Step 3 closed with a structured JSON payload representing the Acme Warehousing submission: 47 locations across the Dallas-Fort Worth Metroplex, $182M TIV, $24M BI exposure, six losses over five years ($14K-$142K severity), COPE narrative summarized for the three largest locations, EagleView roof-age fills for 15 of 18 SOV gaps, three locations pending producer follow-up. Steps 5, 6, and 7 - enrichment, appetite scoring, triage - consume that payload and produce three artifacts: an enriched submission record with external signals appended, an appetite score with structured JSON reason chain, and a routing decision (auto-quote, UW-assistant, specialist UW, or decline) with §4 reason codes attached. Total elapsed time at this stage of the pipeline: under 24 minutes from intake. The platforms are Cytora (with the Cytora Autopilot agentic addition expanding in 2026) and Convr competing for the orchestration role, Send entering the market with broker-side enrichment, FEMA NRI providing cat-aggregate input at the county and tract level, EagleView providing roof and footprint data, D&B providing financial stability, Moody's Orbis providing ownership and ultimate parent identification, Verisk AIR / Moody's RMS / KCC providing modeled wind/hail/quake PML at the 100/250/500/1000-year return periods, ICEYE SAR providing emerging parametric flood capability. The Tier-1 wind aggregate consumption is at 78% pre-bind; this submission contributes 0.8 points; post-bind 78.8% - within tolerance but flagged for portfolio review. This lesson is the artifact-level build of Steps 5-7 with the worked Acme example: appetite reason codes, knockout reasons, triage routing logic, the structured JSON output that feeds the UW workbench at Step 8 in Federato RiskOps or in-platform within PolicyCenter.

The Step 5 Enrichment Stack

Step 5 enrichment runs in parallel API calls against external data providers. The orchestration platform (Cytora or Convr) manages the parallel calls, handles retries with exponential backoff, normalizes the outputs against the carrier's canonical schema, and writes back to the structured submission record. The 14-minute SLA target requires parallel execution - sequential calls would consume 30-45 minutes and break the SLA on every submission, eroding the carrier's hit-ratio advantage and the broker-side perception of speed that drives broker share.

FEMA NRI (National Risk Index). County-level natural hazard risk scoring across 18 perils (wind, hail, tornado, hurricane, flood, earthquake, wildfire, drought, lightning, winter storm, coastal flooding, ice storm, landslide, riverine flooding, snowstorm, tsunami, volcanic activity, heat wave) with the risk index decomposing into expected annual loss, social vulnerability, and community resilience. For Acme's 47 DFW locations, FEMA NRI returns county-level risk index for Dallas, Tarrant, Collin, Denton counties: 5 locations at very high (90+), 18 at relatively moderate (50-75), 22 at low (0-50), 2 at very low. The data drives the cat-aggregate analysis at Step 13 and feeds Convr's Risk 360 scorecard if Convr is the orchestration platform.

EagleView. Aerial imagery integration delivers roof age, building footprint, roof slope and area and condition, and the pre-loss imagery baseline that supports claims-side workflow at Step 11 of the FNOL pipeline (covered in Lesson 8). For Acme, EagleView fills 15 of 18 SOV roof-age gaps; 3 rural locations without coverage flag producer follow-up; the roof-material classifier returns confidence scores for each location that feed the cat-PML model.

D&B (Dun & Bradstreet). Financial stability rating, PAYDEX score, ownership structure, years in business, employee count verification, NAICS validation, prior bankruptcy filings, and the D&B small-business risk score where applicable. For Acme, D&B confirms $47M annual revenue, 280 employees, NAICS 493110 (general warehousing and storage) verified, no bankruptcy filings, A2 credit rating, PAYDEX 78 (paying obligations within agreed terms).

Moody's Orbis. Ownership chains, ultimate parent identification, related-entity mapping, sanctions screening against OFAC and the EU consolidated list. For Acme, Orbis confirms the ultimate parent is Acme Holdings LLC (Texas), no foreign ownership, no related-entity coverage gaps, no sanctions hits. Confirms the named-insured legal name matches the PAS expectation exactly including the LLC corporate form.

Verisk AIR / Moody's RMS / KCC (Karen Clark & Co). Three cat-model providers with comparable but distinct methodologies. Most carriers run Verisk AIR as primary with Moody's RMS as comparison and KCC as a third validation source. Each returns modeled PML at 100/250/500/1000-year return periods by peril. For Acme: 250-year wind PML $2.18M aggregate across 47 locations per Verisk AIR; Moody's RMS comes in at $2.05M (within tolerance of Verisk); KCC at $2.31M (slightly higher, reflecting KCC's tornado treatment); hail PML $1.7M; tornado PML $880K. The three-model spread is informative; the carrier uses the median plus the highest-PML model as the basis for treaty cession decisions.

OSHA inspection history. For commercial accounts with employee exposure, OSHA inspection records flag prior citations, severity, abatement status, and any open enforcement actions. For Acme warehousing, no current OSHA actions; 2 historical inspections both closed without citation; no open enforcement actions across any of the 47 locations.

ISO ClaimSearch claim history. Cross-carrier loss history beyond the producer's loss run. For Acme, ISO returns 3 additional claims not on the producer-provided loss run - 2 small property claims at predecessor entities, 1 GL claim closed without payment. The submission's effective loss history adjusts to 9 total losses with revised frequency (1.8/year) and revised severity (range broadened to $4K-$142K).

ICEYE SAR parametric flood capability. Emerging in 2026 as a parametric flood signal source. ICEYE's synthetic-aperture-radar satellite constellation provides flood-extent imagery within hours of a major event. For underwriting, ICEYE's historical flood-extent data informs the location-specific flood-exposure assessment beyond the FEMA flood-zone designation. For Acme's Dallas-area locations, ICEYE confirms low historical flood exposure consistent with the FEMA non-SFHA designation across 44 of 47 locations.

The Cytora vs. Convr Orchestration Comparison

Cytora and Convr both deliver enrichment-orchestration plus appetite-scoring plus triage-routing in integrated platforms. The architectural and operational differences matter for vendor selection.

Cytora. UK-founded; expanded the U.S. market 2023-2025; introduced Cytora Autopilot agentic capability in 2026 that augments the underlying orchestration with autonomous submission triage, missing-info follow-up, and quote-and-bind path execution within configured authority. Architecture: rules engine plus ML scoring with carrier-configurable appetite guides. Strength: appetite-guide ingestion accuracy (the carrier's v2026-04 PDF translates cleanly to structured rules including the exception line items), structured JSON reason-chain output (the §4 documentation artifact is publication-ready), pre-built connectors for Federato RiskOps and Guidewire PolicyCenter. Weakness: U.S. cat-model integration is younger than Convr's; the FEMA NRI integration is excellent, Verisk AIR integration is solid, Moody's RMS integration is improving through 2026. Pricing 2026: subscription-based, $400K-$1.5M annually for a mid-size carrier depending on premium volume and module scope.

Convr. U.S.-founded; longer U.S. cat-model integration history. Architecture: ML-first with rules overlay; appetite-guide structured representation requires more carrier-specific configuration than Cytora. Strength: U.S. cat-model integration depth (Verisk AIR + Moody's RMS + KCC all native; Convr's Risk 360 scorecard with FEMA NRI is the deepest in the market), broader external data partnerships, broker-side workflow integration through Convr's expanding broker connector library. Weakness: the appetite reason-chain output is less publication-ready than Cytora's; §4 documentation requires post-processing for the carrier's audit packet format. Pricing 2026: subscription-based, $350K-$1.4M annually for a mid-size carrier.

The carrier's vendor selection. Cytora wins where appetite-guide complexity is high (specialty carriers, MGAs with delegated authority constraints, brokers needing reason-chain output, carriers preparing for AISET Exhibit B/C audits). Convr wins where U.S. cat-model integration is decisive (Tier-1 wind specialty, earthquake specialty, wildfire specialty in California with the 2025-2026 market dislocation). Most mid-size specialty carriers in 2026 are running one or the other; the platforms overlap rather than complement, so a hybrid deployment is rarely cost-justified. The carrier's selection often follows the chief underwriting officer's prior platform experience plus the chief actuary's cat-modeling preference plus the data team's integration capacity.

Federato RiskOps as the orchestration layer above either. Federato RiskOps + agentic underwriting can sit above Cytora or Convr and orchestrate the underwriter workbench, the triage routing, and the §4 reason-chain capture across the entire pipeline. The Federato vs. Cytora-or-Convr choice is not "or" - many 2026 carriers run Federato as the workflow orchestrator with Cytora or Convr as the scoring engine underneath.

The Appetite Scoring Engine and Acme Worked Example

Step 6 appetite scoring takes the enriched submission and scores against the carrier's appetite guide (v2026-04 PDF maintained by the chief underwriting officer). The score is 0-100; bands map to triage routing at Step 7.

Acme appetite scoring. Cytora ingests the appetite guide structured rules. For Acme's 47 locations: NAICS 493110 (General Warehousing and Storage) is in-appetite. TIV $182M is within band (the carrier accepts up to $250M aggregate TIV with no single risk above $30M). Loss frequency 1.8/year (revised after ISO ClaimSearch enrichment) is acceptable. Loss severity range $4K-$142K is within band. Geographic concentration in the DFW Metroplex is acceptable per the appetite guide's geographic-spread rule for warehousing. Three locations sit on the habitational-frame appetite exception (Buildings 14, 27, 41 with frame construction in habitational-adjacent territory).

Score: 78 - in-band with appetite exceptions. Cytora's structured JSON output:

{ submission_id: 'ACME-2026-05', appetite_score: 78, classification: 'in_band_with_exceptions', line_item_references: ['appetite_guide_v2026_04_section_3_warehousing', 'section_4_2_habitational_frame_exception'], in_band_drivers: ['naics_493110_match', 'tiv_under_threshold', 'loss_frequency_under_threshold', 'geographic_appetite_match'], exception_overrides: [{ location_ids: ['B14', 'B27', 'B41'], exception_rule: 'habitational_frame_non_tier1_wind', conditions: 'roof_age_under_15 + sprinkler_present + no_prior_wind_loss' }], cat_aggregate_contribution: { wind_250yr_pml: 2180000, hail_250yr_pml: 1700000, contribution_points: 0.8 }, recommendation: 'specialist_uw_routing', section_4_reason_chain: ['appetite_match_with_three_exceptions', 'no_proxy_detected', 'producer_followup_required_three_locations'] }

Why specialist UW routing. Despite the 78 in-band score, the three appetite exceptions push routing to specialist UW (senior underwriter with §4 reason-chain workflow) rather than UW-assistant. The specialist UW handles the exception analysis, the CUO referral if needed (the appetite-guide exception line item may require chief underwriting officer sign-off depending on the version-specific authority schedule), and the treaty cession decision on the three locations whose TIV approaches the $25M single-risk threshold even though no single location exceeds it.

Appetite Knockout Reasons and Decline Rationale

Not every submission scores in-band. The appetite engine's knockout reasons drive decline decisions; each carries §4 documentation that must survive the post-decline FCRA analysis and any subsequent E&O or fair-conduct review.

Hard knockouts. NAICS not in appetite (the carrier doesn't write the class); TIV exceeds the aggregate threshold; geographic exclusion (the carrier doesn't write in certain states or zones - California wildfire, Florida Tier-1 wind, coastal Texas); prior-carrier non-renewal for cause (confirmed via ISO ClaimSearch carrier-contributed data); loss frequency above threshold (e.g., 4+ losses in 5 years on a property book); loss severity above threshold (e.g., $500K+ single loss); regulatory action against the named insured (DOI complaint history, market-conduct findings, suspension or revocation); known fraud history (NICB pattern reference or prior fraud-based denial).

Soft knockouts (require referral). Multiple appetite exceptions on a single submission; geographic concentration above 60% in a single county or single ZIP cluster; treaty single-risk above $25M; cat-aggregate consumption pushing post-bind above 85% tolerance on any peril; loss-development pattern suggesting unreported claims (the loss run reports closed at a pace inconsistent with the carrier's experience for the class).

Decline letter generation. Hard-knockout submissions route to the decline-letter draft queue. Cytora drafts the decline letter from a template referencing the specific appetite-guide line item; the UW reviews; §4.4 documentation captures the decline rationale with the structured JSON exported as the audit artifact. The decline letter must avoid disparate-impact language (no reference to characteristics that could proxy for protected class) and must reference specific appetite-rule line items rather than subjective judgment. The FCRA §615 pre-notice and adverse-action analysis attaches if the decline relied on consumer-report data; for Acme, FCRA does not apply because the appetite score derived from policy application data plus ISO ClaimSearch plus public-record data, none of which is a consumer report under FCRA §603(d).

The Triage Routing Decision and the Four Paths

Step 7 triage routes to one of four paths based on the appetite score and the exception structure.

Path 1 - Auto-quote eligible. Score 85+ with no exceptions, no treaty or aggregate concerns, no missing-info gaps, no producer-follow-up pending. The submission flows through the compressed sub-pipeline: the pricing model runs, the quote letter generates from template, the UW reviews only the exception flags at Step 14. Elapsed time intake-to-quote: 90 minutes. Roughly 25-35% of clean small-mid commercial submissions qualify. The Cytora Autopilot agentic capability is the natural fit here in 2026; the agentic mode executes the auto-quote path under the configured authority and surfaces only the exceptions to a human UW.

Path 2 - UW-assistant routing. Score 65-85 with limited exceptions or missing-info gaps. Routes to the UW assistant (junior underwriter) with an AI-assisted workbench in Federato RiskOps or in PolicyCenter. AI drafts the missing-info request, the loss-run analysis, the pricing rationale. The junior UW reviews and signs; the specialist UW reviews on referral. Roughly 30-40% of mid-market commercial submissions qualify.

Path 3 - Specialist UW routing. Score below 65, multiple exceptions, treaty implications, large-loss exposure, complex coverage structures. Routes to the senior UW with the full §4 reason-chain workflow. Roughly 25-35% of submissions including Acme. The specialist UW handles the chief-underwriting-officer referral if the appetite-guide exception requires it, the treaty actuary referral for cat-aggregate decisions, and the coverage counsel referral for any manuscript endorsement requests.

Path 4 - Decline. Hard knockout fires. Routes to the decline-letter draft queue. Roughly 5-15% of submissions. The decline letter ships with the §4.4 documentation packet preserved for any subsequent FCRA dispute, DOI complaint, or producer relationship inquiry.

The Tier-1 Wind Aggregation Analysis at Step 13 Preview

While the full treaty-constraint enforcement happens at Step 13, the Step 5-7 stage produces the cat-aggregate input that feeds the Step 13 analysis. For Acme: 250-year wind PML $2.18M aggregate per Verisk AIR (with Moody's RMS at $2.05M and KCC at $2.31M for triangulation); the account contribution is 0.8 points to the carrier's 78% consumed Tier-1 wind cat aggregate; the post-bind projection is 78.8%. Within the 85% tolerance but flagged for portfolio review.

The aggregation analysis at Step 5-7 surfaces the contribution; Step 13 enforces the constraint via the treaty-cession decision (quota-share retention plus cat-XOL aggregate consumption plus facultative single-risk threshold); Step 14 UW judgment integrates the treaty implication into the quote-with-restriction structure (per Lesson 6 of L2). The three steps form an integrated treaty discipline that prevents the carrier from over-consuming the cat aggregate on a single account or accumulating Tier-1 wind exposure beyond the reinsurance capacity layer that the treaty broker placed at the prior 1/1 or 4/1 renewal.

The treaty broker's view. The cat-aggregate consumption tracking matters not only for the immediate quote but for the next treaty renewal. The treaty broker reviewing the carrier's portfolio in October for the 1/1 placement reads the cat-aggregate consumption profile across the year; a carrier that drifts above 85% mid-year and has to buy expensive last-minute cat capacity at the next renewal pays a relationship premium. The Step 5-7 enrichment plus the Step 13 enforcement plus the Step 14 UW judgment together produce the discipline that protects the next renewal's treaty pricing.

The Handoff to the UW Workbench at Step 8

Steps 5-7 close with the appetite score, the structured JSON reason chain, the triage routing decision, and the enriched submission record. The handoff to Step 8 UW workbench populates Federato RiskOps (or the in-platform workbench in PolicyCenter / Duck Creek / Sapiens IDIT) with:

The Acme submission summary (per the Lesson L2 artifact). The per-location COPE rollup. The BI exposure rollup with contingent BI flags from the BI worksheet. The natural-cat aggregation contribution (Verisk AIR, Moody's RMS, KCC PMLs side by side). The treaty constraints (single-risk threshold check on three locations whose TIV approaches $25M). The missing-info status (3 rural locations pending producer follow-up plus Building 27 roof inspection plus Building 41 prior-loss closure documentation). The appetite-match summary with the §4 reason chain. The triage routing decision (specialist UW) with the rationale.

The UW at Step 8 opens the workbench and sees everything from Steps 1-7 in a single screen. The specialist UW reviews the appetite analysis, accepts the routing, queues the missing-info request at Step 9, and begins the pricing-and-treaty analysis at Steps 11-13. The pipeline continues.

Key Takeaways

• Step 5 enrichment runs parallel API calls against FEMA NRI, EagleView, D&B, Moody's Orbis, Verisk AIR / Moody's RMS / KCC, OSHA, ISO ClaimSearch, and emerging sources like ICEYE SAR for parametric flood capability. Sequential calls would consume 30-45 minutes; parallel orchestration meets the 14-minute SLA and protects the carrier's broker-perceived speed advantage.
• Cytora vs. Convr architectural comparison. Cytora: rules engine + ML scoring; appetite-guide accuracy and reason-chain output strong; UK-founded; the 2026 Cytora Autopilot agentic capability fits the auto-quote path. Convr: ML-first; U.S. cat-model integration deeper with Risk 360 scorecard; Convr's broker connector library expanding. Both at $350K-$1.5M annually for mid-size carriers. Selection depends on appetite-guide complexity vs. cat-model centrality. Federato RiskOps can orchestrate above either.
• Acme appetite score: 78 - in-band with three exceptions (Buildings 14, 27, 41 on habitational frame). Cytora structured JSON output captures the appetite-guide line items referenced, the exception conditions, the cat-aggregate contribution, and the §4 reason chain. Routing: specialist UW because the exceptions trigger senior workflow including potential chief underwriting officer referral.
• Hard knockouts include NAICS-not-in-appetite, TIV-over-aggregate, geographic exclusion, prior-non-renewal for cause confirmed via ISO ClaimSearch, loss-frequency-above-threshold, regulatory action, known fraud history per NICB. Soft knockouts include multiple exceptions, geographic concentration above 60% in a single county or ZIP cluster, treaty single-risk above $25M, cat-aggregate post-bind above 85% on any peril, loss-development pattern suggesting unreported claims.
• Triage produces four paths: auto-quote eligible (25-35% of clean submissions; the natural Cytora Autopilot fit), UW-assistant routing (30-40%), specialist UW (25-35%), decline (5-15%). Path determines the sub-pipeline shape and the elapsed time downstream.
• Tier-1 wind aggregation analysis surfaces at Step 5-7 and enforces at Step 13. Acme: $2.18M 250-year wind PML per Verisk AIR with Moody's RMS at $2.05M and KCC at $2.31M for triangulation, 0.8 points contribution, 78.8% post-bind consumption - within 85% tolerance, portfolio-review flag. The treaty broker's view of cumulative consumption matters at the next 1/1 or 4/1 renewal.
• Decline-letter generation uses Cytora template referencing specific appetite-guide line items. §4.4 documentation captures the decline rationale with structured JSON as the audit artifact; the letter avoids disparate-impact language; references specific rules, not subjective judgment. FCRA §615 pre-notice and adverse-action analysis attaches if the decline relied on consumer-report data; for Acme, FCRA does not apply.
• The Step 5-7 handoff to Step 8 UW workbench populates the full submission summary with COPE rollup, BI exposure with contingent BI flags, three-source cat aggregation, treaty single-risk constraint check, missing-info status, appetite analysis, triage decision. The UW sees everything in a single screen in Federato RiskOps or the in-platform workbench and continues the pipeline.