> For the complete documentation index, see [llms.txt](https://comunity.gitbook.io/learning.comunitynexus/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://comunity.gitbook.io/learning.comunitynexus/26.2/toolkit-guides/intelligence/set-up-a-provider-and-deploy-a-model.md).

# Set up a provider and deploy a model

A provider is the Azure AI Foundry infrastructure your project's agents run on. Creating one provisions that infrastructure. Everything else in Intelligence belongs to a provider: models are deployed under it, and tools, agents and knowledge are built on those models.

## Before you start

* **Sign in as an Organisational Administrator.** Intelligence is not available to the Developer, Lead Developer, Operations or Viewer roles, which do not see the menu at all. See [Roles and Permissions](/learning.comunitynexus/26.2/getting-started/organisations/roles-and-permissions.md).
* **Know which environment you are working in.** The AI Provider Manager is organised by environment: you select an environment, and then configure that environment's models, tools and agents. See [Environments](/learning.comunitynexus/26.2/getting-started/manage-your-project/deploy/environments.md).

{% hint style="warning" %}
A provider and an available model are the prerequisite for everything else. An agent can be created without tools, and used without a trigger, but it cannot run without a model.
{% endhint %}

## Providers are organised by environment

The Providers screen is titled **AI Provider Manager**, and it asks you to select an environment in order to configure that environment's models, tools and agents. A provider therefore belongs to an environment rather than to the project as a whole, in the same way that most project configuration is environment-specific.

Until a provider exists the screen reports that none are configured, and invites you to create your first AI provider to start syncing models, registering tools and creating agents.

## Create a provider

1. Open your project and select **Intelligence (Preview)**, then **Providers**.
2. Click the  **Create Foundry Provider** button.
3. Complete the **Create Azure Foundry Provider** form:<br>

   <figure><img src="/files/TYt64zGYwXVnXleByWKx" alt=""><figcaption><p>Create Azure Foundry Form</p></figcaption></figure>

| Field             | Description                                                                                                                                 |
| ----------------- | ------------------------------------------------------------------------------------------------------------------------------------------- |
| **Project Name**  | Shown for confirmation and taken from your project. It cannot be changed here.                                                              |
| **Provider Name** | Required. The name for this Foundry provider, for example `prod-foundry` or `gpt-provider`. It also forms part of the Azure resource names. |
| **Azure Region**  | The region the Foundry resources are created in. Defaults to **East US**.                                                                   |

4. Read the confirmation line, which names the resources that will be created.
5. Select **Create Provider,** wait for the platform to connect with Azure to create a provider and display its details as shown in the screenshot below:<br>

   <figure><img src="/files/vvEtTZfKGcEbR3bM3wUB" alt=""><figcaption><p>AI Provider Manager</p></figcaption></figure>

### What creating a provider creates

The Toolkit creates the Azure AI Foundry project for you and configures it as an AI provider. From your project name and the provider name it provisions:

* a Foundry Account, named `<project>-<provider>-account`
* a Foundry Project, named `<project>-<provider>-project`
* the AI Provider record the rest of Intelligence hangs from

These are real Azure resources, created in the region chosen on the form, so the Azure subscription behind your platform installation has to permit their creation.

The provider is reached at a Cognitive Services endpoint formed from the same two names with their separators removed. A provider named `jac_test_provider` in a project named `gothamcity1` is reached at `https://gothamcity1jactestprovideraccount.cognitiveservices.azure.com/`.

{% hint style="info" %}
The provider name is not only a label, because it is used to name the Azure resources and forms part of the endpoint address. A project can hold more than one provider, which keeps separate sets of agents on separate infrastructure. Where more than one exists, one is marked **Default**.
{% endhint %}

## What the provider shows

Each provider is listed as a card, and the heading above it records when the provider was last validated.

| Element                           | Reports                                                       |
| --------------------------------- | ------------------------------------------------------------- |
| **Connected**                     | Whether the Toolkit can reach the provider.                   |
| **Models**, **Tools**, **Agents** | How much has been configured against this provider.           |
| **Enabled**                       | Whether the provider is in use.                               |
| **Service Health**                | The health of the underlying Cognitive Services resource.     |
| **Last sync**                     | When models were last synced from the provider, or **Never**. |

{% hint style="info" %}
**Connected and Service Health are different things.** A provider can report **Connected** while Service Health reads **Unknown**, with a note that the health of the Cognitive Services resource could not be determined. Unknown health is not the same as an unreachable provider, and on its own it does not mean anything is wrong.
{% endhint %}

The card also carries its actions: **Sync Models**, **Test Connection**, and icons to edit or delete the provider. **Test Connection** re-checks the connection and updates the last validated time.

## Bring in the models

Models are brought into the Toolkit from the provider by selecting **Sync Models**. Until that has been done the card reports `Last sync: Never` and no models are available to agents, which is why a newly created provider shows `Models: 0`.

A deployment records the following.

| Setting               | What it governs                                                                                                                |
| --------------------- | ------------------------------------------------------------------------------------------------------------------------------ |
| **Model**             | Which model the deployment provides.                                                                                           |
| **Capacity**          | The throughput reserved for the deployment.                                                                                    |
| **Provisioning type** | How that capacity is held.                                                                                                     |
| **Guardrails**        | The content filters applied to the deployment, including content safety, jailbreak detection and protected material detection. |

All of these remain editable once the deployment exists, so capacity and guardrails can be revised without redeploying.

## Where to go next

* [Create a tool](/learning.comunitynexus/26.2/toolkit-guides/intelligence/create-a-tool.md): give an agent something to act with
* [Create an agent](/learning.comunitynexus/26.2/toolkit-guides/intelligence/create-an-agent.md): the role that runs on this model