QueryBot

Data is locked behind SQL

Answering a simple question often means writing SQL, which means knowing the query language and the schema. That puts a wall between the people who have questions and the data that holds the answers. Most of the friction is not the question itself, it is the translation step.

Ask a question, get an answer

1. Connect. Point QueryBot at a database or a local file.
2. Ask. Type your question in plain English.
3. Translate. A language model turns the question into the right query.
4. Answer. It runs the query and returns the result, no SQL required.

Packaged for real use

A Python package with a JavaScript front end, published to PyPI so anyone can pip install it and start querying. Shipping it as a real package, not just a repo, was the point: make natural-language data access something you can actually depend on.

Published and installable

QueryBot is live on PyPI and open source on GitHub. It is the most direct expression of a theme that runs through my work: natural language as the interface to systems that used to demand specialist syntax.