← Back to work
Redash natural-language chatbot
LLMsFlaskRedashDocker

Redash natural-language chatbot

2024Training project

A training project that converts natural-language questions into SQL and returns the result through Redash.

The problem

Exploring data in Redash meant writing SQL, which blocked non-technical users from answering their own questions.

How it works

  1. 01, Input
    Plain-English question

    A business question from a non-technical user.

  2. 02, Model
    Text to SQL

    An LLM turns the question into a SQL query.

  3. 03, Tool or code
    Redash query

    Redash runs the query against the data.

  4. 04, Output
    Results and charts

    The answer comes back as results and visualizations.

InputModelTool or codeOutputA Flask add-on, containerized with Docker.

What I did

  • Built a Redash add-on that turns plain-English questions into SQL.
  • Returned results and visualizations without the user writing a query.

Results

  • Demonstrated a natural-language interface for querying and visualizing data in Redash.
View code ↗
← All projects