
## Atlan AI

URL: https://docs.atlan.com/product/capabilities/atlan-ai/concepts/what-is-atlan-ai

Atlan AI is the intelligence layer of the Enterprise Context Layer: agents that build, maintain, and activate enterprise context at scale.

:::info **Did you know?** 
Some capabilities shown here may require additional enablement or licensing. Contact your Atlan representative for details.
:::

Atlan AI works in two directions. On the enrichment side, Context Agents read evidence from your Enterprise Data Graph (lineage, SQL patterns, usage signals, business definitions) and generate structured context: descriptions, term linkages, glossary overviews, completeness diagnostics, and data quality rules. On the activation side, the [Remote MCP server](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt) exposes that enriched context to any downstream AI tool—Claude, ChatGPT, Cursor, Gemini, Codex, Snowflake Cortex, n8n, and more.

Atlan AI runs on a multi-model architecture routed through a centralized AI gateway supporting multiple LLM providers.

:::info important!
Atlan doesn't send any data to the AI service and only uses metadata for supported capabilities. For questions about data security, see [Atlan AI security](https://docs.atlan.com/llms/governance/atlan-ai/security/llms.txt).
:::

## Enable Atlan AI

Only [admin users](https://docs.atlan.com/llms/governance/access-control/delegate-administration/llms.txt) in Atlan can enable Atlan AI for their organization. Once enabled, each user's existing [permissions](https://docs.atlan.com/llms/governance/access-control/delegate-administration/llms.txt) and [access policies](https://docs.atlan.com/llms/governance/access-control/what-are-purposes/llms.txt) in Atlan determine how they can use Atlan AI. For example, a user must have the permission to edit metadata to use Atlan AI for updating asset descriptions.

To enable Atlan AI for your Atlan users:

1. In your Atlan workspace, click **Settings** from the left menu.
 * If you are using the **Old UI (Classic)**, from the left menu, click **Admin**.
2. Click **Labs**.
3. On the _Labs_ page, under the _Atlan AI_ heading:
 - Turn on **Enrich metadata** to enable your users to use Atlan AI for [documenting assets](#use-atlan-ai-to-document-assets), [explaining lineage transformations](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-lineage-analysis/llms.txt), and [generating aliases](https://docs.atlan.com/llms/catalog/discovery/add-an-alias/llms.txt).

4. (Optional) For _Customize Atlan AI_, click the **Add instructions** button to make Atlan AI suggestions more relevant to your organization. In the _Enhance suggestions_ dialog, for _General instructions_, describe your organization and add details about your product, mission, and more, and then click **Save**.

 These instructions shape enrichment suggestions, such as generated descriptions and READMEs. To shape how conversational AI answers questions in chat, use the separate _Customize Conversational AI_ setting instead—see [Customize conversational AI](https://docs.atlan.com/llms/governance/conversational-ai/customize-conversational-ai/llms.txt).

You have now unleashed the power of Atlan AI for your users! 🎉

If you'd like to disable Atlan AI from your organization's Atlan workspace, follow the same steps to turn it off.

:::info **Did you know?** 
 If [Atlan AI is disabled](#enable-atlan-ai), the feature is no longer available in your workspace. However, any descriptions previously generated by Atlan AI and added to your assets or saved SQL queries are still available.
:::

## Current capabilities

Atlan AI builds context from evidence in your Enterprise Data Graph and makes it available to any downstream AI tool. Enrichment capabilities include AI-generated descriptions, lineage transformation explanations, glossary intelligence (AI-generated overviews, completeness diagnostics, contextual gap-fixing), and data quality rules. Activation capabilities include the Remote MCP server that makes all of this callable by Claude, ChatGPT, Cursor, Gemini, Codex, Snowflake Cortex, and any other MCP-aware agent or automation framework.

### Use Atlan AI to document assets

Atlan AI puts you in control of your data estate in Atlan, helping you curate meaningful context for your data assets. Accept, reject, or edit any AI-powered suggestions, the choice is _yours_.

You can use Atlan AI to:

- [Document tables and views with AI-generated descriptions](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-documentation/llms.txt)
- [Document columns with AI-generated descriptions](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-documentation/llms.txt)
- [Document terms and categories with AI-generated descriptions](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-documentation/llms.txt)
- [Document terms with AI-generated READMEs](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-documentation/llms.txt)
- [Add an Atlan AI-generated alias to supported assets](https://docs.atlan.com/llms/catalog/discovery/add-an-alias/llms.txt)

:::info **Did you know?** 
 For full transparency, any changes made using Atlan AI are marked as _Updated using Atlan AI_ in the [activity log](https://docs.atlan.com/llms/catalog/discovery/faq/llms.txt).
:::

### Use Atlan AI for glossary intelligence

Beyond authoring individual descriptions, Atlan AI acts as a completeness and comprehension layer for your entire business glossary. Every glossary, category, and term carries an AI-generated overview. Editors see aggregate completeness diagnostics—missing descriptions, missing READMEs, draft-status terms—and can fix detected gaps with contextual AI assistance directly in place.

You can use Atlan AI to:

- [View AI-generated overviews on glossaries, categories, and terms](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-glossary-management/llms.txt)
- [Identify AI-generated glossary content via the AI icon](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-glossary-management/llms.txt)
- [Get completeness diagnostics for editors](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-glossary-management/llms.txt)
- [Fix detected gaps with contextual AI assistance](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-glossary-management/llms.txt)
- [Explore your glossary with conversational search](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-glossary-management/llms.txt) (Conversational AI Private Preview)

### Use Atlan AI for lineage analysis

Atlan AI can help you understand [lineage transformations](https://docs.atlan.com/llms/catalog/lineage/what-are-processes/llms.txt) using natural language. You can use Atlan AI to create a natural language explanation for assets with SQL attributes and help you better understand the transformation logic.

You can use Atlan AI to:

- [Explain lineage transformations](https://docs.atlan.com/llms/governance/atlan-ai/use-atlan-ai-for-lineage-analysis/llms.txt)

### Use Atlan AI for data quality

Atlan AI can automatically suggest data quality rules based on your asset's metadata structure. This helps you establish quality checks quickly without manually creating each rule.

You can use Atlan AI to:

- [Generate AI-suggested data quality rules](https://docs.atlan.com/llms/governance/data-quality/use-ai-suggested-rules/llms.txt)
- [Generate custom SQL rules with Copilot](https://docs.atlan.com/llms/governance/data-quality/use-ai-suggested-rules/llms.txt)

### Activate context for AI agents

The context Atlan AI builds doesn't have to stay inside Atlan. The [Remote MCP server](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt) exposes Atlan's Enterprise Data Graph as a callable resource for any MCP-aware AI agent or automation framework. Connect Claude, ChatGPT, Cursor, Gemini, Codex, Snowflake Cortex, n8n, or any agent you build—they can search assets, traverse lineage, read business definitions, apply tags, run data quality checks, and update metadata through the same tools Atlan uses internally.

You can use the Remote MCP server to:

- [Connect Claude Code to Atlan](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt)
- [Connect Claude Desktop to Atlan](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt)
- [Connect ChatGPT to Atlan](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt)
- [Connect Cursor to Atlan](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt)
- [Connect Snowflake Cortex to Atlan](https://docs.atlan.com/llms/governance/atlan-mcp-overview/remote-mcp-overview/llms.txt)
- [Automate metadata operations with Python, n8n, or LangChain](https://docs.atlan.com/llms/governance/atlan-mcp-overview/agent-automation-use-cases/llms.txt)

---
