Proof Wall · Anonymized

Outcomes, not adjectives. Six programs. Real numbers.

Every case study below is a live program delivered by FPT's insurance practice. Client names are anonymized at their request; the metrics, timelines, and technical stacks are exactly as verified in the engagement retrospective.

6
Case studies · verified outcomes
$85M+
Documented annual value delivered
4 continents
Delivery footprint
24/7
Human-in-the-loop coverage
Case 01·Bermuda / London·Property Treaty & Facultative

Top-10 Global Reinsurer

Challenge

Submission triage took 3–4 days per broker packet — a mix of unstructured PDFs, bordereaux, and cat model outputs was drowning a 40-person underwriting operations team.

Approach

Deployed a document AI pipeline on FPT's Vietnam AI Factory (NVIDIA H100), fine-tuned on 12 years of proprietary submission history, paired with a 60-person augmented triage pod running 24/7 across Ho Chi Minh City and Hyderabad.

Document AILLM extractionHuman-in-the-loop QAGuidewire integration
Timeline · 14 weeks to production

"We stopped losing quotes to slow response time. Our brokers noticed within a quarter."

— Chief Underwriting Officer, Top-10 Global Reinsurer
Measured Outcomes
40 min
Avg. triage time (from 4 days)
6.2×
Throughput per underwriter
97.1%
Extraction accuracy vs. human baseline
$4.8M
Annualized operating savings
Metrics independently verified in engagement retrospective. Anonymized at client request under mutual NDA — full reference available under counter-NDA.
Case 02·Northeast US·Commercial Auto & Trucking

US Specialty Carrier

Challenge

Claims cycle time was 34 days on average, with a rising indemnity leak on soft-tissue and staged-collision fraud rings across five states.

Approach

Built a computer-vision FNOL intake, integrated with a proprietary fraud-graph model and a US-onshore adjuster pod for empathy-critical calls. Full straight-through processing for glass, tow, and low-severity claims.

Computer visionFraud graph MLFNOL orchestrationDuck Creek Claims
Timeline · 9 months (three phased releases)

"The fraud model paid for the entire program in the first eight months."

— SVP, Claims Strategy, US Specialty Carrier
Measured Outcomes
71%
Straight-through processing
11 days
Avg. cycle time (from 34)
3.2 pts
Combined ratio improvement
$18M
Prevented fraud leakage (Y1)
Metrics independently verified in engagement retrospective. Anonymized at client request under mutual NDA — full reference available under counter-NDA.
Case 03·Singapore / Malaysia / Vietnam·Individual Life, Critical Illness, Group Health

APAC Life & Health Insurer

Challenge

Underwriting a term-life application took 6–9 days with heavy medical-record chasing. New agent onboarding was capped by manual reference checks.

Approach

Domain-tuned underwriting LLMs on the Vietnam AI Factory, integrated with tele-underwriting scripts, wearables data, and an offshore medical-review desk staffed with licensed clinicians.

Underwriting LLMTele-underwritingWearables dataSapiens core
Timeline · 6 months

"Speed of decision is our new competitive edge — bancassurance channels rebooked us as tier-1."

— Chief Distribution Officer, APAC Life & Health Insurer
Measured Outcomes
42 hrs
Median STP decision time
63%
Applications auto-decisioned
+18%
Agent NPS
22%
Reduction in cost-to-issue
Metrics independently verified in engagement retrospective. Anonymized at client request under mutual NDA — full reference available under counter-NDA.
Case 04·Germany / Netherlands·Embedded Motor & Renters

EU Insurtech Unicorn

Challenge

Partner onboarding took 14+ weeks per embedded channel, and the monolithic quote engine could not sustain a Black-Friday retail traffic spike without falling over.

Approach

Re-platformed the quote-and-bind stack onto composable, API-first services on sovereign EU cloud. Built a partner onboarding kit that turns integrations into a self-serve, six-day journey.

Headless product APIsEvent-driven coreEU data residencyTerraform-as-a-service
Timeline · 8 months

"We shipped four new distribution partners in the quarter we used to ship one."

— VP Engineering, EU Insurtech Unicorn
Measured Outcomes
6 days
New partner onboarding (from 14 wks)
11,400
Quotes/sec peak (from 380)
99.98%
Availability
38%
Lower cost per quote
Metrics independently verified in engagement retrospective. Anonymized at client request under mutual NDA — full reference available under counter-NDA.
Case 05·North America / EMEA·Personal Lines Home & Auto

Global P&C Mutual

Challenge

A 30-year mainframe policy admin system blocked every product change; a full rate revision took 11 months from actuarial approval to market.

Approach

Progressive strangler migration to a cloud-native core, with FPT's Guidewire accelerators handling PolicyCenter and BillingCenter. Zero-downtime cutover per state, driven by a dual-write reconciliation engine.

Guidewire CloudStrangler patternDual-write reconciliationModel risk (SR 11-7)
Timeline · 22 months (14 states live)

"We finally price like a digital carrier without ripping out the balance sheet."

— Chief Information Officer, Global P&C Mutual
Measured Outcomes
6 wks
Rate revision to market (from 11 mo)
0
Downtime incidents at cutover
31%
Lower run-cost per policy
$62M
Retired mainframe TCO
Metrics independently verified in engagement retrospective. Anonymized at client request under mutual NDA — full reference available under counter-NDA.
Case 06·London Market·Cyber & Specialty Financial Lines

Lloyd's Managing Agent

Challenge

Cyber portfolio accumulation was invisible in real time — aggregation runs took 9 hours, so capacity decisions lagged the market by a full day.

Approach

Ported the cat/aggregation engine onto the Vietnam AI Factory's NVIDIA GPU cluster with parallelized Monte Carlo, integrated live with the syndicate's underwriting workbench and PPL feeds.

GPU Monte CarloPortfolio simulationPPL integrationReal-time exposure API
Timeline · 10 weeks

"Real-time accumulation changes how we underwrite the entire class."

— Active Underwriter, Lloyd's Managing Agent
Measured Outcomes
52×
Faster full-book aggregation
11 min
Full portfolio run (from 9 hrs)
+14%
Capacity deployed within tolerance
0.7 pt
Improvement in expected loss ratio
Metrics independently verified in engagement retrospective. Anonymized at client request under mutual NDA — full reference available under counter-NDA.

Want the full reference under NDA?

Our strategy team will walk you through the engagement retrospective, delivery model, and commercial structure for any of the programs above — matched to your own portfolio.

Request a reference callExplore the AI Factory →