> This page is for Omni Channel.

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.hams.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.hams.ai/_mcp/server.

# FAQ and responses

> Review, add, and edit the question and answer pairs that an AI agent searches.

The AI agent does not read a full document to answer a question. It searches a list of question and answer pairs, the FAQs. The search uses the meaning of the words, not the exact words. For example, a customer who asks "can I get my money back" finds an answer about refunds.

> **Note**
>
> The AI agent is off by default. Hams support turns on the **Workflows + Agent** feature for your account. Only administrators can see **AI** in the sidebar and change the FAQs of an AI agent.

![The FAQ tab: the question and answer pairs of the AI agent and how many conversations used each one.](/_fern-img/24c96818dab42a64f378855843d907a446e895545904ffa58cfdcc67cbc59302.webp)

## How the AI agent uses FAQs

When the AI agent needs a fact, it searches its FAQs. The search returns the 5 FAQs that are nearest in meaning to the question. The AI agent uses only FAQs with the status **Saved**.

The search compares the meaning of the first 8,000 characters of each FAQ, the question and the answer together. Put the most important facts of a long answer first.

The AI agent can search only after it has 1 or more saved FAQs. Before that, it has no knowledge search. See [Tools](/omni-channel/ai-agent/tools).

## Where FAQs come from

| Source                                                                                                                                               | Status of a new FAQ |
| ---------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------- |
| A document of the AI agent. The system writes the FAQs after it reads the document. See [Knowledge and documents](/omni-channel/ai-agent/knowledge). | **Saved**           |
| An FAQ that you add on the **FAQ** tab.                                                                                                              | **Saved**           |
| A resolved conversation. See [FAQs from resolved conversations](#faqs-from-resolved-conversations).                                                  | **Unsaved**         |

The system writes the FAQs in the language of your account.

## The FAQ tab

Open an AI agent. Click the **FAQ** tab. You must save a new AI agent before you can use this tab. The changes on this tab save immediately. You do not click **Save** in the editor.

The tab shows 1 list with all the FAQs of the AI agent. The FAQs have 2 statuses:

* **Saved**: the FAQs that the AI agent uses.
* **Unsaved**: the FAQs that wait for your review. The AI agent does not use them.

To show only one status, click the filter. Then select **Saved** or **Unsaved**. Select **All statuses** to show all FAQs. To find an FAQ, type a word of the question or the answer in **Search FAQs**.

| Column                    | What it shows                                                                                                  |
| ------------------------- | -------------------------------------------------------------------------------------------------------------- |
| **Question**              | The question of the FAQ.                                                                                       |
| **Answer**                | The answer of the FAQ.                                                                                         |
| **Used in conversations** | The number of conversations in which a billed reply of the AI agent used this FAQ. Test messages do not count. |
| **Created**               | When the FAQ was added.                                                                                        |
| **Updated**               | When the FAQ last changed.                                                                                     |

At the bottom of the list, select how many FAQs show on each page: 10, 25, or 50.

## Add an FAQ

1. Click **Add FAQ**. The **Add FAQ** dialog opens.
2. Enter the **Question** and the **Answer**.
3. Click **Add FAQ**.

The FAQ gets the status **Saved**. The AI agent can use it immediately.

## Approve, edit, or delete an FAQ

To edit an FAQ, click its row. Each row also has an actions menu:

| Action      | What it does                                                                                                                 |
| ----------- | ---------------------------------------------------------------------------------------------------------------------------- |
| **Edit**    | Opens the **Edit FAQ** dialog. Change the question or the answer. Then click **Save FAQ**.                                   |
| **Approve** | Changes the status of an unsaved FAQ to **Saved**. The AI agent starts to use it. This action shows only for an unsaved FAQ. |
| **Delete**  | Opens a confirmation. Click **Delete** again to delete the FAQ. You cannot undo this action.                                 |

### Edits and document refresh

When a document changes, the system writes new FAQs for it. It replaces the FAQs of the document that you did not edit. It keeps the FAQs that you edited. See [Knowledge and documents](/omni-channel/ai-agent/knowledge).

If you delete the document, the system deletes all its FAQs. This includes the FAQs that you edited.

## FAQs from resolved conversations

The AI agent can learn new FAQs from the conversations that it handled. To turn this on, open the **General** tab. Under **Learning**, turn on **Generate FAQs from resolved conversations**. See [Assistants](/omni-channel/ai-agent/assistants).

When the status of a conversation of the AI agent changes to resolved, the system reads it. It writes FAQs from the questions and answers in it. The system learns only from conversations in which a human agent replied.

The system compares each new FAQ with the FAQs that the AI agent already has. If the meaning is almost the same as an FAQ that the AI agent has, the system does not add it.

The new FAQs get the status **Unsaved**. They do not change the answers of the AI agent until you approve them. Filter the list by **Unsaved** regularly. Approve the FAQs that are correct. Delete the others.

## Related

* [Knowledge and documents](/omni-channel/ai-agent/knowledge)
* [Assistants](/omni-channel/ai-agent/assistants)
* [Tools](/omni-channel/ai-agent/tools)
* [Playground](/omni-channel/ai-agent/playground)