Automations 1 min read

AI agents join Automations: put a skill on a schedule

The same AI that answers your questions can now work without you: a new agent task runs a skill — exec summary, data-quality check, anomaly explanation — inside your scheduled automations, with read-only tools, query budgets, and a full audit trail on every run.

Databasin One answers questions when you ask. As of this release, it also works when you don't: AI agents are now a task type in Automations.

A skill is a versioned, shareable package of instructions. An agent task runs one against your governed data, on your schedule, in your workflow. Chain it after your pipelines refresh; deliver what it produces to email, Slack, or Teams.

Starter skills, ready to schedule

  • Executive summary reads the fresh numbers, writes the narrative, builds the charts, ships the PDF. Monday, 7am, already in Slack
  • Data-quality check sweeps your tables for gaps, drift, and surprises, and files the report
  • Anomaly explanation takes a metric that moved, traces which segment and which source, and shows the query that proves it

Skills are managed like code. They're versioned, shareable across the team, and curated by your admins. Run one with a click in Databasin One, or let an automation run it forever.

Automation you can audit

Agents run with guardrails, not vibes: read-only tools, query budgets, and row caps. Every run writes a full audit trail of what it read, what it ran, and what it produced. If a number's in the summary, the receipt is in the log.

One more task card on the same canvas as your SQL, dbt, and notebook tasks — drag it into a stage and go.

NewerQuery one Apache Iceberg table from Trino, Spark, and DuckDB → News & insights ← Two weeks of polish: global dashboard filters, connection credentials, Doris everywhereOlder

See it on your own data — five minutes, $50 in credit, no card.