seKer AI
Live demo · RAG

Document Q&A with sources

Ask a question, get the answer with the passage it came from, or a refusal.

Interactive demo on sample data. Runs in your browser. Not a client project. All live demos

Sample: Harbor Studio partner delivery handbook (fictional)

What runs in this demo

  • Real retrieval code: TF-IDF weighting and cosine ranking over passages, running in your browser.
  • Answers are extracted from the best passage and shown with its section reference. There is no language model in this demo.
  • A refusal threshold: if nothing in the document is close enough, it says it cannot answer.
  • A live evaluation panel that scores ten test questions against the sample document, so quality is measured, not claimed.

What a partner project adds

  • Document ingestion for PDF, Word and web pages, with chunking tuned to the content.
  • Embeddings and a vector store, with reranking, in place of keyword weighting.
  • A language model that writes the answer from the retrieved passages, keeping the citations.
  • Access control per document, an evaluation set built from the client's real questions, and a REST endpoint for the partner's product.

Built for: Software houses whose client has a pile of documents nobody can search.

Discuss a similar build

Try another demo

Have a project that needs additional AI or engineering expertise?

Share a short brief. We’ll tell you whether the work is a fit and how a collaboration could be scoped.