Hesper AI
BlogBuyer Playbook
Buyer PlaybookMay 12, 2026·12 min read·Nitish Badu, COO

The Claims VP playbook for deploying AI investigation: budget, timeline, change management

A Claims VP buying guide for AI investigation: pilot and 12-month budget ranges, phased rollout, stakeholder map, change management, and board outcomes.

$150k - $1M
90-day pilot budget range
1,000 - 5,000 cases, 1-2 LOBs
$3M - $15M
12-month full rollout
scales with flagged-claim volume
25% to 100%
Flagged-claim coverage shift
manual SIU baseline to AI-investigated
14+ days to 2-4 hours
Cycle time per case
manual SIU vs AI investigation

A Claims VP at a Top 50 P&C carrier sits with three numbers on the desk every quarter: the SIU caseload, the cost per investigated claim, and the loss ratio on lines where fraud is concentrated. AI claims investigation moves all three. The harder problem is scoping the buy in a way the CFO will sign, the CIO will integrate, and the SIU director will actually run.

This playbook is for the Claims VP who has decided AI investigation is in the budget cycle and now has to defend the line item. It walks the pre-RFP diagnostic, the pilot and 12-month budget ranges, the phased timeline, the stakeholder map, the change management commitments, and the four metrics that belong on the board report at month 12.

The buy sits inside the wider claims-operations stack covered in our autonomous AI claims investigation pillar, which is the upstream context for the procurement decision below.

The pre-RFP diagnostic: why this is a Claims VP decision

Three signals tell a Claims VP the AI investigation buy belongs in the claims-org budget, not the IT capital plan. SIU capacity is the binding constraint on loss-ratio improvement. Cost per investigated case is the lever the CFO will model. And the antifraud-plan filing under NAIC Model Act #680, adopted in 48 states, is a claims-org responsibility, not an IT one.

The current state at most Top 50 carriers: a manual SIU investigation takes 14+ days per case, an investigator carries 200+ cases, and only about 25% of flagged claims actually get investigated. The other 75% close on adjuster judgment without an investigation file. Meanwhile, total US insurance fraud loss runs at $308 billion annually per the Coalition Against Insurance Fraud, with roughly 10% of P&C claims involving some form of fraud. SIU staffing growth, per the Insurance Information Institute, slowed to 1.4% in 2021-2022 from 2.5% the prior year. The capacity gap is widening, not closing.

A Claims VP also needs to be clear on what AI investigation is not. It is not a detection-vendor swap. FRISS, Shift Technology, and Verisk ISO ClaimSearch keep doing what they do at FNOL. The investigation layer is downstream, covered in detail in our piece on prevention, detection, and investigation as three distinct layers. The new spend sits on the investigation tier, where SIU headcount has historically been the only available lever.

What the Claims VP actually owns

Three things sit on the Claims VP's P&L: loss-ratio impact on lines where fraud is concentrated, cost per investigated claim, and the antifraud-plan filing with the state DOI. AI investigation moves all three. IT, compliance, and finance are reviewers, not owners.

Budget framework: what to ask for and how to model ROI for the CFO

A defensible budget ask has three numbers: a 90-day pilot, a 12-month rollout run-rate, and the coverage-corrected loss-ratio math the CFO will actually scrutinize. Quote the pilot at $150k to $1M for 1,000 to 5,000 cases on one or two lines, the 12-month rollout at $3M to $15M depending on flagged-claim volume, and the ROI in basis points of loss ratio on AI-investigated lines.

Build the CFO model on coverage-corrected unit cost, not headline per-case price. Manual SIU runs at roughly $2,500 per case with about 25% coverage of flagged claims. AI investigation runs at roughly $150 per case with 100% coverage. On a carrier flagging 50,000 claims per year, manual investigation costs about $31M to cover 12,500 cases. AI investigation covers all 50,000 cases for about $7.5M. The fraud caught in the previously uninvestigated 37,500-case tier is the loss-ratio story, not the per-case savings.

Investigation coverage and unit economics

Manual SIU coverage of flagged claims25%
AI investigation coverage of flagged claims100%
Manual cost per case~$2,500
AI cost per case~$150

Two budget pitfalls show up in CFO reviews. First, services markup on integration: the vendor quotes $150 per case, but the SOW lands closer to $300 per case once implementation services, evidence-pack storage, and adjuster training are loaded in. Second, parallel-run costs: most carriers underfund the 60 to 90 days when SIU runs the old workflow next to the AI-assisted one for audit confidence. The full set of line items the CFO will ask about is in our breakdown of hidden integration costs for legacy claims AI. For full ROI scenarios across three carrier sizes, see our ROI case studies for AI claims investigation.

The 12-month timeline: pilot, scale, mature

A defensible rollout has three phases and the Claims VP should commit to phase-gate decisions, not a 12-month lock. Phase 1 is a 90-day pilot on one or two lines, typically auto and workers compensation because referral volume is highest and red flags are well-defined. Phase 2 scales to homeowners, commercial property, and bodily injury once the pilot signal on cycle time and false-positive resolution is clean. Phase 3 retires manual triage on the covered lines and integrates cost-per-case into the monthly claims operating review.

PhaseMonthsScopeBudget rangeOutcomes to measure
Pilot1 - 31,000 - 5,000 cases; 1-2 LOBs (auto, workers comp)$150k - $1MCycle time, false-positive resolution, integration latency
Scale4 - 920,000 - 50,000 cases; add homeowners, commercial, BI$1.5M - $7.5MCoverage %, cost per case, SIU role redesign
Mature10 - 1250,000 - 100,000 cases; all flagged claims in scope$7.5M - $15M run-rateLoss-ratio delta, antifraud-plan refresh under NAIC #680

The pilot exit criteria the Claims VP should write into the SOW before signature: investigation cycle time under 8 hours on 95% of pilot cases, false-positive resolution rate above 80% on rules-flagged claims (since rules-based detection runs 60-85% false positives in industry studies), and zero unresolved compliance flags from the SIU director on audit-trail completeness. If any of the three miss, the scale phase does not auto-fund.

Stakeholder map: the question each one will ask

A Claims VP who runs this procurement without a written stakeholder map will lose 60 to 90 days of cycle time to ad-hoc objection handling, mostly from CIO security review and legal review of third-party data use. Pre-empt every reviewer with the one question they actually care about. The 12-point evaluation framework in our AI fraud investigation vendor checklist is the working document for these conversations.

StakeholderThe question they will ask
SIU DirectorDoes this give me audit-defensible cases at 800+ per investigator per month without breaking my case routing?
CIO / CTOWhat is the integration surface to Guidewire ClaimCenter or Duck Creek, and does the vendor have a current SOC 2 Type II?
ComplianceDoes the audit trail satisfy NAIC Model Act #680 and align with state-level rules like California 10 CCR 2698.36?
LegalWho owns the investigation work product, how is third-party data licensed, and how does EUO evidence flow into the case file?
Operations ManagerWhat does adjuster training look like, and how does day-one workflow change for front-line claims?
CFOWhat is the coverage-corrected cost per case, and where does the loss-ratio impact show up in the next four quarters?

Compliance is the reviewer most likely to be skipped and most likely to slow the deal in week 10. Bring them in at week one. The relevant state rule for California carriers is California 10 CCR 2698.36, which requires the SIU to investigate each credible referral of suspected fraud and to document any decision not to investigate. AI investigation makes that easier, not harder: every flagged claim gets a documented decision in the audit log, which is the exact artifact the state DOI asks for in market conduct exams.

Change management: deploying AI without burning out SIU adjusters

The change management failure mode at carrier rollouts is anxiety, not pushback. SIU adjusters who handled roughly 10 investigations per month manually do not know what 800+ in the queue means for their job. The Claims VP needs three written commitments before the first case routes, all communicated to SIU staff in the same memo.

  1. No SIU headcount reduction in year one. The freed capacity covers the 75% of flagged claims that were previously uninvestigated, which is where the incremental fraud recovery sits.
  2. The investigator role is redefined as decision-maker over AI-produced evidence packs. The work that requires judgment stays human; the work that requires data gathering, document collection, and database lookups moves to the AI layer running 15+ phases in parallel.
  3. A named escalation path for any AI output the investigator disputes, with the SIU director as the final authority. Every disputed output gets logged and reviewed monthly for model-quality signal.

Named-insurer evidence helps with the SIU town hall. AXA Switzerland's deployment with Shift Technology caught a CHF 120,000 false claim involving intoxicated workers in the first week after deployment, per Samuel Klaus, Head of Fraud at AXA Switzerland, in the published Shift Technology case study. The point for SIU is that AI in production fraud workflows is now standard practice at named European carriers, not a beta experiment.

Shift does in a matter of minutes what would take days for a team of analysts to complete.

- Samuel Klaus, Head of Fraud, AXA Switzerland

The 12-month outcomes report: what to bring to the CEO and Board

Four metrics belong on the month-12 board report, and only four. Investigation coverage as a percent of flagged claims. Cost per investigated case. Cycle time per case. Loss-ratio delta on AI-investigated lines, isolated from rate action and other claims-handling changes.

MetricManual baselineMature-state target
Investigation coverage of flagged claims~25%100%
Cost per investigated case~$2,500~$150
Cycle time per case14+ days2-4 hours
Throughput per investigator per month~10 cases800+ cases

The board narrative the Claims VP brings is not a per-case cost story. It is a coverage story: the carrier is now investigating four times as many flagged claims at lower total cost, with documented decisions on every claim that satisfies state DOI referral requirements and the NAIC #680 antifraud-plan filing. The loss-ratio delta lands in months 6 to 12 as the previously-uninvestigated tier gets worked. Pricing-side actuaries can isolate the signal by line and by quarter.

The forward question for the Board is no longer whether to deploy AI investigation but whether the carrier's antifraud plan, SIU role definitions, and reinsurance reporting reflect a 100%-coverage operating model. Carriers that lock that operating model in 2026 will price 2027 books on a different loss-cost curve than carriers still triaging at 25% coverage. That is the procurement decision the Claims VP is actually making this cycle.

Key takeaways

  • AI claims investigation is a claims-org budget item, not an IT line; the loss-ratio lever and the antifraud-plan filing under NAIC Model Act #680 both sit with the Claims VP.
  • A defensible 90-day pilot scopes to 1,000 - 5,000 cases on 1-2 lines at $150k - $1M; the 12-month rollout lands at $3M - $15M depending on flagged-claim volume.
  • The CFO math the Claims VP brings is coverage-corrected: manual SIU at 25% coverage and ~$2,500 per case versus AI at 100% coverage and ~$150 per case, with the incremental fraud caught in the previously-uninvestigated 75% as the loss-ratio story.
  • SIU headcount does not get cut in year one; the productive move is reallocating capacity from data gathering to decision-making over AI-produced evidence packs.
  • The four metrics on the month-12 board report are investigation coverage, cost per case, cycle time, and loss-ratio delta on AI-investigated lines.

Frequently asked questions

Budget the pilot in three layers. Software at roughly $150 per case for 1,000 to 5,000 flagged claims puts the software line at $150k to $1M. Integration to your claims system and SIU case management typically lands as a fixed implementation fee in the $50k to $250k range. Internal stakeholder time covers SIU director hours, CIO sponsor hours, and compliance review. Most Top 50 carriers land between $300k and $1.2M for a 90-day pilot covering one or two lines, usually auto and workers compensation because referral volume is highest and red flags are well-defined. Avoid scoping smaller than 1,000 cases; the statistical signal on cycle time and false-positive resolution will be too noisy to support a scale decision.

Cycle time and coverage move in the pilot; loss ratio moves at scale. Expect investigation cycle time to drop from 14+ days to 2-4 hours within the pilot's first 60 days, and investigation coverage of flagged claims to move from roughly 25% to 100% on the covered lines by month three. Loss-ratio impact lags because it depends on the share of the previously-uninvestigated 75% of flagged claims that turn out to be confirmed fraud or recovered subrogation. Most carriers model a meaningful loss-ratio delta on AI-investigated lines starting in months 6 to 9 of the rollout, with full-portfolio impact by month 12. Track cost per case and coverage percent monthly; report loss ratio quarterly to the CFO and board.

Three integration points need scoping. The claims system of record (Guidewire ClaimCenter, Duck Creek, or in-house), SIU case management, and any detection vendor (FRISS, Shift Technology, Verisk ISO ClaimSearch) feeding flagged claims into investigation. Most AI investigation platforms expose a webhook or API into the claims system and an outbound document store for evidence packs. Standard CIO checklist: SOC 2 Type II, data residency and PII handling, role-based access control, audit-log export to your SIEM, and a documented incident-response runbook. Plan on 60 to 90 days from signed SOW to first production case, with the bulk of CIO effort in weeks one through six. The hidden costs to flag in the budget are services markup and adjuster training time, not integration engineering.

The failure mode is anxiety, not workload. SIU adjusters who handled roughly 10 investigations per month manually do not know what 800+ in the queue means for their job security. Make three commitments before kickoff. First, no SIU headcount reduction in year one; reallocate the freed capacity to the 75% of flagged claims that were previously uninvestigated. Second, redefine the investigator role from data gatherer to decision-maker over AI-produced evidence packs. Third, name an escalation path for any AI output the investigator disputes, with the SIU director as the final authority. Communicate all three in writing before the first case routes, and review disputed outputs monthly for model-quality signal.

Less than most Claims VPs expect, if the platform produces a complete audit trail. NAIC Model Act #680, adopted in 48 states, requires an antifraud plan with documented SIU procedures, training programs, and standard operating procedures for investigating suspected fraud; it does not prescribe whether the investigator is human or AI-assisted. State rules like California 10 CCR 2698.36 require the SIU to investigate each credible referral and to document any decision not to investigate, which AI investigation actually makes easier because every flagged claim gets a documented decision in the audit log. The compliance review to run: verify the platform's audit log, evidence preservation, and report generation meet your state DOI requirements, and refresh your antifraud-plan filing to reflect the AI-assisted workflow.

Lead with coverage-corrected math, not headline unit cost. Manual SIU runs at roughly $2,500 per case with 25% coverage of flagged claims; AI investigation runs at roughly $150 per case with 100% coverage. On a carrier flagging 50,000 claims per year, manual investigation costs about $31M to cover 12,500 cases while AI investigation costs about $7.5M to cover all 50,000. The CFO question is not whether the per-case cost is lower but what the incremental recovery is from investigating the 37,500 claims that previously closed without an investigation file. Bring two scenarios: a conservative model at a 5% confirmed-fraud rate on the incremental coverage and a base case at 10%, matching the P&C fraud rate cited by the Coalition Against Insurance Fraud. Tie payback to loss-ratio basis points.

Plan for 8 to 16 weeks from RFP issue to signed pilot SOW at a Top 50 carrier. Compress this by running the stakeholder map in parallel, not in sequence: SIU director defines functional requirements, CIO runs security review, compliance reviews antifraud-plan alignment, and legal reviews audit trail and EUO handling, all kicked off in week one rather than week six. Most slippage happens in legal review of data use and indemnification, and in CIO security review when the vendor cannot produce a current SOC 2 Type II report. Pre-qualify on those two items before issuing the RFP. The named-insurer reference call with an existing customer at a peer carrier is the highest-signal step in the evaluation and should happen before, not after, the security review.

← More articles on the Hesper AI blog

See Hesper AI on your documents

Request a demo and we'll run an analysis on your real document samples.