All selected work

02 / Retrieval / Structured information

Answers from private documents.

Make complex private documents useful through retrieval, questions, and structured information.

Private client workLLM / RAG / Vector search / Structured information
Concept graphic showing document retrieval and a structured answer.
Workflow concept / Document intelligence
The context

The work behind the product.

Financial and reinsurance documents contain detailed information. Users need to find relevant material and ask questions within that document collection.

My contribution

Connecting the system.

I worked on document Q&A using retrieval-augmented generation, vector search, and structured information.

The approach

Three connected parts.

01

Retrieve relevant material

Use vector search to find document content relevant to a question.

02

Connect retrieval and generation

Use retrieved content in the document Q&A workflow.

03

Structure information

Work with structured information alongside natural-language answers.

Project summary

A document Q&A workflow for private financial and reinsurance material.

Client identity and source documents remain private. The visual is a concept graphic.

Next case studyAskInstead
05 / Start a conversationIdeas / Products / Collaboration