> ## Documentation Index
> Fetch the complete documentation index at: https://docs.omni.co/llms.txt
> Use this file to discover all available pages before exploring further.

# Suggestions

> Apply prioritized recommendations to improve your model's AI context, based on user feedback.

**Suggestions** turn user feedback on AI-generated queries into a prioritized list of recommended improvements to your shared model. For example, a suggestion might add `ai_context` to a field that confused the AI, or refine `synonyms` on a measure users call by another name.

Suggestions live in the [**AI Hub**](/administration/ai-hub). Each suggestion targets one object in your model (a topic, a view, or a field) and includes a rationale plus the chat sessions that motivated it.

## What suggestions can change

Suggestions only recommend changes to **AI-context fields**:

* `ai_context` on the model, a topic, a view, or a field
* `description`, `synonyms`, or `sample_values` on a field

Suggestions never create, rename, or restructure model objects, and they don't touch joins, relationships, or schema.

## Requirements

* **Organization Admin** permissions (required for the AI Hub).
* Omni AI enabled in [**AI Hub > Setup**](/administration/ai-hub) for your organization.
* A **shared model** with at least one prompt that shows signal. Signals include explicit feedback (a thumbs up/down or a comment) and automatically-detected signals in the user's prompt (positive or negative sentiment, frustrated tone, or mild or strong profanity).

## Open the suggestions page

1. Open the **AI Hub** from the left navigation.
2. Select **Suggestions** in the AI Hub sidebar.
3. Choose a shared model from the model selector.

If suggestions have been generated for the model, you'll see a prioritized list of cards. Otherwise, see [Generate suggestions](#generate-suggestions) below.

## Read a suggestion

Each card collapses to a **title** and a **priority badge**. Expand a card to see:

* **Rationale** — why this change would help, written against the feedback patterns that motivated it.
* **Suggested change** — what the AI is recommending, in plain language (for example, *"Set AI Context on Net New ARR in Opportunity"*).
* **Proposed diff** — a preview of the change to your model YAML. The actual edit is drafted by the Modeling Agent and may differ.
* **Sources** — the chat sessions whose feedback drove this suggestion. Click a source to open the chat.

Suggestions are bucketed into three priority tiers:

| Priority | What it means                                                                 |
| -------- | ----------------------------------------------------------------------------- |
| **P1**   | Highest impact. Strong, repeated signal in feedback or a clear quality issue. |
| **P2**   | Medium impact. Useful improvement supported by some feedback.                 |
| **P3**   | Lower impact. Worth considering but less urgent.                              |

Use the **All / P1 / P2 / P3 / Ignored** tabs above the list to filter by priority or view ignored suggestions.

## Generate suggestions

Suggestions are generated on a schedule.

Once you enable a schedule for a model, Omni regenerates suggestions automatically **once per day at 2:00 AM**. The timezone is captured from the browser you used to enable the schedule. The list refreshes the next time you open the AI Hub Suggestions page.

To enable a schedule:

1. Open **AI Hub > Suggestions** and select the model.
2. Click **Schedule AI suggestions…** in the empty state, or open the **⋮ Suggestion options** menu and select it.
3. Confirm the timezone shown in the dialog and click **Enable**.

<Note>
  To change a schedule's timezone, open the dialog and click **Turn off**, then re-enable it from a browser in the desired timezone. To turn scheduling off entirely, open the dialog and click **Turn off**.
</Note>

## Implement a suggestion

To apply a suggestion to your model:

1. Open the suggestion card.
2. Click **Implement in IDE chat**.
3. Omni opens the Model IDE for that shared model, with the suggestion pre-filled in the **Modeling Agent** chat panel.
4. Review the proposed change, then let the Modeling Agent draft the YAML edit.

The Modeling Agent treats the suggestion as a starting point. It may rewrite the suggested text for concision or skip a description that restates the field name. Review the edit before applying, just like any other Modeling Agent change.

Applying a suggestion does not remove it from the list — it stays until the next generation run detects the change and clears it. See [Lifecycle](#lifecycle).

## Ignore a suggestion

If a suggestion isn't relevant, you can ignore it to remove it from the main list. Ignored suggestions are moved to the **Ignored** tab, where you can review or restore them later.

To ignore a suggestion:

1. Open the suggestion card.
2. Click **Ignore** next to the **Implement in IDE chat** button, or open the **⋮** menu and select **Ignore** or **Ignore with reason…**.
3. If you selected **Ignore with reason…**, enter a note explaining why you're ignoring this suggestion, then click **Ignore**.

Ignored suggestions remain in the **Ignored** tab until you restore them. They persist across generation runs.

## Restore an ignored suggestion

To restore a suggestion you previously ignored:

1. Select the **Ignored** tab above the suggestions list.
2. Open the suggestion card.
3. Click the **⋮** menu and select **Restore**.

The suggestion returns to the main list with its original priority.

## Manage suggestions in bulk

To handle several suggestions at once, use select mode:

1. Click **Select** to the right of the tabs.
2. Check the suggestions you want to act on. The **select-all** checkbox in the bulk-action row selects or clears every suggestion in the current tab.
3. Click a bulk action:
   * On an active tab (**All / P1 / P2 / P3**): **Implement** or **Ignore**.
   * On the **Ignored** tab: **Restore**.

A live **N selected** count appears next to **Cancel**. Click **Cancel** to leave select mode without making changes.

Selection is scoped to the active tab:

* Actions apply only to the suggestions visible under the current filter.
* Switching between an active tab and the **Ignored** tab clears your selection.

Select mode closes once an action completes.

Bulk **Implement** combines the selected suggestions into a single Modeling Agent prompt and opens the Model IDE chat with every selected suggestion included.

## Ask about suggestions in the Modeling Agent

You can also browse suggestions conversationally from the Model IDE. Open the [Modeling Agent](/ai/model-assistant) chat panel and ask:

```txt theme={null}
What AI suggestions do we have for this model?
```

The agent fetches the current list, summarizes the highest-priority items, and includes links to the chat sessions that motivated each one.

## Lifecycle

Each time suggestions are generated, Omni reviews the existing list against the current state of your model and the latest feedback:

| Outcome                                                               | What happens                                             |
| --------------------------------------------------------------------- | -------------------------------------------------------- |
| Suggestion is still valid and unchanged                               | Kept as-is                                               |
| Suggestion's recommendation needs to be revised based on new feedback | Updated in place; previous source sessions are preserved |
| Suggestion's target field no longer exists in the model               | Removed                                                  |
| Suggestion is already implemented in the model                        | Removed                                                  |
| Suggestion is ignored                                                 | Remains in the **Ignored** tab until you restore it      |
| New patterns appear in feedback                                       | Added as new suggestions                                 |

Ignored suggestions persist across generation runs and are not automatically removed — they stay in the **Ignored** tab until you restore them.

## Troubleshooting

<AccordionGroup>
  <Accordion title="I don't see a Suggestions tab in the AI Hub.">
    Verify the following:

    * You have **Organization Admin** permissions.
    * Omni AI is enabled in [**AI Hub > Setup**](/administration/ai-hub) for your organization.
  </Accordion>

  <Accordion title="I generated suggestions but the page is empty.">
    Generation runs in the background and typically takes one to three minutes. Reload the page after a minute or two.

    If the page is still empty after several minutes, there may not be any recent prompts that meet the [signal requirements](#requirements).
  </Accordion>

  <Accordion title="I applied a suggestion but it's still in the list.">
    Applied suggestions remain in the list until the next generation run detects the change in your model. The next scheduled run, or a manual regenerate, will clear it.
  </Accordion>

  <Accordion title="A suggestion's target field no longer exists.">
    Suggestions whose target field has been renamed or removed are dropped from the list on the next generation run. You don't need to manage these manually.
  </Accordion>
</AccordionGroup>

## Next steps

* [AI Hub](/administration/ai-hub) — Monitor AI adoption, quality, and credit usage.
* [Modeling Agent](/ai/model-assistant) — Apply suggestions and edit your model from chat.
* [Optimize models for Omni AI](/modeling/develop/ai-optimization) — Add the metadata that makes the AI answer better.