AI can only act on what it can see.
That was one of the central ideas behind our Digital Eyes session at Digital Insurance CEE 2026 in Warsaw, where Rohit Tandon, Founder & CEO of Elevondata, spoke about connecting real-world inspection evidence with underwriting and claims workflows.
The insurance conversation around AI often starts with models: what they can detect, how accurate they are, and where they can support or automate decisions.
There is an earlier problem that deserves just as much attention.
Was the right evidence captured in the first place?
A strong damage model cannot compensate for the wrong photograph. An underwriting model cannot assess a condition that was never documented. And a claims team gains little from receiving more images if those images arrive without enough context to understand what was captured, when, and under what circumstances.
For inspection technology, the quality of the downstream decision starts upstream.
One inspection, more than one way to complete it
Not every risk needs the same inspection method.
Some policyholders can complete a guided Self Inspection themselves. Others may need help from an agent. Certain risks still justify an on-site professional.
The important point is that these do not need to operate as separate processes.
Digital Eyes supports Customer Self Inspection, Remote Assisted Inspection, and Agent Field Inspection within the same platform. An inspection can move between these modes while retaining the same inspection order and evidence record. Changing the inspection method does not require the carrier or customer to start again.
This is as much an adoption issue as a technology issue.
Self Inspection does not have to work for every customer in every situation. If the customer needs help, the inspection can move to Remote Assisted. If the risk genuinely needs someone on-site, it can move to Field Inspection. The Field Inspection application also supports offline working where connectivity is limited.
The objective is not to force every inspection into a digital channel. It is to use the right channel for the risk and the customer.
Use the right inspection mode for the risk and customer.
The goal is not more photos
Capturing evidence is only useful if the information can support a decision afterwards.
An inspection may include images, video, answers, required fields, location information, timestamps, and other metadata. Digital Eyes brings these elements together within the inspection record, while allowing the evidence requirements to vary depending on the risk being assessed.
A simple way to think about the workflow is:
Capture. Validate. Analyze. Deliver.
If an underwriter simply receives another folder containing 30 photographs, the work has not necessarily been reduced. It may only have moved from the field to the desk.
The value comes from preserving the context around those captures and turning them into information that can support review, AI analysis, reporting, or a handoff into the insurer's existing systems. Digital Eyes can return that output through configurable reports, structured data, and system integrations, depending on the carrier workflow.
That is a different objective from simply digitizing photo collection.
The same evidence problem appears after a loss
The same design principle applies in claims.
At First Notice of Loss, a carrier may need the accident scene, areas of damage, close-up images, vehicle registration details, and structured incident information. Capturing these elements earlier gives the claims team a more useful starting point before the first physical assessment takes place.
Property underwriting and motor claims are very different insurance moments, but the evidence problem is similar.
The carrier needs to define what should be captured, guide the person collecting it, preserve the context, and make the result usable by the next person or system in the process.
The pattern remains consistent:
Guide the capture. Structure the evidence. Apply intelligence. Deliver the result.
AI is a journey, not a one-time vendor decision
The AI layer introduces another consideration.
Models are changing quickly. Different use cases may also require different specialist capabilities. The model used for property risk may not be the one a carrier wants for motor damage, OCR, or document intelligence.
That makes tight coupling between the inspection workflow and a single AI provider a potentially limiting design choice.
A more durable approach is to separate the evidence workflow from the intelligence applied to it.
Digital Eyes can work with Digital Eyes AI, insurer-owned models, and third-party AI services. The inspection and evidence workflow can remain consistent while the intelligence layer changes over time.
For carriers, this matters because AI adoption is unlikely to be a one-time technology decision. Models will improve, specialist services will emerge, and internal AI capabilities will mature.
The inspection infrastructure should not need to be rebuilt each time that happens.
Inspection workflows need to change as the business learns
Inspection requirements rarely stay fixed.
An underwriting team may want to add two roof questions. Claims may need another damage photograph. A carrier may decide that a field should become mandatory, introduce a new validation rule, or change the output sent downstream.
Those changes should not automatically become software-release projects.
Digital Eyes can reproduce existing carrier forms and inspection structures, including sections, zones, required media, conditional logic, and validation rules. Routine changes are managed through backend configuration and do not require a new application release.
That means straightforward changes to the inspection workflow can typically be made in days rather than weeks.
The same applies to outputs. Reports, structured data, and downstream handoffs can be configured around the carrier's operating requirements. More substantial integrations or genuinely new business logic may still require engineering work, but day-to-day changes to the inspection process do not need to wait for another product release.
This is an important part of time to market. Getting a workflow live is only the first step. The greater operational value comes from being able to improve it as the business learns.
Connecting the physical world to the insurance decision
Digital inspection is sometimes framed mainly as an alternative to sending someone into the field. That is too narrow.
The larger opportunity is to create a reliable connection between something that exists in the physical world and the people, systems, and AI making decisions about it.
That requires more than a camera.
It requires the right evidence to be captured, its context to be preserved, and the resulting information to be made usable by underwriting or claims.
That was the central idea behind our discussion in Warsaw, and it is the problem Digital Eyes is designed to address.
Learn more about Digital Eyes at digitaleyes.app.
