Answer company data questions with an agent that combines Snowflake and workspace context
David Thomas of Notion describes Data Scout, an internal agent that answers employees' plain language questions from governed data in Snowflake. It pairs warehouse numbers with business context from Notion pages, account notes and Slack, so a rep preparing a renewal can check whether AI adoption in an account matches the customer's rollout plan and keep asking follow ups in the same thread. Each user signs in to Snowflake with their own credentials, so answers respect their personal data permissions. Teams built further agents on its shared MCP connection, including a reporter that schedules recurring reports. Two months in, the data team was getting about 80% fewer weekly requests, answers take a couple of minutes, and about half of Notion staff now use it.
Tools used
- Notion Custom Agents
- Snowflake MCP
- Slack
Sources (1)
- Originalhttps://www.notion.com/blog/how-we-built-a-personal-data-scientist-for-every-notion-employee (opens in a new tab) ↗
David Thomas (Notion) · notion.com