Agent system for automated processing of insurance claims

Claim adjusters had to process a wide range of unstructured documents (handwritten reports, data-less PDFs, photos, invoices, powers of attorney). Each case took an average of 15 minutes of manual work. Detecting errors and inconsistencies (e.g., wrong account number, suspicious items, incomplete documentation) was difficult. There was no system capable of preparing a complete case summary and supporting fast payout decisions. Processing was limited to adjusters’ working hours, leaving clients waiting.

BigHub helped design and deliver a modular system based on AI agents capable of automating claims document processing. The solution combined a pragmatic iterative development model with modern AI tools: Hybrid approach: A rules engine enforces business rules combined with AI for attribute extraction and decision-making. Document data extraction: Azure Document Intelligence extracted data from invoices, forms, and handwritten notes. Agent orchestration: LangChain and LangGraph served as the core framework for agent orchestration and management. Event-driven architecture: Kafka processed commands and events across the multi-layered architecture. Cursor IDE integration: Deep use of LLMs in the development environment enabled fast iteration and strong business involvement.

Claim adjusters

9

Azure Document Intelligence, LangChain, LangGraph (Python), Kafka, Cursor IDE, hybrid rules engine + LLM

Head of Innovation / Digital Transformation

Joint implementation with the client

We are available to the client.

Trained poweruser

Trained poweruser

Case processing time was reduced from ~15 minutes to ~2 minutes (approx. 87% time savings). Simple cases are resolved fully automatically without adjuster intervention. The system runs continuously 24/7, unlike adjusters’ working hours.

“Customer feedback on automated claims processing is very positive. Significant acceleration of case resolution, simplified communication, and the system’s ability to independently decide on payouts within minutes improves the overall customer experience.” – Jakub Lada, AI Digitalization Expert, Direct Insurance.

Effective project management – clearly defined roles, responsibilities, and regular communication. Team collaboration – high level of engagement and open knowledge sharing among team members. Flexibility and adaptability – ability to respond quickly to changes and new requirements.

When planning an AI project, we recommend starting with a clearly defined business goal and a realistic estimate of benefits and risks.