Steps 8-14 of the claims pipeline - recorded statement through reserve approval - are where Tractable photo damage AI (with the 70-75% digital-completion rate Admiral Seguros reported in 2026 deployment), CCC Intelligent Solutions auto estimating (with the April 2026 report of 23.1% total-loss frequency record across the 35,000+ repair facilities and 350+ insurance companies in the CCC network), EagleView aerial roof reports, and Snapsheet's ClaimXperience virtual inspection do their work. The artifacts: a structured recorded-statement summary at Step 8; an investigation plan at Step 9 with specific factual questions and source-document requests; a Tractable damage estimate on the Atlanta auto claim at $14,200 within 30 minutes of photo submission; an EagleView pre-loss / post-loss roof comparison on a property hail claim where the field adjuster's photo shows architectural shingles but EagleView estimated three-tab - the $4,200 RCV delta that materially affects the reserve; a reserve recommendation with severity-model output integrating comparable-claim references; and a supervisor reserve approval against the carrier's authority schedule. The worked scenario this lesson uses is the Iowa hail-storm property claim: a 32-year-old residential property, multiple roof penetrations from baseball-sized hail, EagleView aerial called three-tab asphalt shingle, the on-site adjuster's photos show architectural laminate shingle, and the model interpretation matters because three-tab RCV is $9,800 and architectural is $14,000 - and reserve adequacy at Step 20 depends on getting it right early. This lesson is the artifact-level build of Steps 8-14 with the Iowa hail-storm scenario worked end-to-end, the Tractable vs. CCC platform comparison on auto, the EagleView reconciliation methodology for roof-type discrepancies, and the monthly reserve-adequacy review at Step 20 that drives severity drift detection on long-tail property and BI claims.

Step 10 - Tractable Photo Damage AI on the Atlanta Auto Claim

The Atlanta claim (Lesson 7) continues. Step 10 fires when the insured submits vehicle photos via the Hi Marley conversational link or the adjuster collects via the ClaimXperience invitation. Tractable processes within 30 minutes from photo submission to estimate output.

Tractable output for Atlanta. Damage-area heatmap on the insured's 2023 vehicle: front-end damage including bumper, hood, grille, left headlight assembly, deployed airbags (driver and passenger), structural deformation indicator on the left frame rail. Repair vs. total-loss probability: repair at 74% confidence vs. total-loss at 26% - the total-loss threshold is typically 70% of ACV; the vehicle ACV is $28,400; the threshold is $19,880; the Tractable estimate is $14,200 (50% of ACV). Parts and labor cost prediction: $9,200 parts, $5,000 labor, including the airbag-system reset and the structural alignment. The 70-75% digital-completion rate Admiral Seguros reported in 2026 reflects the share of claims that move from photo submission through estimate to settlement without requiring a physical inspection; the Atlanta claim falls in that band. SLA: under 30 minutes from photo submission. Atlanta photos received 9:30 a.m. Day 2; Tractable estimate complete 9:55 a.m. Day 2.

Tractable vs. CCC architectural comparison on auto. Tractable: photo-first model trained on millions of vehicle-damage images; strong on aesthetic damage assessment and total-loss probability; integrates with carrier estimating workflows including direct write into the carrier's claims management system. CCC Intelligent Solutions: OEM-repair-guide-grounded estimating with line-item parts and labor hours referenced to specific manufacturer procedures; deeper integration with body-shop networks across the 35,000+ repair facilities in the CCC network; mature ALAE workflows; the April 2026 report of 23.1% total-loss frequency record reflects the network-wide trend that more vehicles are reaching the total-loss threshold faster as parts costs and safety-system complexity increase. Most carriers run both in parallel: Tractable for fast initial assessment within 30 minutes, CCC for line-item estimate validation that the body shop can execute against. The adjuster uses both for triangulation; the mid-2026 industry split is roughly 40% Tractable-primary, 35% CCC-primary, 25% parallel use with cross-validation.

The total-loss decision logic in 2026. With CCC reporting a 23.1% total-loss frequency record, the total-loss decision is increasingly outcome-determinative. The decision integrates the Tractable estimate, the CCC line-item validation, the ACV from Kelley Blue Book / J.D. Power / Manheim auction data, the salvage value from Copart / IAA, the state-specific total-loss formula (some states use a fixed percentage; others use a more complex formula incorporating salvage), and the policyholder's preference (some prefer total-loss with cash settlement; others prefer repair to retain the vehicle). The adjuster's call on borderline cases - like the Atlanta claim at 50% of ACV well below the 70% threshold - is straightforward; the borderline cases at 65-72% require explicit documentation of the decision logic.

Step 11 - EagleView Pre-Loss vs. Post-Loss on the Iowa Hail Claim

The Iowa hail-storm property claim. FNOL on 2026-05-18 reporting hail damage from the 2026-05-17 storm. Loss location Des Moines, IA. A 32-year-old single-family residential property. The insured reports multiple roof penetrations, soft-tissue damage to vinyl siding, broken skylight, and gutter damage on the south elevation.

EagleView retrieval. Pre-loss imagery from the carrier's pre-bind EagleView capture (2025-09-18, 8 months before the loss). Post-loss imagery captured 2026-05-21 (4 days after the loss) by an EagleView contractor flight scheduled in response to the storm event. AI comparison computes: roof area 2,180 sqft, slope 8/12, condition pre-loss good (granule coverage normal for age, no visible prior damage, no obvious deferred maintenance), condition post-loss showing multiple impact damage points (estimated 28 hail strikes per 100 sqft, severity moderate-to-severe per the impact-mark sizing classifier), some shingle displacement on the south elevation, broken skylight visible, gutter damage at the south fascia. EagleView's roof-material classifier identifies: three-tab asphalt shingle at 78% confidence based on the pre-loss imagery resolution.

The reconciliation discrepancy. The on-site field adjuster's photo (taken Day 6 after the EagleView post-loss flight) shows a close-up of the damaged shingles. The photos clearly show architectural laminate shingle - thicker profile, varied tab dimensions, multi-color granule pattern characteristic of laminate construction. The field adjuster correctly classifies the material as architectural. EagleView's pre-loss classification (three-tab at 78% confidence) is wrong; the aerial imagery confused the tab pattern at altitude, particularly because the architectural shingle on this particular roof has a tab geometry that visually resembles three-tab from the EagleView flight altitude.

The $4,200 RCV delta. Three-tab asphalt shingle RCV per sqft: $4.50. Architectural laminate RCV per sqft: $6.40. 2,180 sqft × $1.90 delta = $4,142, plus the underlayment, drip edge, and ridge cap upgrade required for laminate ($60-$80), totaling roughly $4,200 RCV difference. A material reserve impact. The adjuster updates the Tractable estimating workflow and the Xactimate file with the corrected roof material; the estimate increases from $9,800 (three-tab) to $14,000 (architectural).

The EagleView Reconciliation Methodology

EagleView's roof-material classification at 78% confidence is meaningful but not definitive. The reconciliation methodology surfaces discrepancies and resolves them with field documentation.

The four-source triangulation. (1) EagleView aerial roof-material classification with confidence score and the specific feature weights that drove the classification. (2) Pre-loss insured-provided photos (if available from the binding inspection or the homeowner's records). (3) Field-adjuster post-loss photos at close-up resolution. (4) Roof-installer records (if accessible through the homeowner or the installation contractor's records). When all four agree, the classification is settled. When discrepancies surface, the field-adjuster's photos are authoritative for the current condition; the pre-loss photos and the installer records are authoritative for the original installation; EagleView's confidence drives the default but yields to direct evidence when direct evidence contradicts.

The Iowa Day-6 reconciliation. Step 11 EagleView output recorded Day 4 (post-loss flight). Step 9 investigation plan ordered the field-adjuster inspection Day 5. The field adjuster on-site Day 6; photos uploaded same day; the adjuster flags the discrepancy in the claim file with a specific note about the architectural-vs.-three-tab call. Reconciliation analysis Day 7: the field adjuster's architectural classification wins; the reserve updates upward $4,200.

Why the reconciliation matters for downstream §4.4. The reserve adequacy at Step 20 monthly review tracks severity-model accuracy. Material misclassifications produce systematic underestimation if not caught. The reconciliation discipline at Step 11 surfaces classification errors and corrects them at the individual-claim level; the aggregate effect on book-level severity accuracy is meaningful. The 2026 claims-actuary team that uses the reconciliation log as a feature for model retraining sees a 4-7 point improvement in property-severity model accuracy within 12 months.

Step 13 - Reserve Recommendation with Severity Model

The reserve recommendation produces an initial indemnity and ALAE reserve based on the claim characteristics, the comparable-claim outcomes, and the severity-prediction model output. The reserve is documented in the file note with the §4 reason chain attached.

Atlanta auto BI reserve recommendation. Severity model inputs: claim type (auto BI/PD), injury severity indicator (Emory hospitalization plus airbag deployment), liability profile (other-driver-fault per claimant assertion, currently unverified pending police report), claimant employment status (TBD pending recorded statement), comparable-claim references (Atlanta auto BI claims with hospitalization plus airbag in the last 24 months, n=47, average settlement $38,400, median $34,200, 90th percentile $58,400). Severity model output: PD $14,200 (Tractable estimate), BI indemnity range $18,000-$42,000 with $25,000 mid-point recommendation, ALAE $5,000 reflecting expected investigation plus recorded statement plus medical-records review plus potential SIU evaluation. Total reserve recommendation: $44,200.

Iowa hail property reserve recommendation. The property loss reserve includes: dwelling repair $14,000 (the corrected architectural shingle estimate after the EagleView reconciliation), additional living expense (ALE) $2,800 (estimated 4-6 weeks displacement during the roof replacement), personal property limited (no contents damage from the hail event), ALAE $1,200 (estimated for the inspection, estimating, and adjuster cycle time). Total reserve recommendation: $18,000. The §4 reason chain documents the severity model output, the EagleView reconciliation rationale, and the supervisor authority threshold check.

The §4 reason chain at Step 13. The reserve recommendation captures: severity model output with the feature values, comparable-claim references with the cohort definition and the percentile placement, ALAE benchmark with the jurisdiction-specific cell, model version and features, supervisor authority threshold check, file note attached. For Atlanta: supervisor approval required (over $40K total reserve). For Iowa: under-authority reserve (under the $20K threshold for the senior adjuster); auto-approves with batch supervisor review.

Step 14 - Supervisor Reserve Approval

Reserves above adjuster authority require supervisor approval. Authority schedules vary by line but typically: fast-track adjuster $25K, senior adjuster $75K, claims supervisor $250K, claims manager $1M, claims VP higher.

Atlanta supervisor approval. J. Martinez (senior auto BI adjuster) authority $75K. Reserve $44,200 within authority but routes to supervisor review because the BI indemnity component carries the severity-trajectory monitoring (Step 5.3.4 fired at triage in Lesson 7 because of the Emory indicator). The supervisor reviews the severity model output, the comparable references, the expected medical trajectory, and the settlement-range bounds. Approves at the recommended $44,200 with an adjustment trigger if BI severity escalates beyond $35K (at which point the reserve increases to the $65K range). The reason chain captures the rationale, the adjustment trigger, and the next-monitoring review date.

Iowa supervisor batch review. The under-authority reserve approves automatically through the workflow. The daily supervisor batch review samples 10 of the under-authority approvals for quality check. Iowa reserve quality-checked Day 8; passes review with the note "EagleView reconciliation correctly captured; corrected estimate applied; documentation complete."

Step 20 - Monthly Reserve-Adequacy Review

The claims actuary owns the monthly reserve-adequacy review. The review aggregates across the portfolio, tracks reserve development vs. actual outcomes, and surfaces drift signals that may indicate model degradation, book-of-business drift, or external-environment changes (medical-cost inflation, litigation severity escalation, regional reconstruction-cost inflation).

Severity drift detection. Each month, the severity model's predicted ranges compare against the settled claim outcomes in that month. PSI (Population Stability Index) tracks distribution drift on key features. KS-statistic tracks predictive distribution drift. AUC tracks predictive power. Drift on any metric triggers investigation: model drift (the model has degraded), book-of-business drift (the claim mix is changing), inflation drift (medical or litigation or reconstruction costs are escalating beyond the model's trend assumption). Each driver requires different remediation: model drift triggers a retraining cycle; book drift triggers a feature-engineering review; inflation drift triggers a trend-factor recalibration.

The Iowa hail-storm review at month +3. The Iowa claim closed at $14,400 actual settlement vs. $14,000 reserve estimate (after the Step 11 reconciliation). A 2.9% positive variance. The aggregate hail-claim severity for the quarter: 312 claims, average estimate $11,200, average actual settlement $11,900. A 6.3% positive variance - a modest book-level severity miss attributable to the architectural-shingle reconciliation pattern surfacing on multiple Iowa hail claims. The claims actuary flags the pattern; investigation reveals EagleView's three-tab classification systematically erring on Iowa's architectural-shingle prevalence; the recommendation: tune the EagleView classifier for the Iowa-specific shingle market, or adjust the reserve baseline upward 6% for Iowa hail until the classifier is tuned. The claims actuary coordinates with EagleView's customer-success team for the classifier retraining.

The Atlanta auto BI review at month +6. The Atlanta claim still open. Reserve at $44,200; partial paid (PD settled at $14,400; BI investigation continuing). BI trajectory monitoring: claimant treatment ongoing, MRI showed concussion plus bruised ribs as expected at intake, no escalation indicators (no surgical referral, no extended rehabilitation, no permanent-impairment work-up). Reserve adequate at this stage; review continues monthly.

The reserve-adequacy KPI dashboard. The claims actuary maintains a portfolio dashboard with the PSI / KS / AUC drift metrics, the severity-variance trend, the loss-development triangle progression, the ALAE-to-loss ratio by jurisdiction, and the inflation-trend indicators (medical CPI, litigation severity, reconstruction cost index). The dashboard feeds the chief actuary's quarterly reserve-adequacy memo to the board and the chief financial officer; the memo informs the next ORSA narrative and the AM Best annual interaction.

The Handoff to Step 15 Fraud Screen and Step 17 Medical Records

Steps 8-14 close with structured reserves, completed photo and aerial assessment, and supervisor approval. Concurrent: Step 15 fraud screen (Shift Technology + ISO ClaimSearch + NICB) - covered in Lesson 9. Step 17 medical-records review when records arrive - covered in Lesson 10 with the HIPAA §164.502 minimum-necessary discipline.

The 22-step pipeline branches: claims-side parallel processing of investigation + estimating + fraud screening + medical review accelerates resolution. Without parallel processing, the pipeline would serialize at 80-120 days minimum on complex claims; with parallel processing, complex claims resolve in 60-90 days when liability and damages are clear, and the §4.4 documentation discipline at every step preserves the audit trail for any subsequent DOI exam, FCRA inquiry, or bad-faith defense. The Five Sigma deployment metrics that Starr Insurance reported in 2025 (40% faster resolution, 35% claims-handling cost reduction, 60% faster general-queue email, 70% accuracy) reflect the parallel-processing compression at scale; the 2026 industry trajectory continues toward parallel processing as the default architecture.

Key Takeaways

• Tractable photo damage AI produces auto damage assessment in under 30 minutes. Atlanta vehicle: $14,200 estimate, 74% repair / 26% total-loss probability, $9,200 parts + $5,000 labor. Vehicle ACV $28,400; total-loss threshold typically 70% of ACV = $19,880; the estimate at 50% of ACV remains in repair range. Admiral Seguros reported 70-75% digital-completion rate in 2026 deployment, reflecting claims that move from photo through estimate to settlement without physical inspection.
• CCC vs. Tractable architectural comparison on auto. Tractable: photo-first, strong aesthetic + total-loss probability, fast initial assessment. CCC Intelligent Solutions: OEM-repair-guide-grounded line-item estimating, deeper body-shop integration with 35,000+ repair facilities and 350+ insurance companies in the network, mature ALAE workflows, the April 2026 report of 23.1% total-loss frequency record. 2026 split: 40% Tractable-primary, 35% CCC-primary, 25% parallel with cross-validation.
• EagleView pre-loss vs. post-loss roof comparison drives the Iowa hail-storm reserve. 2,180 sqft roof, 8/12 slope, EagleView classified three-tab at 78% confidence; the field adjuster's Day 6 close-up photos show architectural laminate. $4,200 RCV delta from the corrected classification; estimate jumps from $9,800 to $14,000. The classification difference is material to reserve adequacy.
• The four-source roof-material triangulation: EagleView aerial + pre-loss insured photos + field adjuster post-loss photos + roof-installer records. Discrepancy resolution: field adjuster photos authoritative for current condition; pre-loss photos and installer records authoritative for original installation; EagleView's 78% confidence yields to direct evidence. The 2026 claims-actuary team that uses the reconciliation log as a feature for model retraining sees a 4-7 point improvement in property-severity model accuracy within 12 months.
• Step 13 reserve recommendation with severity-model output. Atlanta auto BI: PD $14,200 + BI indemnity $25,000 mid-point + ALAE $5,000 = $44,200 total. Iowa hail: dwelling $14,000 + ALE $2,800 + ALAE $1,200 = $18,000 total. Each carries the §4 reason chain with severity model output, comparable-claim references, ALAE benchmark, supervisor authority check, file note attached.
• Supervisor reserve approval at Step 14 against authority schedules. Atlanta $44,200 within senior adjuster $75K authority but routes to supervisor because the BI severity-trajectory monitoring fired at triage; the reason chain documents the adjustment trigger if BI escalates beyond $35K. Iowa under-authority auto-approves with daily batch sampling for quality check.
• Monthly reserve-adequacy review at Step 20 surfaces drift signals. Iowa quarter: 312 hail claims with 6.3% positive variance attributable to EagleView three-tab vs. architectural systematic miss. The claims actuary tunes the classifier for the Iowa-specific market or adjusts the reserve baseline. PSI / KS / AUC drift metrics distinguish model drift vs. book drift vs. inflation drift. The dashboard feeds the chief actuary's quarterly reserve-adequacy memo to the board and the chief financial officer.
• Parallel processing of investigation + estimating + fraud + medical review at Steps 8-17 accelerates complex-claim resolution from 80-120 days to 60-90 days when liability and damages are clear. Without parallel processing, the pipeline serializes; with parallel processing, claims resolve faster while preserving §4.4 documentation discipline at every step. The Five Sigma metrics at Starr Insurance (40% faster resolution, 35% cost reduction) reflect parallel-processing compression at scale.