Skip to content

Natural-language analytics

DataQuery AI

Upload a dataset of up to a million rows and ask questions in plain English. Everything runs in the browser.

What I built

  • The LLM extracts the schema and generates SQL that is validated before it runs.
  • Chooses table or chart output from the shape of the query result.
  • In-browser query execution tuned to stay inside browser memory limits.
Stack
Next.js, LLM APIs, SQL, In-browser query engine
Kind
Natural-language analytics

Next: EduVid AI or the minaz case study