Document question answering

Ask a question of a document set and the answer comes back with the passage it was drawn from, so it can be verified rather than taken on trust. Where the corpus does not contain an answer, that is what it reports instead of inventing one. Questions can be asked across a whole corpus or scoped to a single file, and answers keep their citations.

You can use the models we operate or bring your own model and we run it on your cluster. Work arrives as an API call, a batch over a folder or bucket, or a resident on-premise service, and results return as structured output with a per-job audit record.