Skip to main content

Underwriting & Pricing — FDE Playbook (overview)

A Forward-Deployed-Engineer field guide to ktayl's underwriting & pricing domain — how the business works, where the pain is, and where software + AI deliver measurable value. This page is the map; the full detail lives with the code in the product repo.

📖 Full playbook → ktayl-underwriting/docs/fde-underwriting-playbook.md

Domain: Underwriting workbench (#2) + Pricing (#3) in the EA Blueprint · Board: #12 · Initiative: Insurance LOB

:::note Two-layer model This is the ktayl-solution insurance IS (business context) — not the certification. The RNCP deliverable is Retrieva, a separate product. Underwriting appears nowhere in Retrieva. :::

What the playbook covers​

ktayl is a commercial-lines / large-risk IARD insurer, so underwriting means technical underwriting of complex business risks — judgement-heavy, document-heavy, referral-heavy. The full playbook walks the domain end-to-end:

  • Actors — broker, UW assistant, underwriter, technical/senior UW, manager, actuary, cat modeller, committee, compliance (primary user = the underwriter).
  • Process — submission → triage → appetite/eligibility → risk assessment → pricing → aggregate/cat check → quote → referral/committee → negotiation → bind → handoff to Policy Admin (+ renewal and endorsement variants).
  • Data objects — submission, risk/insured, exposure, guideline, rate table, quote, referral, decision, aggregate, authority matrix.
  • As-is vs target systems — the pain (unstructured submissions, no single file, Excel rating) mapped to the epic backlog UW-01 … UW-05.
  • Pain points → manual tasks — ranked by leverage.
  • AI / automation opportunities — traced through the FDE chain, split into build-now (thin AI inside the workbench: submission extraction, guideline check, renewal pre-fill, referral routing) vs the parked AI Ops Copilot capstone; grounded doc-Q&A flagged as Open WebUI, not code.
  • KPIs — quote turnaround, hit ratio, referral/straight-through rate, rate adequacy, aggregate utilisation — the "measurable outcome" every automation must move.
  • A recommended thin slice — one LOB, one broker path, that actually binds.

How it connects​

  • Build order — business tools first, AI/automation last. The workbench/guidelines/pricing come before the AI Ops Copilot (parked capstone). See the EA Blueprint build order.
  • Bind target — ktayl-policy-service (Policy Admin), already live.
  • MDM — client/broker/insured are shared master data; MDM (#20) is parked and emerges as this domain needs shared entities — UW-01 is not blocked on it.