We are happy to present our most recent integration Weaviate

Weaviate is a Dutch company specializing in providing vector databases that excel at rapid searches across billions of vectors. We were approached by Weaviate to implement their solution into our agentic workflow. The implementation of their system was perfect for our use case of improving our agents' similarity search.  Similarity search is used by the agent to relate new incoming claims to past cases with human reviewed decisions.

1. Initial Analysis

When a claim comes in, our Claims agent gets to work instantly.

 It starts by extracting and structuring all key information. From there analyzing supporting documents, and running built-in validity and fraud checks. Automating the manual review that normally slows insurers down.

Valid documents move straight into automated processing.
 Suspicious or incomplete ones are escalated to Stage 3 for specialized handling.

2. Detailed Analysis

With clean, structured data from Stage 1, our agent moves into deeper reasoning.

It applies insurer-specific rules, policy conditions, and situational logic to interpret the claim in context just like a human handler would.

3. Create a Summary of the analysis

Stage 3 uses Weaviate to supercharge decision-making.

The detailed analysis from Stage 2 is embedded into vectors and compared against thousands of previously resolved claims through Weaviate’s high-speed similarity search.

Instead of relying on rigid rules or keyword matching, Weaviate lets the agent instantly identify contextually similar historical cases. including edge cases, special conditions, and rare claim patterns.

This gives the agent real precedent-based intelligence. The agent then enriches the summary with any external data needed (e.g., flight logs, weather data, official reports), creating a complete, context-aware snapshot of the claim.

Together with Weaviate, we’re enabling context-aware, precedent-driven automation for insurers. Here’s how the integration works.

4. Final decision.

Using the enriched summary and similarity insights surfaced through Weaviate, our agent produces a fully reasoned final decision on the claim

The decision is backed by:

• contextual precedent from similar historic cases

• policy-specific logic

• external data (flight logs, weather info, reports)

• fraud/validity signals from earlier stages

The agent outputs a clear Decision + Explanation package, mirroring the structure of a human adjuster’s reasoning but delivered instantly and with full traceability.

Book a demo or get in touch to learn more.