Native Databases

Tables are now native databases: structured data that lives inside a space and can be queried like a real database, not just displayed as a grid. A new import tool loads JSON files directly, inferring column types automatically as the database is created — in one case, three years of GitHub activity across thousands of PRs, actors, and reviews imported in a single step.

Every database operation — create, delete, upsert, query, and aggregate — is available through the public API, and the aggregation endpoint supports cross-database queries from custom tools. Cell editing isn’t in the UI yet, though it’s already possible through the API. Databases are scoped to a space and follow space membership for permissions; a dedicated place to browse a space’s databases is coming next.