Insurance · Pilot
Run WikiSure against your own wordings, SOPs and risk memos.
A 4–6 week structured pilot with a single line of business. We ingest your real source documents, surface semantic drift, run governance workflow with your stewards, and hand back a governed registry plus a Semantic Stability Score baseline.
The problem
Insurance runs on language. The same term — Coverage, Claim, Risk Exposure, Reinsurance Recoverable — carries different meanings across underwriting, claims, legal, risk and finance. Semantic drift mis-prices coverage, breaks Solvency II / IFRS 17 reporting consistency, and turns AI agents into compliance liabilities.
Sample findings (anonymised, prior pilot)
- 5 conflicting definitions of Coverage across 3 departments.
- Claims SOP and Policy Wording disagreed on Notification trigger.
- Risk Appetite memo introduced an undefined Material Exposure term.
- Reinsurance schedule used Recoverable inconsistently with Finance.
Governance workflow
- Ingest source documents into a tenant-scoped registry.
- Extract concepts; cluster competing definitions; classify drift severity.
- Route conflicts to nominated stewards (Underwriting, Claims, Legal, Risk, Finance).
- Capture approved canonical definitions with owner, version, effective date.
- Publish governance decisions to the audit log; export CSV for committee review.
Learning workflow
- Every subsequent scan reuses approved canonicals; drift only appears for new or changed wording.
- Governance history accumulates per concept — first seen, decisions, occurrences, last drift.
- Semantic Stability Score tracks improvement over the pilot quarter.
- AI agents and downstream systems consume the registry via the Resolve API.
Pilot scope
In scope
- One line of business (e.g. Property, Liability, Motor).
- 10–40 source documents (wordings, SOPs, memos, schedules).
- 3–5 nominated stewards across business and risk.
- Weekly review cadence with WikiSure governance lead.
- Tenant isolation, EU hosting, signed DPA.
Out of scope (pilot)
- Production integration with core policy / claims systems.
- Custom domain models beyond the agreed line of business.
- Full SOC 2 / ISO 27001 evidence pack — see Compliance evidence.
Expected outcomes
- Governed registry of canonical definitions for the chosen line of business.
- Quantified semantic drift baseline + Semantic Stability Score trajectory.
- Auditable governance decisions, exportable to your steering committee pack.
- Resolve API endpoint your downstream systems and AI agents can query.
Request a pilot
Send us a short note describing the line of business and steward team. We respond within 2 business days with a scoping call slot and the standard pilot SOW.