Project

Ask questions of your own data

A chat assistant connected to a BigQuery database. You type a question in plain English and it returns the answer as a table with a short explanation.

Analytics

Project snapshot

Industry
Analytics
Input modalities
Questions in plain EnglishA BigQuery database
A chat assistant answering a question with a table of dealers and lead counts, followed by notes on the row count and missing values.
An answer from the assistant: the table, the row count, and a note on which values are missing.

Situation

The situation

Data in a database is only useful to the people who can write a query. Everyone else waits for someone who can.

The build

What I built

I built a chat assistant on top of a BigQuery database. It turns a plain-English question into a query, runs it, and explains the result.

In use

What it does

  • You type a question in your own words.
  • The assistant writes the query, runs it, and returns the answer.
  • An answer can carry notes on the data, such as the row count and missing values.

Watch it run

A 40-second run. Someone asks for a breakdown of inventory by dealer and gets the table back. No sound.

In detail

A closer look at the problem and how the build works.

The situation

Data in a database is only useful to the people who can write a query. Everyone else waits for someone who can.

What I built

I built a chat assistant connected to a BigQuery database. You type a question the way you would ask a colleague.

How it works

  • The assistant turns the question into a database query.
  • It runs the query against the database.
  • It reads the result and writes the answer.
  • The answer can include notes on the data, such as the number of rows and any missing values.