Can AI Catch the Proposal Errors That Turn Into E&O Claims?

By /

The short answer

That’s the claim Johnson Kendall Johnson is making about its Outmarket rollout, and the proposal speed is not the interesting part of it. JKJ, an independent, employee-owned brokerage in Pennsylvania, says putting together a client proposal dropped from a couple of days to a matter of hours after it adopted Outmarket’s AI platform. Fine — every vendor story this year has a clock in it. The line in the release that matters to a broker’s E&O carrier is a quieter one: the platform “catches missed forms or policy gaps that human review occasionally overlooks,” and seasoned veterans are using it to double-check their own work. That is not a speed feature. That is an error-catching layer sitting upstream of the proposal, before the mistake has a chance to become a claim.

Everything below is drawn from the two companies’ announcement, and it should be read that way: these are JKJ’s and Outmarket’s reported results from their own rollout, not independently measured outcomes. The desk’s job here is to say what the claim is, where it sits in the E&O chain, and what a broker should ask before treating it as proven.

What did JKJ actually report?

JKJ didn’t flip a switch firm-wide. Per the release, the agency ran a six-month pilot with a focused group of 10 employees, and the pilot is what paved the way for a full-team rollout. The reported results:

  • Proposal assembly: time spent putting together a client proposal “dropped from a couple of days to a matter of hours.” Workflows like manually cross-referencing complex Excel spreadsheets and coverage documents, which previously took hours, now take minutes.
  • Error catching: the AI “successfully catches missed forms or policy gaps that human review occasionally overlooks.” Even veterans with decades of experience are using it to double-check their work, per the release.
  • Consolidation: JKJ housed AI workflows, data structuring, and data visualization in one platform instead of buying multiple point products.
  • Client work: time saved on manual work is being reinvested in consultative work, and JKJ says it can now scale its claims analytics reports to a broader range of clients.

Outmarket’s own boilerplate is explicit about the pitch: it says over 300 agencies use the platform “to reduce E&O exposure, accelerate client delivery, and unlock new revenue.” Note the order in that sentence — and note that it is the vendor’s description of itself, not an audited result.

Where does this sit in the E&O chain?

Upstream of the control this desk has already covered. The policy check — comparing the carrier-issued policy against what was sold — is the classic E&O front line, and it is rapidly becoming an agent’s job. But the policy check happens after issuance. It catches the discrepancy between what was promised and what the carrier bound.

The JKJ story is about the layer before that: the proposal and submission package itself. A missed form, a gap in the coverage schedule, a limit that doesn’t match what the producer discussed — those errors originate inside the agency, in documents the agency produced. If they’re caught at proposal stage, they cost an edit. If they’re caught after a loss, they cost a claim against the agency’s own errors and omissions insurance — and the client’s version of events will be that the proposal said one thing and the policy delivered another.

That is the honest framing of the E&O angle here. An upstream check doesn’t replace the policy check; it shortens the list of ways the two documents can disagree in the first place. The fewer original errors in the proposal, the fewer discrepancies there are for anyone — human or agent — to hunt for later.

Why does catching a missed form matter more than the speed?

Because speed is recoverable and errors compound. A proposal that goes out a day late is a service embarrassment. A proposal that goes out with a missing form or an unnoticed coverage gap is an exposure the client is relying on and the agency has, in writing, appeared to arrange. E&O claims are born in exactly that gap between what the client reasonably believed was handled and what was actually in the paperwork.

The detail worth sitting with in the JKJ release is who is doing the double-checking: “even seasoned veterans with decades of industry experience.” The conventional pitch for checking tools is that they protect the agency from its newest hires. JKJ’s framing runs the other way — the experienced producer uses the machine as a second set of eyes precisely because experience doesn’t exempt anyone from the end-of-day miss. That’s consistent with how these claims actually happen: rarely a producer who didn’t know, often a producer who was busy.

What does this do to training the next generation of producers?

This is the release’s own second headline, and it’s the part that makes the story an operations story rather than a software story. JKJ says it recruits out of college and that entry-level roles used to consist of “menial, tedious data-entry tasks” that frustrated new hires. With the manual assembly work automated, entry-level staff reach strategic, client-facing work faster, and JKJ projects — its word is “projected” — that training time to become a fully trained employee will drop from a year to six months.

Chad Czapor, a JKJ vice president, said the agency believes its account managers’ capacity “will almost double,” and that the freed time goes to “training and development for our people.” Patrick Mullen, JKJ’s president of commercial insurance, put the agency’s thesis in one line: “It’s not just about time savings, it’s about empowering people to do things they couldn’t do before.”

Hold both of those as what they are: a projection and a belief, stated by the agency’s leadership in a partnership announcement. But notice the shape of the claim, because it’s testable in a way most AI announcements aren’t. If the tedious first year really was where new producers learned the documents — the forms, the endorsements, the places gaps hide — then removing that year creates a training question, not just a training saving. The producer who never assembled the package by hand has to learn error recognition some other way. The agencies that get this right will teach the checking as a discipline. The ones that get it wrong will have fast proposals reviewed by people who’ve never seen the slow version fail.

What should a broker ask before believing the headline?

Four questions, in order:

  1. Pilot or proof? JKJ’s rollout started with 10 employees over six months. That’s a responsible shape — and it means the days-to-hours figure comes from a focused group, not a whole book of business. Ask any vendor which population a number describes.
  2. Measured or projected? The proposal time is reported as a result. The training halving is a projection, and the capacity doubling is a stated belief. The release itself keeps those categories separate; summaries of it often don’t.
  3. Where’s the E&O evidence? Neither company offers E&O loss data in this announcement — no claim frequency, no near-miss counts. “Reduce E&O exposure” is Outmarket’s positioning language. The mechanism is plausible: catching gaps upstream should mean fewer downstream disputes. Plausible mechanism is not the same as a demonstrated loss record, and your E&O carrier will price the difference.
  4. Who answers when the checker is wrong? An error-catching layer becomes part of the workflow people rely on, which makes its own misses a new category of problem — the territory this desk covered in who pays when an AI agent makes the mistake. The approval and review discipline around the tool is not a footnote; it’s the control.

FAQ

What did JKJ say Outmarket changed about proposals?
Per the companies’ announcement, time spent putting together a client proposal dropped from a couple of days to a matter of hours, and cross-referencing spreadsheets and coverage documents went from hours to minutes. These are the agency’s reported results from its own rollout.

How does catching missed forms reduce E&O exposure?
The mechanism: a missed form or policy gap caught at proposal stage costs an edit; the same gap discovered after a loss becomes a dispute over what the client was promised versus what was delivered — the classic shape of an agency E&O claim. Outmarket markets this as E&O exposure reduction; the announcement contains no E&O loss data demonstrating it.

Is this the same as AI policy checking?
Related, but a different layer. Policy checking compares the carrier-issued policy against the agency’s records after binding. The JKJ use case sits earlier, in assembling the proposal and submission package itself — catching the agency’s own errors before they leave the building.

What did JKJ say about training and capacity?
JKJ said entry-level hires reach client-facing work faster with data entry automated, and projected training time to fully trained will drop from a year to six months. A vice president said the agency believes account manager capacity will almost double. Both are forward-looking statements from the agency, not measured results.

Should my agency read this as proof the platform works?
Read it as a credible-shaped claim: a named, 70-year-old agency, a six-month pilot before full rollout, and specific operational details. That’s more substance than most partnership releases carry. It is still a partnership announcement — the results are self-reported, and the E&O benefit in particular is argued from mechanism, not from loss data.

Sources

This article is informational only and is not insurance, legal, or financial advice for any specific agency or policyholder.

Scroll to Top