Unlocking Financial Insights With Finch
- Source
- uber.com
- Author
- Uber Engineering 1
- Date
Context
Finch helps finance staff ask for company data in ordinary language by translating business terminology into database queries. Uber Engineering gives the assistant curated tables and a searchable mapping between everyday terms, column names and stored values. That mapping matters because selecting the right table still leaves the problem of expressing the user's intended filters.
Finch uses the retrieved metadata to write and execute a query, and can revise it when the database returns an error. Uber evaluates generated queries by comparing their results with reference queries, allowing different SQL wording to count as correct. The architecture account explains this connection between business vocabulary and structured data.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- RAG — Retrieval-augmented generation — fetching relevant documents first and pasting them into the model's context so it answers from your data instead of memory.
Checking sign-in…
Loading comments…