GitHub built Qubot, an AI agent powered by GitHub Copilot that lets employees ask questions about company data in plain language and get answers in seconds, removing a major bottleneck for product teams.

What happened: GitHub created Qubot, an internal analytics agent accessible via Slack, VS Code, and the Copilot CLI, that allows employees to query the company's data warehouse by asking questions in plain English rather than writing SQL. The tool connects to Kusto and Trino query engines and automatically switches between them based on what the question requires.

Why it matters: Hundreds of GitHub employees are now using Qubot to explore data autonomously, which has reduced the volume of questions in the data and analytics Slack channels dramatically. This solves a long-standing problem: product teams previously struggled to access insights without dedicated analytics support. The shift means data teams can focus on complicated questions while broader teams gain self-serve access to decision-making data.

What to watch: The context layer—structured documentation about data models, query examples, and metric definitions—proved critical to accuracy and speed; GitHub found that well-curated context makes Qubot three times faster at returning the right answer. This suggests that investing in clear, organized data documentation has become a first-class priority for analytics engineering, not an afterthought.