Capability Map & Registry — ktayl IS
The fine-grained functional target: 19 L1 domains → ~150 L2/L3 capabilities, each mapped to our actual system, status, criticality and resilience target. This is the machine-readable companion to the EA Blueprint (the narrative map). The full registry is the spreadsheet
insurance_enterprise_capability_registry.xlsx; this page is its published summary + the why (the four problems) and the order (the roadmap).
:::note Two-layer model This is the ktayl-solution IS (the insurer). Retrieva is a separate product — but its provider→dependency→nth-party graph is a reusable pattern for portfolio accumulation (Problem 3 below). :::
1. The north star — four existential problems (the why behind priority)
A capability earns priority by how much it attacks the industry's structural problems, not by parity with a generic map.
| # | Problem (AXA/Chubb/HDI/Allianz) | What attacks it | Our position |
|---|---|---|---|
| P1 | Insurability crisis — climate & secondary perils blow past actuarial models | Risk Engineering (prevention), CAT, Exposure & Accumulation, portfolio geo-steering | 🔴 our weakest area vs the #1 problem |
| P2 | Margin squeeze — soft market + claims/social inflation | Pricing Engine, portfolio analytics (loss/combined ratio), UW discipline, Data Platform | 🔴 largely gap |
| P3 | Systemic cyber/AI accumulation — one vuln → cascade of claims | portfolio dependency graph, exposure aggregation, cyber LOB, scenario analysis | 🟡 we own the exact pattern (Retrieva graph) |
| P4 | Legacy tech + talent gap | a modern automated core + AI copilots | 🟢 our strength — no legacy debt, mature AI platform |
The shape of the gap: strong bottom-of-stack + AI (answers P4), a unique graph asset (answers P3), and most of the insurance business core missing (which is exactly what attacks P1/P2).
2. The 19 domains (B2B industrial model) ↔ the 12-domain blueprint
The EA Blueprint groups the IS as 12 business domains + 4 transversal layers; the registry uses the finer 19 L1 domains of the B2B-industrial capability model. They reconcile — the 19 just split what the 12 bundled:
- Submission (02) and Pricing & Portfolio (04) are broken out of Underwriting.
- Data (12), Document (14), Integration (16), IT Ops (17), Cyber/IAM (18), AI (19) are the blueprint's transversal layers made explicit L1 domains.
- Compliance (13) and International (06) are first-class L1s.
3. Status — the honest headline (150 capabilities)
| Status | Count | % | Where |
|---|---|---|---|
| 🟢 LIVE | 22 | 15% | Policy, Claims (prod), IAM, AI platform, Data Platform (Metabase BI), parts of Finance |
| 🟡 PARTIAL | 34 | 23% | Underwriting (live on dev), ERPNext finance, compliance/DORA, integration primitives, DMS storage |
| 🟡 SCAFFOLD | 40 | 27% | Risk-Eng, Reinsurance, International, ITSM |
| 🔴 GAP | 54 | 36% | Submission, CRM/MDM, Pricing, Billing (insurance), Channels |
~15% of the insurer's capabilities are live. The GAPs cluster exactly where P1/P2 bite (pricing, portfolio, prevention). Criticality mix: 31 Critical · 80 High · 39 Medium.
Updated 2026-09-30 (verified against the running cluster): since the last registry count, Claims promoted to prod (GlobalCore strangler shipped), the Data Platform (Metabase BI, Slice 1) came live, and Underwriting is live on dev — so the live/partial mix slightly understates reality; the exact counts await a registry re-count. Stage-by-stage live-vs-planned is in §3.1 below.
Per-domain status is in the EA Blueprint gap analysis; per-capability detail (owner, data, API, pain, AI potential, priority score) is in the registry spreadsheet.
3.1 Contract lifecycle coverage — the B2B industrial policy end-to-end
Corporate insurance is a risk-engineering process, not a product sale — the insurer prices the
exposure, and the submission dossier IS that exposure. This walks the full industrial-policy lifecycle
(courtier → submission → underwriting → quote → bind → policy → billing → vie du contrat → renouvellement)
and marks what actually runs. 🟢 prod · 🔵 dev only · 🟡 partial · ⚪ planned (verified 2026-09-30).
| Lifecycle stage | System / domain | Board | Status |
|---|---|---|---|
| Courtier / relationship | Distribution & CRM | #13 | ⚪ planned (no CRM) |
| Submission / appel d'offres (dossier d'exposition) | Submission Hub | #27 | ⚪ not built — board #27 opened 2026-09-30; build not started (the #1 front-door gap) |
| Underwriting — analyse · rating · capacité · T&C | ktayl-underwriting | #12 | 🔵 live on dev (intake→appetite→rating→quote→bind) |
| Pricing (dedicated) + portfolio analytics | Pricing & Portfolio | #04/#12 | ⚪ planned |
| Capacité (traité / facultative) | Reinsurance & Captive | #22 | ⚪ scaffold |
| Quote | ktayl-underwriting | #12 | 🔵 live on dev |
| Négociation courtier ↔ assureur | Broker portal | #13/#15 | ⚪ planned |
| Binding (→ écrit la police) | ktayl-underwriting → PAS | #12 | 🔵 live on dev |
| Police / Policy | ktayl-policy-service | #6 | 🟢 live prod |
| Facturation de la prime (échéances · IPT · commissions) | Insurance Billing | #14 | ⚪ planned — ERPNext is general finance, not insurance billing |
| Vie — avenants / endorsements | ktayl-policy-service | #6 | 🟡 live prod (depth partial) |
| Vie — certificates | Policy / International | #6/#23 | ⚪ planned |
| Vie — Risk Engineering (continu) | Risk Engineering | #21 | 🟡 scaffold |
| Vie — sinistres / Claims | ktayl-claims (GlobalCore ACL) | #11 | 🟢 live prod |
| Renouvellement annuel | UW / Policy renewal | #12/#6 | ⚪ planned |
Verdict: no single tool spans the lifecycle end-to-end. It is live in the middle — Underwriting (dev) → Policy (prod) → Claims (prod) — while the front door (Submission), the financial tail
(insurance Billing, Renewal) and the prevention layer (Risk Engineering, Exposure/CAT/Accumulation)
are gaps. The highest-leverage next build is the Submission Hub: it is the entry point of the whole
corporate flow and feeds the already-live Underwriting workbench — and it is exactly the "prices-the-
exposure" capability (P1/P2). See the roadmap (Stage C).
4. Resilience by design — DORA RTO/RPO tiers (business-capability layer)
DORA is a transversal layer over every business capability, distinct from Retrieva's third-party-ICT slice. Each capability carries a criticality + RTO/RPO target (in the registry):
| Tier | Domains | RTO | RPO | DORA |
|---|---|---|---|---|
| Critical | Policy · Claims · Billing/Premium · Cyber/IAM | 2–4 h | 5–15 min | CIF |
| High | Underwriting · Submission · Pricing · Reinsurance · Finance/Actuarial · Data · Compliance · Risk Eng · International · Integration | 4–8 h | 15–60 min | Important |
| Medium | Distribution · Documents · Channels · IT Ops · AI | 8–24 h | 1–4 h | Supporting |
These are targets, not yet verified DR capabilities for the unbuilt domains — a capability's DORA row is only "done" once backup + DR are tested (the DoD in the blueprint). Retrieva covers the ICT-third-party obligation; this table covers the business capabilities.
5. The order — problem-prioritised roadmap
We are past the tech foundation and the AI platform already, so the backlog re-sequences around P1–P4:
- Stage A — foundation registry (this work): capability registry populated · System of Record · Canonical Data Model · DORA/RTO-RPO per capability. (done)
- Stage B — unlock P1/P2/P3 analytics: Data Platform (keystone) + MDM (Customer/Entity/Broker).
- Stage C — operational core + AI showcase (P2 + P4): Submission Hub → Underwriting Workbench + Pricing Engine (UW Copilot lands here).
- Stage D — the P1/P3 differentiators: Risk Engineering (prevention → P1) + Exposure/Accumulation/CAT (reuse the Retrieva graph → P3).
- Stage E — the rest of the core + channels: Insurance Billing, International/DIC-DIL, Reinsurance, Actuarial; then Broker + Client portals. (Claims — via the GlobalCore legacy strangler — has since shipped and is LIVE in prod.)
This refines — does not replace — the blueprint's business-tools-first, AI-last build order. The difference is the problem lens: within "business tools first," build the ones that move P1/P2/P3 first (Data + prevention + accumulation), not simply the operational order.
6. Sources
- Registry (authoritative, fine-grained):
insurance_enterprise_capability_registry.xlsx— 150 capabilities × 28 fields (owner, data, API, criticality, DORA, RTO/RPO, pain, automation/AI potential, priority score, status). - Narrative map: EA Blueprint · Architecture at a Glance.
- Foundations: System of Record · Canonical Data Model.
- Detail: Business Applications Catalog · Regulatory Operating Model.