New keynote for 2027 · Who Pays When the Agent Is Wrong?

A keynote on the question every AI agent demo skips.

Insurers are putting agents into underwriting, claims, and service faster than anyone is deciding who answers for them. My company runs these systems for carriers, MGAs, and TPAs. This talk is the accountability map I want every board to have before go-live.

What the talk covers

Somebody pays. Decide who before it happens.

Every demo ends with the AI making a decision. Every business ends with a licensed human, a vendor contract, and a regulator. When the agent is wrong, and it will be, one of them pays. Most organizations I talk to cannot say which.

I split every AI use into two kinds. Decision support buys back mistakes. Workflow automation buys back hours. They carry different risk, they need different controls, and they belong on different rungs of what I call the trust ladder, where an agent starts on drafts and summaries and earns its way up to acting on its own.

Then the part for risk managers and boards. What an agent has to log. What your vendor contract says about errors. What your own insurance does and does not cover when software makes the call. The market is still writing those answers, and the questions are already on conference programs, from RISKWORLD 2026 to the 2027 InsurTech America Symposium.

You leave with a one-page map you can take back to your general counsel, your CIO, and your next vendor meeting.

What the room walks out with

Four things to take back to the office.

The accountability map

Who answers at each step: the licensed human, the vendor, the carrier, and the regulator.

The trust ladder

What an agent may do alone at each rung, and what it has to prove before it moves up.

The log

What to record so you can explain a decision to a regulator, an auditor, or a jury.

The vendor questions

What to put in the next contract about errors, data, and who carries the loss.

Who it's for

Built for the people who sign for the outcome.

The audience

Risk managers, general counsel and compliance leaders, carrier and TPA executives, captive and self-insured fund boards, and CIOs.

Formats

Keynote, in person or virtual. Before every event I do a thirty-minute prep call with your program lead on who's in the room and the three things you want them to remember. I'm also around for VIP dinners, sponsor meet-and-greets, a podcast interview on site, and Q and A after the talk.

Proof

Rooms I've spoken in.

This is a new keynote for 2027. If you want to be one of the first rooms to hear it, the prep call is where we shape it for your audience.

Rooms I've spoken in

More than four hundred keynotes since 2005, including these.

  • 2026 RIMS Cumberland Symposium
  • 2026 ITC Vegas
  • 2025 RISKWORLD
  • 2026 World Captive Forum
  • 2026 InsurTech Insights NYC
FAQ

Questions organizers ask about this one.

Who is Who Pays When the Agent Is Wrong? built for?

Risk managers, general counsel, compliance, boards, and the executives who approve AI deployments. It works for risk and insurance conferences, captive and self-insured forums, and carrier leadership meetings.

Is this a legal talk?

No. I am an operator, not your lawyer. I show you the questions to ask and the records to keep, and your counsel decides what they mean for you.

How is it different from The Agentic Revolution?

The Agentic Revolution is what agents can do. This one is who answers for them when they get it wrong.

How far ahead should we book?

Premium slots fill four to nine months out, so earlier is better. Send the date, the audience, the slot length, and the three things you want the room walking out remembering.

Other keynotes

Looking for a different angle?

Or see every topic, the sizzle reel, and where I'm speaking next. Or watch full talks and clips.

Book James

Bring Who Pays When the Agent Is Wrong? to your room.

Send the date, the audience, and the slot length, and we'll go from there. If this isn't the right fit, the other keynotes are on the speaking page.