry
  • Powered by LayerJot
Two robotic arms with cameras inspecting a surgical instrument, with close-up views and a Pass result
Inspection

Robotic visual inspection

Cameras on robotic arms examine each instrument from every angle and return a result, with the evidence attached for review.

  • Surface wear and tear
  • Mechanical damage and alignment
  • Rust, corrosion, pitting, and tint
OR.FLOW tablet app showing live analysis of an operating room with detected events
OR.FLOW

Operating room flow analysis

AVA watches the room and logs what happens: staff and patient detected, room entry, patient transfer, and positioning, with identities blurred. You get a notification when the report is ready.

  • Events detected and timestamped
  • Identities blurred by default
  • Report delivered when ready
AVA console tracking a presenter in a lecture hall, with a watch list of presenter, whiteboard and projector screen
Lecture capture

Automatic lecture recording

A reference camera sees the whole room while the PTZ camera follows what matters. Set a watch list and AVA locks on and frames it, with no camera operator.

  • Watch list: presenter, whiteboard, projector screen
  • Locks on and tracks the presenter live
  • A showcall sequence cuts between subjects

Use cases, organized by the action they drive.

Different environments, one architecture.

Observe & Measure

Understand what is happening in a physical process.

Operating-room flowLine sequences

Inspect & Verify

Check objects, conditions, sequence, or completeness.

Instrument tray IDRobotic inspection

Track & Record

Create structured records of physical events.

Room eventsLecturesProcedures

Guide & Control

Use visual understanding to influence cameras, devices, or workflows.

PTZ trackingRobotic arms

Trigger & Escalate

Start a business process when a real-world condition occurs.

Skipped-step alertReview request

What AVA produces

From what the camera sees to data you can act on.

Structured data

Observations become usable data

Where permitted and appropriate:

  • Objects
  • Identities
  • State
  • Timestamps
  • Counts
  • Sequences
  • Events
  • Measurements
Evidence

Visual context behind the result

So a result can be reviewed, not just trusted:

  • Images
  • Video clips
  • Event history
  • Timestamps
  • Relevant scene context
Action

Observation drives the next step

AVA does not stop at detection:

  • Update a record
  • Trigger a workflow
  • Notify a person
  • Escalate an exception
  • Control a camera or compatible device
  • Request approval
  • Create a task

Privacy built in

Faces and identities can be blurred in the analysis, so the data is useful without exposing people.

Local or cloud processing

Process footage and prompts on your premises or in the cloud, whichever fits your requirements.

An API is on the way

We are working to open an API so you can build on AVA’s scene understanding. Talk to us about early access.

AVA can also control how the scene is captured.

Video and PTZ control are examples of actions AVA can take once it understands the scene.

Follow a person or object

Lock on and re-acquire through movement and occlusion.

Maintain framing

Keep the subject framed as the scene changes, with no operator in the room.

Switch viewpoints

Move between subjects, such as presenter, whiteboard, and screen.

Control compatible PTZ cameras

Direct PTZ cameras from a stated intent, in plain language.

Capture event-specific video

Record the moments that matter, wide context and close-up detail in sync.

Create review-ready records

Paired assets ready for training, inspection, QA, and incident review.

Also in Vision & Robotics

From seeing to doing.

Cameras on robotic arms let AVA examine, verify, and respond to physical objects, with a person reviewing what matters.

Explore robotics →
Two robotic arms with cameras inspecting a surgical instrument

Partners in the physical world.

AVA works with proven hardware and specialist technology partners, so the cameras and catalogs are theirs and the intelligence is ours.

Camera partner
PTZOptics logoPTZOptics

Professional PTZ cameras, controllers, and software for broadcast, education, healthcare, and manufacturing. AVA directs PTZ cameras to frame what matters.

Visit PTZOptics →
Customer & technology partner
LayerJot logoLayerJot

Powers instrument identification and the catalog matching behind AVA's instrument tray application.

Visit LayerJot →

Frequently asked questions

What is AVA?

AVA stands for Agentic Visual Applications. It is the visual intelligence layer of JotVision’s system of action. It observes a physical scene, identifies relevant objects, people, states, and events, creates structured data and evidence, and triggers the workflow that should happen next.

How is AVA different from computer vision?

Computer vision detects what is visible. AVA understands what matters in the context of your stated intent, and JotVision turns that understanding into action.

How do I tell AVA what to look for?

In natural language. Describe the observation or outcome you care about, such as when a tray is complete or when a step is skipped. Intent can change without rebuilding the visual application.

Can AVA use my business data as well as video?

Yes. JotVision can combine what AVA observes with enterprise context from your ERP and operational systems, to determine whether something is missing, unexpected, or needs action.

Where is video processed, and what about privacy?

Footage and prompts can be processed locally on your premises or in the cloud, depending on your requirements. Faces and identities can be blurred in the analysis.

Which hardware does AVA work with?

AVA works with compatible PTZ cameras, including PTZOptics, and with robotics hardware such as Trossen Robotics arms for robotic visual inspection.

Can I build my own applications on AVA?

Not yet. We are working to open an API so you can build on AVA’s scene understanding, tracking, and event detection. Talk to us about early access.

Show us a scene worth acting on.

Tell us what AVA should see and what should happen next: instrument checks, inspection, lecture capture, room analytics, or an application of your own. We will walk you through a live demo.

Book an intro call →