Indian Insurers' AI Wins Are All in Motor and Health: Your Property and Liability Claims Are Still Manual

Read the FY26 disclosures and press coverage from Indian general insurers and one pattern is hard to miss. Where an insurer has put a number against artificial intelligence in claims, the number belongs to motor or health.

Business Standard reported on 16 July 2026 that ICICI Lombard moved generative AI out of pilots and into day-to-day operations during FY26, spanning underwriting, claims, servicing, marketing and enterprise productivity. The named initiatives are retail-shaped: One IL One Call Centre, built on AI, voice bots and propensity modelling, and Project Orion, aimed at turnaround times. The same reporting cycle covered Bajaj Allianz's proprietary AI engines, where photo-based damage detection has cut motor settlement times to under two hours in select pilot cities, on a human-in-the-loop model. On the health side, Whalesbook reported on 21 August 2026 that Star Health's stated target is to automate more than 50 per cent of cashless claims within two years.

Those are real results and they are worth taking at face value. What they are not is evidence about your fire, marine or liability programme. No Indian insurer has published a comparable figure for commercial property or casualty claims in this reporting cycle, and the absence looks structural rather than accidental.

Machine learning gets cheap when three conditions hold together: high claim volume, low variance between claims, and a decision that can be reduced to evidence a camera or a form can capture.

Motor own-damage satisfies all three. A dented bumper photographed on a mobile phone looks much like the last hundred thousand dented bumpers, the repair cost sits inside a narrow band, and the parts catalogue is finite. Cashless health is close behind. The hospital is on a network, the tariff is pre-agreed, the diagnosis maps to a code, and the adjudication question is largely about policy terms and exclusions applied to structured data.

Commercial lines invert every one of these:

• Volume is thin. A mid-sized insurer may see a few hundred material fire losses in a year against millions of motor claims. There is no training set.
• Every claim is bespoke. Two factory fires of identical sum insured can differ by an order of magnitude in reinstatement cost, business interruption period and salvage position.
• The evidence is not photographable. Quantum on a large property loss turns on reinstatement estimates, contractor quotations, production records, and an average clause calculation on declared values. On liability, it turns on legal exposure that has not crystallised yet.

So the commercial claim still runs the way it always has, through a licensed surveyor, a loss adjuster on the larger files, and a long correspondence trail between the insured, the broker, the insurer and often a reinsurer.

The honest answer is that it works on the paperwork around the claim, not on the claim decision itself.

The realistic deployments are document handling: pulling structured fields out of a surveyor's report, matching invoices against a claimed schedule, chasing missing documents, checking that an intimation has the policy wording clauses it needs to cite. Useful work, and it compresses the administrative dead time that makes a six-month claim feel like a nine-month claim. It does not compress the parts of the timeline that actually dominate: appointing the right surveyor, getting the site inspected before evidence is disturbed, agreeing the basis of loss, and getting money released while quantum is still open.

The practical reading for a risk manager is that AI has changed your motor fleet claims experience and your group health experience, and has left your property and casualty experience roughly where it was in 2020.

This asymmetry is visible in the supervisory work as well. Insurance Business Asia reported on 23 June 2026 that the terms of reference for IRDAI's AI working group name claims processing and fraud detection as the priority areas for its explainability framework. There is no separate treatment of commercial lines claims in that framing.

That is a defensible allocation of regulatory attention. The consumer harm from an opaque model is largest where the model is deciding at scale against individuals who cannot negotiate: a rejected health pre-authorisation, an automated motor assessment a policyholder cannot interrogate. A corporate insured with a broker and a loss adjuster is not the party the explainability framework is protecting.

But it does mean a corporate buyer should stop waiting for regulation to deliver faster commercial claims. Nothing currently in flight is aimed at your fire claim's cycle time. The IRDAI policyholder protection framework already sets outer limits on surveyor appointment and report submission. Those are ceilings, not service standards, and treating them as targets is how a large loss drifts.

If AI is not going to move your commercial claims, the negotiable levers are the human ones. Two of them carry most of the value, and a third, the escalation ladder, is covered in the next section.

Surveyor appointment and mobilisation

The single largest controllable delay on a large property or marine loss is the gap between intimation and a surveyor standing on site. Push for named commitments in the placement slip:

1. Appointment within a fixed number of hours of intimation for losses above an agreed threshold, not merely within the regulatory outer limit.
2. A pre-agreed panel of two or three surveyor firms per major location, with the sectoral experience your risk needs, so nobody is searching for an engineering surveyor at 2 a.m.
3. Site attendance within a stated window for losses above the threshold, with a documented escalation if the appointed firm cannot mobilise.
4. An agreed protocol on preservation of evidence and on when you may begin clean-up and salvage handling without prejudicing quantum.

A named panel is the highest-yield item on this list and the easiest to concede at placement, because it costs the insurer nothing at inception. It is worth more than any turnaround assurance in a brochure.

Interim payment triggers

On a claim where quantum takes months, cash timing matters more than the final number. Negotiate on-account payments as a mechanical trigger rather than as a favour to be requested. The workable formulation ties a first release to the surveyor's preliminary or interim report and an agreed percentage of the amount not in dispute, with subsequent releases at defined stages. Detail matters: state who authorises, within how many days of the triggering report, and what happens when a reinsurer's approval is in the chain. See our note on on-account interim payments in large property claims for how these clauses are structured in practice.

For programmes above a meaningful size, the escalation path should be settled at placement rather than improvised in month three of an open loss. Agree a named claims manager at the insurer, a defined trigger for appointing an independent loss adjuster, a stated response time on coverage positions, and a standing joint review cadence once a claim passes an agreed age or value.

Each of those has a failure mode it is written against. A named claims manager stops the file being re-explained to a new handler every quarter, and the name should come with a deputy, because a single point of contact who changes roles mid-claim is a common cause of a stalled file. A written adjuster trigger, expressed as a value threshold or a multi-peril test, removes the argument about whether an independent adjuster is warranted at the moment the loss is least understood. A response time on coverage positions matters because a reservation of rights that arrives in month five, after the insured has spent on reinstatement, is worth far less than the same letter in week two.

Set the joint review cadence by claim age as well as value, since a mid-sized claim that has not moved in ninety days needs attention as much as a large one. Fix who attends from the insured, the broker, the insurer and the surveyor, and require a short written note of what was agreed and what each party owes by the next review. That note is what turns a review into progress, and it is also the record that supports the insured's position if the claim later goes to a formal escalation with the insurer or to arbitration under the policy.

Dropping the turnaround promise does not mean dropping AI from the conversation. It means asking for the things the technology genuinely delivers on low-volume, high-value claims.

Reasonable asks at renewal:

• A single document portal for the claim, with automated acknowledgement and a running status of what is outstanding, so document chasing stops being email archaeology.
• Structured extraction from surveyor reports so the insurer's file and your broker's file hold the same numbers, and reconciliation disputes surface in week two rather than month five. Our post on AI extraction from surveyor reports covers what this can and cannot pick up.
• Automated ageing alerts that flag any claim past an agreed number of days without movement, escalating to the named claims manager rather than sitting in a queue.
• Triage routing that gets a complex multi-peril loss to a senior adjuster on day one instead of after a junior handler has spent three weeks on it.

Each of these is a workflow improvement with a measurable output. None requires the insurer to make a promise about a settlement clock it cannot control, which is precisely why an insurer can agree to them.

Bring the asymmetry into the room directly. When an insurer presents its AI capability at renewal, the useful questions are narrow.

Ask which lines the quoted numbers come from. If the answer is motor and health, say so out loud and move to commercial. Ask what proportion of the insurer's commercial property claims closed within six and within twelve months over the last three years, and what the median time from intimation to surveyor appointment was on losses above one crore. Ask how many on-account payments were released on commercial property claims last year and at what average percentage of the eventual settlement. Ask who signs off an interim payment and whether that authority sits in the same city as your risk.

Insurers with genuinely strong commercial claims operations can answer these. Insurers whose AI story is entirely retail will change the subject back to the call centre, and that itself is the answer you needed.