TechEniac built an AI chatbot that handles the entire insurance claims intake process, and generates an FCA-compliant audit record for every conversation, automatically.
ClaimBot was founded by two former insurance claims adjusters who had spent years managing First Notice of Loss (FNOL) intake for UK insurance brokers. Having worked inside the claims process, they understood the core problem: FNOL handling was a manual, time-intensive administrative function that consumed 70+ person-hours per week for brokers handling 200+ claims weekly. The process wasn't complex because claims were complex it was complex because the architecture was broken. A single phone call was doing three jobs simultaneously: gathering structured data, providing customer guidance, and creating a compliance record. No single tool existed to address all three, so claims handlers did everything manually, every time.
The ClaimBot founders identified three compounding problems making the status quo unsustainable:
40% of claim events happen outside business hours. Without 24/7 intake, customers waited to report incidents, leading to missing details, lower data quality, and longer claim processing times.
The FCA’s Consumer Duty requires UK insurers to provide clear, timely information throughout the claims journey. Unstructured phone interactions can make it difficult to prove what was communicated and when, creating potential compliance and audit risks.
When claims moved between teams, key details were often lost in incomplete or conflicting notes. Customers had to repeat information, specialists missed important details, and claims slowed down—leading to lower customer satisfaction.
Handle the entire FNOL process 24/7 across web with zero human involvement for standard claims
Collect every required field, validate data in real-time, process document uploads, and push structured records into the broker's CMS automatically
Generate an FCA-compliant communication record for every interaction every question, every answer, every disclosure
Integrate bi-directionally with both Guidewire ClaimCenter (REST) and Duck Creek Claims (SOAP) the two legacy CMS platforms their target brokers used
A smart AI claims agent, powered by LangChain and Claude Sonnet, guides customers through the FNOL process via web chat or voice. It asks relevant questions based on the claim type, validates details like dates, postcodes, and policy numbers in real time, and uses a 3-attempt rule to avoid loops. If a field cannot be completed, it is marked as pending and the claim continues smoothly.
Customers can upload photos and documents like police reports or medical certificates while the claim continues in the background. Google Cloud Vision extracts important details such as dates, amounts, reference numbers, and vehicle information automatically. For motor claims, AI also assesses damage severity and flags serious cases for priority review.
Every customer interaction is securely recorded with a timestamp and SHA-256 verification. The system captures questions, customer responses, and information shared during the claim. These records are stored in an immutable audit log, ensuring they cannot be changed or deleted and supporting FCA Consumer Duty compliance.
The system connects with Guidewire ClaimCenter and Duck Creek Claims, automatically mapping claim data to each platform’s requirements. A submission queue manages high claim volumes without data loss, while failed submissions are stored separately for manual review.
The voice experience uses the same claims engine as web chat, with shorter questions and clear confirmations. Twilio Voice manages the call experience, while Whisper converts speech to text and helps recognize policy numbers accurately. ElevenLabs provides natural-sounding voice responses for a smooth customer experience.
Agent Looping on Evasive Responses
Implemented a 3-attempt retry limit per field. After 3 attempts, the field marks pending, the agent acknowledges the gap naturally ("We'll follow up on that detail separately"), and intake continues. The claim flags for human review on pending fields only. Outcome: intake completes in under 5 minutes even when some fields are unresolved.
Guidewire Rate Limiting During Storm Events
Built a submission queue capped at 10 API calls per second with exponential backoff and jitter on 429 errors. Submissions exhausting all retries route to a dead-letter queue for manual review. Outcome: simulated storm surge with 50 simultaneous claims resulted in zero submissions lost; all claims pushed successfully within 12 minutes.
Voice Transcript Errors on Policy Numbers
Added a post-transcription phonetic normalization layer that reconstructs probable reference formats and validates against the broker's policy number regex. If validation fails, the voice agent falls back to character-by-character spelling. Outcome: 98% of policy number mismatches resolved without human intervention.
Reduction in FNOL Processing Time
Average intake reduced from 22 minutes (human) to 4.8 minutes (AI-handled). At 200+ claims per week, this eliminates 70+ person-hours of administrative work weekly.
Out-of-Hours Availability
24/7 intake channel ensures all claims submitted outside business hours are captured at time of incident, eliminating reporting delays that previously degraded data quality.
CMS Auto-Population Accuracy
AI-collected intake fields passed Guidewire and Duck Creek validation on first submission without human correction or cleanup.
FCA Audit Compliance
Every interaction generates a timestamped, immutable, hash-verified Customer Communication Record satisfying FCA Consumer Duty documentation requirements. Eliminates manual compliance documentation burden.
AI-Resolved Claims
Standard claims processed entirely by AI without human involvement. Remaining 31% escalated for multi-party disputes, injury claims, and specialist judgment situations.
Customer Satisfaction
Post-interaction CSAT for AI-handled claims matched traditional phone-based intake benchmarks (industry average: 3.8/5 for digital claims channels).
AI / ML
Backend
Frontend
Integrations
Cloud & DevOps
MediaTech / EntertainmentTech
AI-powered music interpretation platform helping Spanish-speaking audiences discover songs, explore lyrics, and understand the stories behind them.
TravelTech / Tourism SaaS
AI-powered business intelligence and operations platform for tour operators. Unified bookings, analytics, CRM, customer communication, marketing, and automation in one connected system.
