GitHub built Qubot, an AI agent powered by Copilot that lets employees ask questions about company data in plain language and get answers within seconds, eliminating the need for specialized analytics support.

What happened: GitHub created Qubot, an internal tool that connects to the company's data warehouse and runs on GitHub Copilot. Employees can ask questions in Slack, VS Code, or the Copilot CLI, and the agent automatically generates the correct query and returns results—storing them as markdown reports that can be refined or shared.

Why it matters: Before Qubot, product and engineering teams struggled to access data without a dedicated analyst, because figuring out the right data model, filters, and query validation was difficult. Qubot removes this bottleneck and has been widely adopted—the number of questions in analytics Slack channels has reduced dramatically because teams can now explore data with greater autonomy and reach out only for complicated questions.

What to watch: The context layer—structured knowledge about data schemas, query examples, business rules, and metric definitions—proved crucial to accuracy and speed; in experiments, well-curated context made Qubot three times faster at returning the right answer. Qubot defaults to Kusto for exploratory questions over recent data and automatically switches to Trino for complex joins and deeper historical analysis.