Agentic AI Just Went Live in Underwriting. Total Loss Is Still Waiting.

A few weeks ago we argued underwriting was racing ahead while total loss stayed manual. This week it stopped being an argument and became a fact. Three agentic underwriting launches landed in a fortnight. Here is what actually shipped, why total loss is still the last manual link, and what that means for the AMC sitting in the gap.

At the start of this month we made a case. Underwriting was automating quickly, total loss settlement had barely moved, and the gap between the two was widening in a way that mattered for the accident management companies sitting between them. That was framed as a direction of travel.

Two weeks later it is not a direction of travel. It is a set of product launches. In the space of a fortnight, three separate agentic underwriting tools went from announcement to live. If you settle total loss claims for a living, this is the clearest signal yet of where the automation frontier actually is, and how far in front of you it now sits.

What actually shipped

Three concrete things, all in the first half of July.

Duck Creek acquired Send, an AI-native underwriting orchestration engine, and described the result as an agentic underwriting-to-core platform. The pitch is that a submission can flow from broker email through to the core policy system with agents doing the orchestration in between.

Hyperexponential launched a tool it calls hyperoperator, an agent built to carry a commercial submission from the broker's first email through to a priced, quoted risk. Its own demonstration ran a cyber submission from email to a triaged and priced risk in under three minutes, with no manual data entry.

And Cytora's chief executive said the quiet part out loud. On the back of its Zurich deployment, running live across five countries after a 90-day rollout, he said agentic AI is now beyond pilots. Zurich reported manual triage time falling by around 80 percent. That is not a proof of concept. That is production.

Three vendors, three launches, one message. Agentic AI in underwriting is no longer coming. It has arrived.

Why underwriting went first

None of this is an accident of who had the better engineers. Underwriting went first because it is the part of the chain best suited to agents.

A submission is structured. A broker email, a schedule, a set of risk factors, a pricing model with defined appetite and delegated authority. The rules are written down. The volume is high and repetitive. That is precisely the profile where an agent earns its keep: lots of similar decisions, a clear rulebook, and a measurable output at the end. Feed it the submission, let it triage and price against the model, put a human on the exceptions.

Total loss is not shaped like that, and that is exactly why it has been left until last.

Why total loss is still the manual link

A total loss settlement is a judgement wrapped in a negotiation. What is the vehicle actually worth. Which valuation guides apply and how do you weight them. Is there a pre-accident condition adjustment. Has the customer got a point about the comparable listings. What will this insurer actually accept, and where is the room. None of that reduces cleanly to a rulebook, because half of it is contested and the other half is customer-specific.

So while underwriting has been handing structured decisions to agents, total loss has stayed where it always was. A skilled person, a set of valuation tools, a spreadsheet, and a back-and-forth with the insurer that runs on experience rather than logic gates. It works, but it does not scale, and it is now visibly the slowest, most manual link in a chain that is automating on both sides of it.

That is the gap. Distribution and underwriting are being rebuilt around agents. Total loss is still hand-built, file by file. And the AMC is the business sitting right in the middle of that mismatch.

What this means for AMCs

Two things follow, and they pull in the same direction.

The first is a pricing and expectation problem. As insurers automate underwriting and intake, their internal clocks speed up. Straight-through processing on the underwriting side resets what fast looks like across the whole operation. A total loss process that takes days of manual work starts to look slower than it did, not because it got slower, but because everything around it got quicker. The AMC that still runs total loss entirely by hand will feel that pressure first.

The second is an opportunity, and it is the more interesting one. The reason total loss resists full automation is also the reason it is defensible. You cannot hand the judgement and the negotiation to an agent and walk away. But you can take everything around the judgement, the data gathering, the valuation assembly, the evidence pack, the first draft of the challenge, and structure it so the skilled person spends their time on the part that actually needs them. That is not automating total loss out of existence. It is taking the manual weight off it so the judgement gets more attention, not less.

The AMCs that come out of this well will be the ones who read the gap correctly, structured the routine work around the judgement, and kept the human exactly where the value is. The vendors proved this month that the frontier has moved. The question for total loss is not whether to catch up on the judgement. It is whether to keep carrying the manual weight around it by hand.

Where TotalSettle fits

TotalSettle is built for precisely this gap. It does not try to replace the settlement judgement, because that judgement is the job. What it does is take the manual weight off the process around it. Valuation evidence assembled in one place, the arguments structured and grounded, the file built so the negotiation starts from a strong, documented position rather than a blank page.

That is the right shape for a part of the chain that is judgement-led rather than rules-led. Underwriting could hand the structured decision to an agent. Total loss cannot, and should not. But it can stop being hand-built from scratch every time, and that is the difference between feeling the speed pressure from automated underwriting and being ready for it.

TotalSettle is the total loss settlement platform for accident management companies, built to take the manual weight off valuation and settlement while keeping the judgement with your team. Book a demo.