For the CRO · Chief Risk Officer

AI you can put in front of your regulator, your board, and your model risk committee.

You carry the model risk, the regulatory exposure, and the reputational risk of every AI decision the business ships. FPTInsure builds AI to your standards, not around them — evidence-first, HITL by design, and mapped to the regulations that apply to your book.

What you're up against

The problems we hear from every Chief Risk Officer we talk to. If two of these match your week, keep reading.

Model artifacts arrive after the model

Business ships an AI feature; risk is asked to bless it three weeks later. Nothing about the process is defensible.

Fairness testing that doesn't hold up

Vendors quote 'no bias' with no methodology, no cohorts, no thresholds. Won't survive a market-conduct exam.

EU AI Act clock is running

High-risk obligations phase in through 2026–2027. Your board wants a plan and evidence, not a slide.

Third-party AI oversight is thin

NAIC Bulletin, NYDFS Circular 7, and SR 11-7 all demand it. You need vendors who arrive with the artifacts, not a shrug.

What good looks like

Every model
Ships with a model card, eval report, and HITL pattern
8 stages
Governed evaluation methodology, artifacts named at each stage
Independent
Second-line validation on every high-risk model
Kill-switch
Documented, tested, on every production model

Where we bet with you

The four moves that shift your metrics fastest — and where in our stack they live.

Model cards, evaluation reports, validation memos

Produced during build — not retrofitted for an audit. Everything your model risk committee needs, in the format they use.

See Responsible AI

EU AI Act + NAIC + SR 11-7 alignment

Risk-tier classification, Annex IV technical file, bulletin control mapping, model documentation — mapped to Articles 9–15 and Board reporting.

See trust & compliance

Bias & fairness testing that survives an exam

Demographic parity, equalised odds, adverse-impact ratio, calibration by cohort. Colorado SB21-169 methodology supported for life carriers.

See Responsible AI

Human-in-the-loop by design

Every coverage decision, decline, cancellation, or SIU referral is a Gate pattern. The AI never denies a claim.

See HITL patterns

Proof to send your team

Three links that get the rest of your organisation onside.

Trust centre

SOC 2, ISO 27001, HIPAA, GDPR, EU AI Act, NAIC, SR 11-7, DORA — control mappings and evidence.

Open →
Responsible AI

Full model governance programme — model cards, evals, fairness, regulatory alignment.

Open →
Insights

POVs on regulatory change, model risk, and post-market monitoring.

Open →

What you'll be asked

The questions your peers, procurement, and board will fire back at you. Our answers.

Every vendor says they're 'EU AI Act ready'.+

We ship the Annex IV technical file, the risk management system record, the data governance record, the logging design, and the post-market monitoring plan. Ask any vendor for those five artifacts. Watch the room go quiet.

Who owns the model?+

The carrier does. We build to your governance framework and deliver a model file that lives in your inventory. On exit, you keep the model, the docs, and the monitoring dashboards.

What if a regulator asks us questions we can't answer?+

That's what the evidence pack is for — model card, eval report, fairness report, validation memo, monitoring log. We'll sit next to your team in the review if needed.

How do you handle a fairness threshold miss?+

Documented remediation workflow: root-cause analysis, retrain or re-scope, re-test, sign-off. No silent overrides, no 'ship and see'.

Two ways to take the next step.

Pick the one that fits where you are.

Model governance walkthrough

60 minutes with our model risk lead — we walk you through the artifacts we produce and how they map to your framework.

Book a walkthrough
Evidence pack request

SOC 2, ISO 27001, sample model cards, EU AI Act technical file template — under NDA.

Request evidence