SageMaker Data Agent vs Data Workers: Notebook Assistant or Agentic Data Platform?
The SageMaker Data Agent helps a person write code and SQL in Unified Studio. Data Workers runs data operations across your whole AWS estate.
The Amazon SageMaker Data Agent and Data Workers both put AI agents to work on AWS data, but they do different jobs. The SageMaker Data Agent is an assistant inside SageMaker Unified Studio: it helps a person write code, SQL and execution plans in notebooks and the Query Editor, and the person runs the result.
Data Workers is the agentic data platform: it runs data operations across Glue, Redshift, S3 Tables, Lake Formation, dbt and BI, from the alert to the verified fix, with approvals and receipts. If your team wants the data platform to run itself, Data Workers is the stronger choice.
Key Takeaways
- •The SageMaker Data Agent is a strong notebook and SQL assistant. It generates code, SQL and plans, diagnoses errors and uses business context from SageMaker Catalog.
- •It works inside SageMaker Unified Studio, for the person at the keyboard, and is billed per credit.
- •Data Workers owns outcomes across services: incidents, schema changes, pipeline recovery, access requests, cost cleanup and catalog upkeep.
- •Data Workers works from the coding agent your team already uses and covers systems outside AWS, such as your dbt project, Airflow, Snowflake and BI.
- •Many teams keep the Data Agent for notebook work and run Data Workers for operations.
What the SageMaker Data Agent does
AWS made the SageMaker Data Agent generally available in November 2025 as part of SageMaker Unified Studio, and has added to it since.
- •Generates code, SQL and multi-step execution plans from natural language in notebooks and the Query Editor.
- •Diagnoses errors in notebook cells and suggests fixes.
- •Uses business context from SageMaker Catalog, such as glossary terms and metadata (June 2026).
- •Adds charting and support for materialized views (March 2026).
- •Is billed per credit, at $0.04 a credit on AWS's pricing page.
It is a good assistant for an analyst or engineer working inside Unified Studio. It speeds up the time from question to working code.
What Data Workers does
Data Workers runs the operations of a data platform. The Autonomous Data Conductor owns each outcome and directs 20+ specialist agents in the Data Agents Swarm, working from one governed context graph built by Data Context Wizard.
- •Traces a broken number from the dashboard back through Redshift, dbt, Glue and DMS to the cause.
- •Drafts or applies the fix where the problem started, at the autonomy level set for that domain.
- •Reruns and rebuilds what is affected, then verifies the downstream result.
- •Catches schema changes in review, before jobs write nulls.
- •Proposes least-privilege Lake Formation grants and works the access queue.
- •Ties spend to the models and schedules behind it and drafts each fix for its owner.
- •Leaves a tamper-evident receipt on every change.
Head to head
| Job | SageMaker Data Agent | Data Workers |
|---|---|---|
| Write a query or notebook cell | Strong, inside Unified Studio | Through your coding agent, with governed context |
| Diagnose an error | The notebook cell in front of you | The incident across services, from source to dashboard |
| Fix and rerun | Suggests code for a person to run | Drafts or applies the fix behind approval, reruns and rebuilds |
| Verify the outcome | Left to the person | Reruns checks, diffs values and records a receipt |
| Schema changes before they break | Not its job | Schema Evolution and Data Change Review agents |
| Access requests | Not its job | Least-privilege grants proposed for Lake Formation |
| Cost cleanup | Not its job | Spend tied to its source and fixes drafted for owners |
| Systems outside AWS | SageMaker Catalog context | dbt, Airflow, Snowflake, BigQuery, Tableau and more in one graph |
| Where people work | SageMaker Unified Studio | Claude Code, Codex, Cursor, Kiro and Q Developer |
| Autonomy | Assistant: a person runs every result | Per-domain levels from observe to autonomous |
| Pricing | Per credit | No usage meter; unlimited seats; see pricing |
Why the difference exists
AWS builds the best primitive for each service and an assistant that helps a person use it. The Data Agent writes the notebook cell; the person decides and runs it. That keeps each service's AI inside its own boundary, which is the right design for a cloud provider.
Owning a data incident across services is a different product. It needs blast-radius scoping across Glue jobs, Iceberg tables, Redshift, dbt and dashboards, approvals that vary by domain, rollback, verification and a record of why each change was made. That is what Data Workers is built to do.
Can you use both?
Yes, and many teams will. Keep the SageMaker Data Agent for notebook and SQL work inside Unified Studio. Run Data Workers for everything that has to be traced, fixed, approved and proven across the estate. Data Workers reads SageMaker Catalog and the Glue Data Catalog as sources for AWS assets, so the business context your team curates there carries through. Our guide You're on Amazon SageMaker covers the setup, and Data Workers on AWS compares the full AWS stack service by service.
Frequently Asked Questions
Is Data Workers an alternative to the SageMaker Data Agent?
For notebook and SQL help inside Unified Studio, the Data Agent is built for that. For running data operations across services, Data Workers is the stronger choice: it fixes, verifies and records the work the Data Agent leaves to a person.
Does the SageMaker Data Agent fix pipelines automatically?
It diagnoses errors and generates code and SQL for a person to review and run inside SageMaker Unified Studio. Data Workers drafts or applies fixes across services behind approval and verifies the result downstream.
How is Data Workers priced compared with the Data Agent?
The SageMaker Data Agent is billed per credit. Data Workers has no usage meter, unlimited seats and no markup on model spend, and its core is Apache 2.0. See pricing.
Can Data Workers use SageMaker Catalog context?
Yes. SageMaker Catalog and the Glue Data Catalog stay the source for AWS assets, and Data Workers joins them with your dbt project, orchestrator and BI in one context graph.