
This article describes the [permissions](/docs/settings/identity-access-management/permissions) needed to use the [AI Hub](/docs/ai-hub).

To set the permissions, you need to create a user role or edit an existing one and open the Permissions matrix. In the matrix, the permissions are collected into groups. Some of these groups can be expanded to set more granular permissions.

## How to read this list

In this article, each heading describes an action. The permissions for that action are described in the form of breadcrumbs.

**For example**, if the required permissions are:
- **Profiles** > **Client list**: `read`
- **Search engine**: `create`, `edit`

The permission matrix needs to look like this:
<figure><img src="/api/docs/image/54176ad07f146575310749eba44b7c2f42c1b327/docs/settings/_gfx/permissions-example.png" class="large" alt="Permissions example"><figcaption>Permission example, see description above figure.</figcaption></figure>

The `read` permission is not explicitly listed when any higher permission is required - the portal enables it automatically.

## I want to...

### work with AI Recommendations

<div class="admonition admonition-note"><div class="admonition-icon"><svg xmlns="http://www.w3.org/2000/svg" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="2.5"><path stroke-linecap="round" stroke-linejoin="round" d="M13 16h-1v-4h-1m1-4h.01M21 12a9 9 0 11-18 0 9 9 0 0118 0z" /></svg></div><div class="admonition-body"><div class="admonition-content">

For permissions needed to configure the AI engine, see ["Settings permissions"](/docs/settings/identity-access-management/permissions/settings-permissions#configure-ai-engine).

</div></div></div>


#### see Recommendations and statistics

- **Communications** > **Recommendations**: `read`
- **Assets** > **Catalogs**: `read`

#### preview a Recommendation
- **Profiles**: `read` - to search or select a profile from the list of all profiles
- **Analytics** > **Segmentations**: `read` - to select a profile from a segmentation

#### see Recommendation details, create and edit a recommendation
- **Communications** > **Recommendations**: `create`, `edit`
- **Settings** > **AI engine configuration**: `read`
- **Analytics** > **Aggregates**, **Expressions**: `read`
- **Search engine**: `read`

This also enables saving and activating a Recommendation.

#### duplicate a Recommendation
**Communications** > **Recommendations**: `create`, `edit`

#### delete a Recommendation
**Communications** > **Recommendations**: `delete`

### work with Recommendation A/B/X tests

- Recommendation permissions as described above
- **Communications** > **Test optimizer**:
    - `read` - to see the tests
    - `create`, `edit` - to create and edit tests
    - `execute` - to run tests

### work with promotions

#### see promotions

- **Communications** > **Promotions**: `read`
- **Analytics** > **Segmentations**, **Expressions**, **Aggregates**: `read`

#### create, update and duplicate promotions
- **Communications** > **Promotions**: `create`, `edit`
- **Assets** > **Tags**: `read`
- **Assets** > **Catalogs**: `read`

#### publish and hide promotions
**Communications** > **Promotions**: `create`, `edit`

#### delete promotions
**Communications** > **Promotions**: `delete`

#### import promotions
See ["Import data" in "Data Management permissions"](/docs/settings/identity-access-management/permissions/data-management-permissions#import-data).

### work with personalized promotions

#### see personalized promotions
**Communications** > **Personalized promotions**: `read`

#### create, edit, and duplicate personalized promotions
- **Communications** > **Personalized promotions**: `create`, `edit`
- **Assets** > **Catalogs**: `read`

#### publish and unpublish personalized promotions
**Communications** > **Personalized promotions**: `create`, `edit`

#### delete personalized promotions
**Communications** > **Personalized promotions**: `delete`

### access AI search

This allows you to see the indexes and suggestion indexes.


<div class="admonition admonition-note"><div class="admonition-icon"><svg xmlns="http://www.w3.org/2000/svg" fill="none" viewBox="0 0 24 24" stroke="currentColor" stroke-width="2.5"><path stroke-linecap="round" stroke-linejoin="round" d="M13 16h-1v-4h-1m1-4h.01M21 12a9 9 0 11-18 0 9 9 0 0118 0z" /></svg></div><div class="admonition-body"><div class="admonition-content">

For permissions needed to configure the AI engine, see ["Settings permissions"](/docs/settings/identity-access-management/permissions/settings-permissions#configure-ai-engine).

</div></div></div>


- **Search engine**: `read`
- **Settings** > **AI engine configuration**: `read`
- **Assets** > **Catalogs**: `read`
- For A/B tests: **Communications** > **Test optimizer**: `read`

### edit indexes, suggestion indexes, and A/B tests
This includes managing synonyms and query rules.

- **Search engine**: `edit`
- **Settings** > **AI engine configuration**: `edit`
- For A/B tests: **Communications** > **Test optimizer**: `edit`
- For query rules: **Analytics** > **Segmentations**: `read`

### create indexes, suggestion indexes, and A/B tests
This includes managing synonyms and query rules.
- **Search engine**: `create`
- **Settings** > **AI engine configuration**: `create`
- For A/B tests: **Communications** > **Test optimizer**: `create`
- For query rules: **Analytics** > **Segmentations**: `read`

### preview search/suggestion results
**Profiles** > **Client list**: `read` - to select a preview context profile

### delete indexes, suggestion indexes, and A/B tests
This includes managing synonyms and query rules.
- **Search engine**: `delete`
- **Settings** > **AI engine configuration**: `delete`
- For A/B tests: **Communications** > T**est optimizer**: `delete`

### access Predictions
This allows you to see the predictions:

- **Predictions**: `read`
- **Settings** > **Predictions**: `read`
- **Settings** > **AI engine configuration**: `read`
- **Search engine**: `read`
- **Assets** > **Catalogs**: `read`

### create and edit predictions
- **Predictions**: `create`, `edit`
- **Analytics** > **Expressions**: `read`

### calculate, deactivate, and duplicate predictions
**Predictions**: `create`, `edit`

### clone a prediction to another workspace

In the source and target workspace, you need:
- **Cloning**: `create` 
- all permissions for the cloned prediction and all nested objects that will also be cloned in the process.

To learn more about cloning, see [Cloning objects to other workspaces](/docs/settings/workspace/cloning-objects).