Datus is the open-source data engineering agent for the modern data stack — one AI data engineering agent that connects your warehouse, catalog, semantic layer and BI, grounded in an evolvable context engine your team owns. Apache 2.0 · self-host or free playground.
Copilots and NL2SQL solve one prompt at a time. A data engineering agent owns the lifecycle — plan, write, run, validate, deploy, monitor.
Delivery on top, Intelligence in the middle, a Data Layer underneath. Three stacked layers that turn Datus from a chat wrapper into a production-ready data engineering agent.
How teams reach the agent
How the agent thinks
What the agent stands on
From SQL development to monitoring — eight lifecycle phases orbit one Datus agent, giving your team autonomous data engineering in a single, always-in-context workflow.
Four workflows that show up on day one — same data engineering agent, same context, different jobs to be done.
Same data engineering agent, same context — four surfaces (Studio, CLI, Chatbot, MCP) so every data engineer can start where they already work.
Point the Datus data engineering agent at what you already run. Governance and dialect handling for every warehouse ship in the box.
Data engineering agents, the Datus context engine, surfaces, pricing and how Datus compares.
A data engineering agent is an AI system that owns data work end to end — not just answering questions, but planning, writing SQL, running pipelines, validating results, deploying models and monitoring what it shipped. Unlike a text-to-SQL copilot, a data engineering agent is grounded in your warehouse, catalog and semantic layer, and it keeps that context across every run.
Datus is the open-source data engineering agent. A chatbot stops at translating a prompt into SQL. The Datus agent orchestrates Catalog, SQL, Pipeline and BI subagents on a shared context engine to plan, run, validate, deploy and monitor real data work end to end.
Without grounded context, any data engineering agent will hallucinate joins and metrics. Datus captures historical SQL, table structures, metrics and semantic definitions, stores them in a dual Tree + Vector memory, and incrementally refines that context from real usage — every run makes the next answer more accurate.
Studio is the free hosted playground in your browser. Data engineers usually start with the CLI for terminal-native workflows. The Chatbot embeds the data engineering agent in Slack or Feishu, and the MCP server plugs Datus context into Claude, Cursor or Windsurf. All four surfaces share the same agent and context engine.
Yes. The open-source Datus data engineering agent is Apache 2.0 and free to self-host — you bring the model and the warehouse. Datus Studio hosts the same agent in the browser as a free playground. Enterprise adds SSO, RBAC, SQL Policy and private / VPC deployment.
Warehouse-native copilots are tied to one platform; open-source NL2SQL projects usually stop at query translation. Datus is a warehouse-agnostic data engineering agent that covers the full lifecycle — SQL, data quality, deployment, monitoring — with a shared context engine, not just a query layer.
Anything that today ping-pongs between SQL editor, dbt project, catalog and BI. An agentic data engineering workflow can draft and review models, run and validate queries, catch schema drift, patch broken pipelines and answer stakeholder questions from the semantic layer — all in one thread, with the data engineering agent keeping context across steps.
Open Datus Studio free in your browser, or self-host the open-source agent on your own warehouse in minutes.
