Accelerating portfolio growth with document processing in seconds

For insurers, bordereaux files are essential for managing liabilities and tracking premiums, assets, and claims. However, their inconsistent formats significantly hindered our client’s operational efficiency. For example:

• Mapping inbound bordereau data to a standard 100-column template could take as long as three hours.
• Slow data mapping delayed access to critical insights such as overall risk exposure and portfolio performance.

As a result, scaling broker onboarding without increasing headcount was impossible, stalling growth. Our client required a single ingestion solution accessible to business users across multiple units.

To address this, we implemented a Data Intelligence Solution using a moden data platform enhanced with GenAI capabilities:

• Created a Databricks environment to consolidate and capture incoming data in real-time.
• Applied a large language model for mapping and similarity matching data to standardize formats, with automated data quality testing.
• Deployed a Streamlit application for end-users to interact with the product in natural language.

• Built, scaled, and transitioned the solution using a phased approach, supported by weekly user engagement.

The result was a Databricks-based ingestion and standardization tool featuring human-in-the-loop validation.

The Kubrick solution allowed the client to map bordereaux files rapidly, accelerating onboarding and unlocking growth potential. Key impacts included:

• Reduced manual mapping time from three hours to under one minute.
• Onboarded 100 new providers and processed
• 1,500 files per month within the first three months.

Expert validation delivered transparency, fostering trust in the solution and driving collaboration across previously siloed teams.