For P&C carriers, MGAs, brokers, claims teams and the vendors that sell to them · Florida first
It asks for an inventory of the AI you use — including features your vendors switched on — who owns each item, and the evidence behind your high-risk systems. In three weeks you have it on file; in thirty days one claims or underwriting workflow runs the same way.
The bulletin itself: 27 states plus DC and Puerto Rico, as of 8 October 2026. Florida hasn't adopted it.
Three weeks: the inventory, named owners and evidence file, in the form examiners ask for. Thirty days: one claims, underwriting or brokerage process rebuilt as a governed workflow to prove it works.
Same method I used to take AI use cases through model-risk, information-security, algorithmic-bias and financial-crime review inside a global bank — cut to the size of a Florida carrier, broker or MGA.
Why this year
The bulletin does not ban anything. It asks you to prove three things — and vendor-run AI counts as yours.
Most carriers' first AI isn't one they built — it's a feature a vendor switched on. You're accountable for it anyway.
Only 24% of insurers say they're very confident they'd pass an independent AI governance review within 90 days; 68% say the controls exist but the evidence is scattered. (Grant Thornton, 2026 insurance AI survey, 100 insurers)
Florida has not adopted the bulletin. The questions reach Florida carriers anyway — through market-conduct exams, reinsurer and carrier-partner diligence, and D&O/E&O renewals.
How you select, validate, monitor and retire AI — including the claims-triage and fraud models your vendors run. Produced on request.
Every AI system touching underwriting, claims or pricing, with a model owner and a governance lead the examiner can call.
Testing records, monitoring logs, vendor documentation, and complaint logs tied to AI-influenced decisions.
Adoption count: 27 states plus the District of Columbia and Puerto Rico, as of 8 October 2026 (NAIC adoption map). Florida has not adopted it. Not legal advice — your compliance counsel confirms what applies to your domicile and lines.
The questionnaire, answered
| The questionnaire asks for | What you'll have after the Governance Setup |
|---|---|
| Exhibit A — AI usage inventory | Every AI system in use, including features vendors switched on, with purpose, data and an owner |
| Exhibit B — governance and risk assessment | Named accountability, the intake and risk-tier rule, and the evidence log behind it |
| Exhibit C — high-risk system detail | A file for each high-risk system: testing, monitoring, human review, rollback |
| Exhibit D — AI data detail | Data sources, lineage notes and the vendor terms that govern them |
Exhibit names per the NAIC AI Risk Evaluation Supplement pilot (2026). This is preparation, not legal advice.
Which one are you?
Find yours. The engagement is sized to it — not to a framework.
| Situation | The question on the board's mind | First move |
|---|---|---|
| New entrant / reciprocal | "How do we start governed, so the first exam isn't a clean-up?" | A 3-week Governance Setup that installs the inventory, owners, intake rule and vendor clauses once — no legacy to unwind. |
| TPA / claims services / MGA | "Can we take hours out of assignment, estimate QA or subro — and prove control to the carrier?" | One process in 30 days, with the control evidence packaged so it answers your carriers' questionnaires, not just your own. |
| Established FL carrier | "What AI is in our claims process today, and who signs off on it?" | A claims-triage or FNOL build — one process rebuilt with human-in-the-loop and a rollback switch; the AI-systems-program paperwork written as a by-product. |
| We build AI in-house | "Who answers the reinsurers' AI questionnaires without putting a gate in front of every release?" | The Governance Setup, run against your release process — risk tier set at the ticket, test evidence from your pipeline; the diligence answers written from it. |
| PE-backed broker | "Which brokerage workflow do we automate first, and how do we show the number by day 100?" | A 30-day Build on one back-office process — submissions, renewals, certificates, carrier reconciliation — metric agreed on day one. |
| We sell AI to carriers | "What will our insurer customers ask us about our AI — and what do we hand them?" | The Governance Setup, run on your product: registry entry, test evidence, and the audit-rights and incident-notice language carriers' risk reviewers ask for. |
I work alongside your counsel and auditors, not instead of them.
Health insurers, health-side TPAs and benefit administrators are outside this practice.
One claims workflow — first notice, document intake or triage — rebuilt in 30 days, with the questionnaire paperwork written while we build, not after.
Book a 30-minute callHow we work together
7-day Check → Governance Setup → 30-day Build → Agent Care / Fractional CAIO
The list of AI actually in use — including the features your vendors switched on — with a name next to each item, a one-page scorecard, the three gaps to fix first, and the questions the examiner, regulator or underwriter will ask. 30-minute walkthrough.
$4,500 · 7 days · credited in full to a Build or a Governance Setup within 90 days · you send the system list and vendor contracts in the first 48 hours
Book a callWhat AI you run, who owns it, and the evidence file — mapped into your GRC and model-risk process. Inventory incl. vendor models, named owners, intake rule, risk tiers, vendor clauses, and a gap list against the NAIC bulletin and the AI Risk Evaluation Supplement.
Fixed fee · 3 weeks · scoped on a 30-minute callFrom $19k · 3 weeks
Book a callOne claims, underwriting or brokerage process rebuilt as a governed AI workflow your team runs on day 31: human-in-the-loop, rollback switch, metric agreed on day one. Runbook, owner, test evidence and sign-off trail.
From $19k · 30 days · one workflow, one system integration · up to 30% paid on the day-30 number
Book a callAgent Care — from $3,500/month. We keep your agents running, fix what drifts, and ship the next small change. Fractional CAIO — from $8k/month, one day a week. Someone senior who owns the AI program, vendor reviews and board and examiner-facing reporting.
3-month initial term, then monthly
Book a call| Check | Governance Setup — examiner edition | Build | Agent Care / CAIO | |
|---|---|---|---|---|
| What you get | The list of AI in use with a name next to each item, a one-page scorecard, the three gaps to fix first, the questions the examiner will ask, a 30-minute walkthrough. | Inventory incl. vendor models, named owners, intake rule and risk tiers, vendor clauses, gap list against the NAIC bulletin and the AI Risk Evaluation Supplement. | One claims, underwriting or brokerage process rebuilt as a governed workflow your team runs on day 31. | Agent Care: agents kept running, drift fixed, the next small change shipped. CAIO: someone senior who owns the AI program, vendor reviews and reporting. |
| Days | 7 | 3 weeks | 30 | Monthly |
| Price | $4,500, credited in full to a Build or a Governance Setup within 90 days | Fixed fee, scoped on a 30-minute call | From $19k, one workflow, one system integration · up to 30% paid on the day-30 number | Agent Care from $3,500/month · Fractional CAIO from $8k/month, one day a week · 3-month initial term, then monthly |
| Who it is for | Any carrier, MGA, broker or claims firm that needs the AI list first. | New entrants, in-house builders, vendors selling to carriers — anyone an examiner, reinsurer or carrier partner is asking. | Anyone who knows the process. | Firms with agents to keep running, or an AI program someone senior should own. |
| What makes it stick | — | — | Named owner and trained operators · decision log and an off switch · evidence pack · 30-day warranty | — |
| Then | A Build or a Governance Setup — the Check's fee comes off either. | A Build on the first use case, or Agent Care / a Fractional CAIO to run the program. | Agent Care, or the next process. | The next small change, every month. |
All four are fixed-scope. The 7-day Check is $4,500, credited in full to a Build or a Governance Setup within 90 days; the Governance Setup is scoped on a 30-minute call; the 30-day Build is from $19k with up to 30% paid on the day-30 number; Agent Care from $3,500 a month and a Fractional CAIO from $8k a month, one day a week — each on a 3-month initial term, then monthly.
What I need in week one: your vendor list and contracts, the system list from claims, underwriting and IT, and a few hours of your leads' time.
What I don't need: policyholder or claimant records.
What I don't do: legal opinions, a new committee, or a build your team can't run on day 31.
What's on record on day 30
| Examiner asks for | What exists afterwards — and which engagement produces it |
|---|---|
| Inventory of AI systems | Governance SetupRegister of every AI touching claims, underwriting or pricing — yours and your vendors' — with purpose, data, and decision impact. |
| Named accountability | Governance SetupA model owner and a governance lead per system, with the line of authority to the executive team written down. |
| Controls mapped to existing process | Governance SetupIntake rule and risk tiers wired into the GRC, vendor-management and model-risk process you already run — no parallel committee. |
| Evidence of testing and monitoring | BuildPre-deployment test record, human-review sampling, monitoring log and the rollback procedure for the rebuilt process. |
| Vendor oversight | Governance SetupDocumentation requests, audit-rights and incident-notice language ready for the next renewal or RFP. |
| One process that proves it works | BuildA live, governed workflow with a day-30 metric — the number you take to the board. |
Why me
Program Director for AI strategy and execution in the third-party risk function of a global bank. I created the AI strategy for a regulated decision function and then ran the program — taking use cases through model-risk management, information security, algorithmic-bias, financial-crime and third-party review, end to end, with the engineers.
The method is the bank's. The scope is one process, one examiner question, three weeks — sized for a carrier, broker or MGA, not a global bank.
Before that: 300+ automation projects for Fortune 500 clients, and a decade of building the workflows myself. I work inside your team, not from the outside — the capability stays in-house when I leave.
Based in Tampa Bay; on-site across Florida.
Questions insurers ask
Because the same questions reach you anyway — through market-conduct exams, reinsurer and carrier-partner diligence, and D&O/E&O renewals — and the inventory takes the same three weeks whoever asks first. Twenty-seven states plus DC and Puerto Rico have adopted the bulletin as of 8 October 2026; Florida has not.
No. The bulletin's position is that the insurer, not its vendor, answers for AI used in its decisions — your counsel confirms how that applies to you. The AI Risk Evaluation Supplement (piloted in 12 states since March 2026; earlier drafts were called the AI Systems Evaluation Tool) asks for the same things. What you need is a register, the vendor's documentation, audit and incident-notice language in the contract, and your own monitoring — none of which the vendor supplies unprompted.
Only inside your environment, under your access controls, and only where the rebuilt process needs it. Nothing leaves your tenant. The governance work itself — inventory, owners, evidence — runs on metadata, not records.
Klek Consulting LLC carries professional and cyber liability cover; NDA and MSA templates are available on request.
Yes. The Governance Setup is three weeks and the Build is one process; both are sized for a single carrier, broker or MGA, not a group. New entrants get the simplest version: inventory, owners and vendor clauses written once, before there is anything to unwind.
We agree one metric on day one — cycle time, touch rate, hours per file. A portion of the Build fee is only invoiced if that number is hit on day 30. Same method as every automation program I have run: the number is the deliverable. The baseline is measured from your system in week one, the at-risk share is written into the statement of work, and the day-30 number is read together from your data, not mine.
Next step
I'll tell you whether it's a 7-day Check, a Governance Setup, or not worth your money yet.
