Product
Five parts, one file your underwriter can actually work.
RiskBorne sits between the people who want cover and the people who write it: we turn the carrier's intake into something an applicant can act on, and the applicant's evidence into what the carrier needs to quote.
The insurability assessment
Twenty minutes, no integrations, no code.
Six guided steps collect everything the requirement corpus needs to evaluate you. You describe each agent in plain language and the assessment structures it for you; you confirm every field before it is stored. Nothing connects to your systems, so nothing has to clear a security review before you can start.
What comes out is an attested record: who said what about which agent, and when. That record is what makes the rest of the product defensible.
- 1Organization profile — sector, footprint, regulated data
- 2Insurance history — current tower, claims, endorsements in flight
- 3Data & security — controls, retention, access, incident history
- 4Agents — one register entry per agent, with autonomy and audience
- 5Human oversight — review, guardrails, drift, disclosure
- 6Attestation — a named person signs what has been stated
The coverage gap map
See exactly which agent breaks which clause.
Agents across the top, carrier requirements down the side, a verdict in every cell. Open a cell and you get the requirement text, the evidence the carrier expects, the inputs that drove the verdict and the clause reference. Structural exclusions that apply to every insured are banded separately, so you can tell what you can act on from what is simply how the market is written.
Illustrative example · coverage gap map
index 66 · grade D
| Requirement | Patient Intake | Claims Triage | Provider Copilot |
|---|---|---|---|
| A written AI usage policy is in force across the organizationCarrier A · Condition 1 | Coveredsatisfied | Coveredsatisfied | Coveredsatisfied |
| Vendor safety guardrails must remain unmodifiedCarrier A · Condition 2 | Coveredsatisfied | ExcludedCondition 2 | Coveredsatisfied |
| Privacy notices reviewed by counsel where protected data is processedCarrier A · Condition 3 | Coveredsatisfied | InsufficientCondition 3 | InsufficientCondition 3 |
| Drift detection with a scheduled model refreshCarrier B · Monitoring duty | Coveredsatisfied | ConditionalMonitoring duty | ConditionalMonitoring duty |
| Role of the model in the decision path is disclosed at underwritingCarrier B · Disclosure duty | Coveredsatisfied | ConditionalDisclosure duty | Coveredsatisfied |
| Human review stands between output and irreversible actionCarrier C · Scheduled condition | Coveredsatisfied | ExcludedScheduled condition | ConditionalScheduled condition |
Inside a verdict
The requirement, and the rule that fired.
Every non-covered result opens onto the requirement text, the rule that fired, the inputs you attested to, and the cure window the carrier allows. Nothing is paraphrased into vagueness, and nothing is decided by a model.
Illustrative example · verdict detail
condition precedentVendor safety guardrails must remain unmodified for the duration of the policy period.
- Rule fired
- Guardrails modified = true
- Your answer
- Prompt filters relaxed for throughput
- Cure window
- 30 days · attestation
A condition precedent is not a preference. Until the guardrails are restored and attested, an agent like this sits outside the grant of coverage — and inside the product, the file shows the requirement that decided it.
The underwriting packet
Hand your broker something they can submit.
One export, print-ready, laid out like a policy document rather than a dashboard screenshot. It is the difference between a submission that gets worked and a submission that gets questions.
- Cover sheet with the organization and assessment date
- Executive summary of the portfolio position
- Coverage gap map across every agent and requirement
- Agent register with autonomy, audience and data profile
- Findings with clause references and evidence status
- Signed attestation
The cure clock
Every gap gets an owner and a date.
A finding is only useful if someone closes it. Each open gap becomes a cure with the evidence required, the person accountable, and a deadline set against your renewal. Overdue items surface first; items falling due inside thirty days sit right behind them.
Re-run the assessment when the evidence exists and the verdict moves on its own. No one marks their own homework — the engine re-evaluates from the stored inputs.
Continuous compliance
Insurability is a moving target. We watch it move.
Continuous compliance monitoring is scheduled re-attestation, the cure clock, the renewal radar, requirement-change alerts, the evidence vault and unlimited underwriting packets. We verify and evidence that your monitoring obligations are being met, in the form an underwriter accepts. We are not an agent observability tool — we connect to the one you already run.
Because requirements are versioned, you can always see what changed, when, and what it did to your position.
How the verdict engine works
Verdicts are computed. They are never generated.
No model decides whether you are insurable. Each requirement is stored with its clause reference, the evidence it expects and a severity, and a deterministic rule evaluates your attested inputs against it. The same inputs always produce the same verdict, and every non-covered result displays the clause that caused it alongside the specific inputs that triggered it.
The insurability index is the share of the applicant’s own requirements that are covered, with the numerator and denominator shown so the result can be audited. Structural exclusions and carrier terms remain visible, but never enter the score — they are part of the policy, not the applicant’s readiness.
AI does the work around the verdict, never the verdict itself: structuring your plain description of an agent into fields you confirm, summarizing computed results for an executive audience, answering questions strictly from the requirement text on screen, and drafting remediation guidance. Every AI-assisted surface is marked in gold so you always know what was computed and what was generated.
Connecting evidence
Connect the observability you already run.
RiskBorne builds on OpenTelemetry, the open standard the industry has converged on, so evidence flows from the tools you already operate — Datadog, Langfuse, Arize, Braintrust, Weights & Biases, Honeycomb, or whatever ships next — without a custom integration on either side. For anything that does not speak OTLP, push three signals to a signed webhook or expose one read-only endpoint we poll hourly.
We read monitoring metadata only: whether monitoring exists, when it last ran, error counts, and model or version changes. Never prompts, completions or customer data, and we store aggregates rather than spans.
Webhook and heartbeat available today · OpenTelemetry coming soon
Coming soon
On the roadmap, not yet available.
We list these because customers ask where the evidence gets stronger. None of it ships today, and nothing in your current position depends on it.
Bring your own wording
PlannedUpload a revised policy form and we encode it into your instance: your team reviews every extracted requirement, publication supersedes the prior version, and historical assessments stay reproducible.
Applicant-facing white-label
PlannedSend an assessment invitation to an applicant under your own brand and domain, with RiskBorne as the engine underneath.
Carrier API
PlannedPull structured risk data directly into your own underwriting workflow, or embed intake in your own funnel.
Agency management system integration, starting with HawkSoft
PlannedReach the retail agencies where they already work, so a submission can originate inside their system of record.
Agent registration
PlannedAgents register directly and emit lifecycle events — deployed, model changed, guardrail modified, incident — so the record updates itself between assessments.
Client MCP ingestion
PlannedWhere an applicant exposes an MCP server, read their agent inventory directly instead of asking them to complete a form.
See where your agents stand.
Request access and get a per-carrier insurability verdict with the exact gaps blocking coverage.