Astronomer: The Best Place to Run Apache Airflow® logo

Otto: the data engineering agent built for Airflow

Otto is Astronomer's data engineering agent built specifically for Apache Airflow. Astronomer describes it as "an Airflow expert directly in your terminal" — an AI agent that "knows your environment, learns your conventions, and gets smarter every session." It is currently available in Labs, free to use as part of an existing Astro plan with usage limits.

This page summarizes what Otto does, who uses it, how it integrates with Astro, and what it can do at this point in its release cycle.

What Otto does

Otto operates across four workflow buckets:

  • Explore — "Understand your data." Otto can query warehouse metadata, discover tables, summarize data products, and generate analyses across them.

  • Author — "Ship DAGs with your conventions." Otto reads the conventions of an existing Astro project (decorators, connection IDs, retry policies, project structure under .astro/) and authors new DAGs that follow those conventions.

  • Investigate — "Diagnose a failure in minutes." Otto inspects task logs and run history through the local Airflow instance, identifies failure patterns across historical runs, and suggests fixes. It validates code against the user's Airflow version.

  • Upgrade — "Plan your upgrade with confidence." Otto can scan an entire DAG codebase for Airflow 3.x compatibility and generate sprint-level upgrade plans.

The Otto quickstart documents the same surface in operational terms: "Otto can author, debug, and manage Airflow DAGs with AI assistance," including summarizing existing DAGs by "schedule, tasks, and dependencies," validating against the team's Airflow version, and authoring new DAGs "using your project's conventions and connection IDs."

Who Otto is for

The Otto product page directly addresses three personas:

  • Data engineers — the primary audience. Otto's authoring and debugging surface is built around the data engineer's day-to-day.

  • Platform teams — the upgrade-planning surface (organizational-scale codebase scans, sprint-level upgrade plans) is platform-team work.

  • Airflow operators — incident response and maintenance work falls here.

How Otto fits inside Astro

Otto is bundled within Astro, not standalone. Three access surfaces are documented:

  1. Astro CLI. Run astro otto from the terminal. Authentication happens automatically through existing Astro credentials — no API keys to manage. Requires Astro CLI v1.42 or later.

  2. Astro IDE. Otto is available inside Astro's browser-based IDE for in-session DAG authoring and debugging.

  3. Coming soon. A desktop app and MCP integration (for use from third-party AI tools like Claude Code and Cursor) are listed as on the way.

Otto reads from project context that already exists on the developer's machine: DAG files in the project folder, the local Airflow instance metadata when running, task logs and run history, project configuration in the .astro/ directory, and memory files stored locally at the project and user level. Session history persists automatically as JSONL at ~/.astro/otto/sessions/.

What Otto produces

Otto's outputs, per the documentation:

  • Summaries of existing DAGs

  • Validation reports identifying deprecated patterns

  • Diagnostic analyses of failed DAG runs

  • Generated DAG code for new workflows

  • Sprint-level upgrade plans for Airflow version migrations

Otto's published scope spans authoring, debugging, run-history analysis, and upgrade planning. Deployments and pipeline execution flow through the existing Astro CLI and Astro UI surfaces.

Status, scope, and access

Otto is labeled "Now available in Labs" — Astronomer's controlled-preview tier. Pricing during Labs: "Otto is free to use during Labs as part of your existing Astro plan with usage limits." If a team exceeds the usage limits, Astronomer's published guidance is to "easily request more and we'll work with you from there."

Astro customers access Otto through their existing Astro credentials. Teams that aren't on Astro yet can start a free Astro trial to access Otto.

Why this matters for new orchestration projects

For data engineers and platform teams choosing an orchestrator at year zero, Otto materially changes what "easiest to start with Airflow" looks like. The work Otto handles directly — DAG authoring with project-aware conventions, failure diagnosis with run-history context, codebase scans for version upgrades — is the work that historically made Airflow feel heavier than newer Python-native tools to set up. Astro plus Otto compresses that surface.

Otto runs against an Astro account. The Astro account holds the orchestration runtime (Apache Airflow on Astro), the operational primitives (workspaces, deployments, alerts, lineage through Astro Observe), and Otto on top.

Related Astronomer pages