> For the complete documentation index, see [llms.txt](https://v2.dataos.info/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://v2.dataos.info/consume/consume-with-ai/overview.md).

# Consume with AI

Use AI assistants and agents with Data Product MCP to discover, evaluate, and query governed data.

Data Product MCP connects Model Context Protocol (MCP)-compatible AI assistants and agentic frameworks to your Data Products. Once connected, you can discover products, ask business questions, inspect health and lineage, and find the right owner, all in natural language. This is the AI mode of the consumer journey: the same discover, evaluate, and activate steps run through conversation.

To start, pick the tool you already work in. From a product page, select **Activate → MCP** to open **Connect with MCP**, then choose a client or framework.

## How it works

Data Product MCP exposes products to AI tools through this protocol. The assistant does not get an unrestricted database connection. It calls Data Product MCP tools, and those tools route through Data Product APIs, authorization, semantic definitions, quality checks, lineage, and ownership metadata. That gives consumers one governed entry point:

<table><thead><tr><th width="138.78173828125">Need</th><th>What Data Product MCP does</th></tr></thead><tbody><tr><td>Find</td><td>Locate products by topic, domain, owner, metric, or business question</td></tr><tr><td>Trust</td><td>Review semantic fields, quality, run history, lineage, profile, limitations, and owner</td></tr><tr><td>Ask</td><td>Query governed metrics and dimensions through the semantic layer</td></tr><tr><td>Act</td><td>Use failure details and owner context to follow up with the right team</td></tr></tbody></table>

## Governance and trust

Data Product MCP inherits product governance. Every tool call now runs through the same policy enforcement point (PEP) as the rest of the platform, in addition to inheriting the upstream governance of the products it queries. Every request carries the user's own credentials, and DataOS authorizes it before returning data. The assistant sees only the products, tables, columns, and values the user is allowed to access; a masked column comes back masked, and an inaccessible product never appears in discovery, lineage, profile, or query responses.

Access is configured the same way as any other platform service, through standard policy grants: the `mcp-tool-read` grant under the `use-mcp-service` policy use case, or a tenant role such as `data_consumer` or `data_developer` on any tenant on the instance. If you hit authentication errors after setup, ask your administrator for one of these. See the [AI activation reference](https://v2.dataos.info/references/interfaces/ai-activation) for the full trust model.

The assistant does not invent data. It translates the question into structured tool calls and returns responses that cite the metric, model, quality, lineage, run, or owner. For analytical answers it names the metric and the source product; when a question cannot be answered from the available semantic surface, it returns a clear limitation instead of fabricating a measure. Query responses also state when row output was capped by the platform's row limit, so you can tell a preview from a full result set.

## Get connected

* [Connect a client](https://v2.dataos.info/consume/consume-with-ai/connect-a-client.md): Claude, Cursor, Copilot in VS Code, ChatGPT Desktop (Codex), or a custom MCP client.
* [Connect an agentic framework](https://v2.dataos.info/consume/consume-with-ai/connect-an-agentic-framework.md): LangChain, Vercel AI SDK, AgentScope, or a custom framework.

Then use the AI mode of each journey step:

* [Discover in natural language](https://v2.dataos.info/consume/consume-with-ai/discover-in-natural-language.md): find relevant products by describing what you need.
* [Understand in natural language](https://v2.dataos.info/consume/consume-with-ai/understand-in-natural-language.md): inspect schema, quality, lineage, profiles, and runs through conversation.

For how Data Product MCP works and its trust boundaries, see the [AI activation reference](https://v2.dataos.info/references/interfaces/ai-activation) in References.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://v2.dataos.info/consume/consume-with-ai/overview.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
