> For the complete documentation index, see [llms.txt](https://codemash.gitbook.io/codemash/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://codemash.gitbook.io/codemash/cloud/ai.md).

# AI

Cloud → AI — connect language models (LLM) and MCP servers to your project, and use them in AI chat.

The AI screens connect AI services to your project. An **LLM integration** is a language model provider (OpenAI, Anthropic, Ollama, …) that your project's AI features run on. An **MCP integration** is an MCP server (MCP = Model Context Protocol, a standard way to give a model tools) — web search, GitHub, Stripe, a browser — whose tools the AI chat may call. You use both for the AI chat your end users get in the Admin Portal, and for the AI chat API.

## Enable the module

AI has no switch of its own: the **Integrations** item is always in the **Project** menu, with two tabs, **LLM** and **MCP**. You need the `ai:read` permission to see it and `ai:create` to add an integration (the **AI — full access** policy has both).

<figure><img src="https://712692422-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FHy372bqsaex2cqS3ikhL%2Fuploads%2Fgit-blob-4caf65e83f931959f7507f9265071f58ffa3506d%2Fintegrations-list-1.png?alt=media" alt="Project menu with Integrations selected; the LLM tab lists Support assistant (OpenAI, DEFAULT), Report writer (Anthropic), Local Llama (Ollama), Fast drafts (Groq), Gemini Flash, Mistral Large, Many models (OpenRouter) and Grok"><figcaption><p>Project → Integrations, LLM tab.</p></figcaption></figure>

What the AI does for your end users (assistants, the default LLM, plans, the **Show AI chat** switch) is set on [Project → Settings → AI](/codemash/cloud/project/settings/ai.md).

## Module map

| Screen                                               | What it is for                                                                                                                                                                                      |
| ---------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| [Integrations](/codemash/cloud/ai/integrations.md)   | The two tabs, LLM and MCP, and what they share.                                                                                                                                                     |
| [LLM integrations](/codemash/cloud/ai/llm.md)        | The language models of the project — list, [test](/codemash/cloud/ai/llm/test.md), [set as default](/codemash/cloud/ai/llm/set-default.md), [disable or delete](/codemash/cloud/ai/llm/disable.md). |
| [LLM providers](/codemash/cloud/ai/llm/providers.md) | One page per provider: Norbix model, OpenAI, Anthropic, Ollama, Groq, Google, Mistral, OpenRouter, Grok.                                                                                            |
| [MCP integrations](/codemash/cloud/ai/mcp.md)        | The MCP servers of the project — list, [test](/codemash/cloud/ai/mcp/test.md), [default](/codemash/cloud/ai/mcp/set-default.md), [disable or delete](/codemash/cloud/ai/mcp/disable.md).            |
| [MCP providers](/codemash/cloud/ai/mcp/providers.md) | One page per server: Brave Search, GitHub, MongoDB, Obsidian, Playwright, Stripe.                                                                                                                   |
| [AI chat](/codemash/cloud/ai/ai-chat.md)             | The chat itself: the Cloud (Hub) chat and the end-user (API) chat.                                                                                                                                  |

## Your first AI integration

1. [Connect an LLM provider](/codemash/cloud/ai/llm/providers.md) — for example [OpenAI](/codemash/cloud/ai/llm/providers/openai.md) with your own API key. The first LLM of an environment becomes its default.
2. [Test it](/codemash/cloud/ai/llm/test.md) — Norbix checks that the key is stored.
3. Optionally [connect an MCP server](/codemash/cloud/ai/mcp/providers.md), for example [Stripe](/codemash/cloud/ai/mcp/providers/stripe.md), so the chat can use its tools.
4. Pick the [default LLM](/codemash/cloud/ai/llm/set-default.md) and the assistants on [Project → Settings → AI](/codemash/cloud/project/settings/ai.md), then [use the AI chat](/codemash/cloud/ai/ai-chat.md).

## Use it from code

The examples on the AI pages use this client. The key is an AI service-user key or a signed-in session (see [Keys and environments](/codemash/get-started/keys-and-environments.md)).

```ts
import { Norbix } from '@norbix.ai/ts';

const projectId = 'pr_1EbxZuxAcCIW9tfBoDfGri';
const norbix = new Norbix({ projectId }); // the key comes from NORBIX_API_KEY

const llms = await norbix.hub.ai.getLlmIntegrations();
const mcps = await norbix.hub.ai.getMcpIntegrations();
console.log(llms.list.items.map((i) => `${i.integrationName} (${i.llmProvider})`));
console.log(mcps.list.items.map((i) => `${i.integrationName} (${i.mcpProvider}, ${i.transport})`));
```

```sh
norbix hub ai llm integrations get
```

The shorter `norbix ai llms` and `norbix ai mcps` commands list the same integrations. The client and the CLI send the project id with every call, so the requests do not repeat it.

## Related

[Integrations](/codemash/cloud/concepts/integrations.md) · [AI in your products](/codemash/ai/products.md) · [AI for developers](/codemash/ai/developers.md) · [SDK](/codemash/sdks/ai.md) · [CLI](/codemash/cli/cli.md) · [API](/codemash/api-reference/ai.md)