How we built an internal data analytics agent

Qubot is an internal analytics agent at GitHub, powered by GitHub Copilot, that enables employees to access and explore data in the company's data warehouse using plain language queries. Designed to be intuitive and requiring zero configuration, Qubot is available through multiple interfaces such as Slack, VS Code, and the Copilot CLI, allowing employees to ask exploratory questions and receive answers quickly. It leverages a context layer to enhance the accuracy and speed of responses, incorporating data at various curation stages, from raw telemetry to curated datasets. Qubot connects to Kusto and Trino query engines, switching between them based on the complexity of the query, and has successfully reduced the reliance on data analysts by enabling self-serve analytics. This innovation has centralized distributed knowledge across GitHub, encouraging contributions from various product and business teams while fostering a collaborative environment for data-driven decision-making.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.