Context Agents Studio
Automate metadata enrichment at scale using AI-powered context agents that generate descriptions, READMEs, and SQL intelligence across your most important data assets.
Most data catalogs fail not because they lack the right tools—but because keeping metadata current is too slow and manual to scale. Assets go undocumented, business terms stay disconnected from technical tables, and analysts spend hours figuring out what data means instead of using it. Context Agents Studio changes that by automating enrichment at scale: AI-powered agents analyze your actual usage signals—query history, lineage, BI activity—to generate descriptions, READMEs, and SQL intelligence across your most important assets. What used to take months of manual curation takes days.
What context agents can do
Generate descriptions
AI analyzes table names, lineage, and query patterns to write business-meaningful descriptions—not generic restatements of column names
Generate READMEs
Produce higher-level documentation for tables, datasets, and BI assets covering purpose, structure, and usage patterns
Surface SQL intelligence
Extract popular joins, common filters, business questions, and foreign key relationships from query history and attach them to the relevant tables and views
Track metadata coverage
Monitor coverage % per attribute across collections—descriptions, READMEs, and SQL Intelligence—to identify and close gaps
Enrich at scale
Trigger enrichment across entire collections with a single click—agents process hundreds or thousands of assets without manual intervention
Generate terms from your docs
Create an agent from a template that reads your wikis and repositories over MCP and turns what's written there into governed context, citing its source every time
Who can use Context Agents Studio
Context Agents Studio is accessible to users with the following roles:
- Admin: Full access to all features
- Governance Admin: Full access to all features—same as Admin for Context Agents Studio purposes
Two ways to build context
Context Agents Studio covers two capabilities. They start from different places and produce different output, so it's worth knowing which one you want. Agent templates are in private preview.
| Question | Collection enrichment | Agent templates |
|---|---|---|
| Where does it start? | An asset collection already in Atlan | A wiki or repository outside Atlan |
| What does it read? | Query history, lineage, and BI activity | Documents, through a connected knowledge source |
| What does it write? | Descriptions, READMEs, and SQL intelligence onto assets you already have | Business terms and categories in a glossary |
| How do you trigger it? | Enrich Now on a collection attribute | A sample run, a full run, or a schedule |
To get started with agent templates, see Understand agent templates.
Learn more
Understand collections
Learn how asset collections are built from usage signals and how coverage is tracked across them.
Understand context agents
Learn about the AI agents that generate descriptions, READMEs, and SQL intelligence.
Enrich metadata
Trigger AI-powered enrichment across your most important asset collections.
Understand agent templates
Learn how an agent reads your existing documentation and turns it into governed business terms.
Have questions? See the Metadata enrichment FAQ.
Context Agents Studio uses credits for enrichment actions. See Credit usage for per-agent costs and how to track consumption.