> For the complete documentation index, see [llms.txt](https://helpcenter.channable.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://helpcenter.channable.com/manage-improve-product-data/generate-with-ai/ai-generated-text/how-to-create-and-use-ai-generated-text.md).

# How to create and use AI generated text

Create, preview, and approve AI-generated product texts using your Gemini or OpenAI connection.

> #### **Who can use this feature**
>
> You need the **Core Pro** plan in your [Channable subscription](/account-billing/manage-your-subscription/channables-pricing-plans.md) to use this feature.

### <i class="fa-memo-circle-check">:memo-circle-check:</i> Before you start

**You'll need**

* [A Gemini connection](/manage-improve-product-data/generate-with-ai/set-up-your-ai-generation-connection/how-to-create-a-gemini-connection-for-ai-generation.md) or [OpenAI connection](/manage-improve-product-data/generate-with-ai/set-up-your-ai-generation-connection/how-to-create-an-openai-connection-for-ai-generation.md) set up in Channable

{% hint style="info" %}
**Note:** You can use a trial Gemini connection to test AI Text Generation if you don't have a Gemini or OpenAI connection yet. A test connection is limited to one generator per project, and you can only generate up to 10 texts.
{% endhint %}

**Good to know**

* Read [AI generation usage and costs](/manage-improve-product-data/generate-with-ai/ai-generation-usage-and-costs.md) to understand token usage, pricing, and data handling.
* **Already have one or more AI text generators set up?** If a generator later in the list uses the same save fields for the same items, it will override the generators that run before it. You can drag and drop to reorder the list.

***

{% stepper %}
{% step %}

### Create your generator

1. Go to **\[your company] > \[your project] > Generate with AI**.
2. Click **Generate with AI**.
3. In the pop-up, select **Text**.
4. At **Text**, select the content you want to create. The text type defines the prompt template used in the **Prompting** step. You can choose from these types:

   * **Titles:** Generate AI text suggestions to improve product titles.
   * **Descriptions:** Generate AI text suggestions to improve product descriptions.
   * **Free text:** Generate AI text suggestions for specific text fields.
   * **Question & answer:** Create short, bulleted selling points that focus on key benefits and features.
   * **Product detail:** Create technical specifications and structured key-value attributes, grouped by section.
   * **Product highlight:** Create frequently asked questions and answers from merchant or customer input.

   <figure><img src="https://501928696-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fu8whzIfip72Kvn3M1bsc%2Fuploads%2FX6bd6meMIum0tVzjdVgj%2FScreenshot%202026-09-21%20at%2015.53.36.png?alt=media&amp;token=cf29ac14-d852-47a0-8017-8f68bc16cc7a" alt=""><figcaption></figcaption></figure>
5. Click **Continue**.
   {% endstep %}

{% step %}

### Configure your generator settings

1. At **Settings**, fill in these details:
   1. At **Name**, add a descriptive name so it's easier to distinguish this generator from others later.
   2. At **AI connection**, you have two options:
      1. If this is your first generator and you don't have a Gemini or OpenAI connection yet, you can continue with a trial (**Trial > Connect**).
      2. If you have a **Gemini** or **OpenAI** account, click either option > **Connect**.
         1. For **Connection type**, select your Gemini/OpenAI connection or create a new one.
         2. Click **Connect with Google Gemini/Connect with OpenAI**.
   3. At **Project fields used to generate attributes**, select the **text fields** that should be used to generate the new texts. Scraping websites or extracting full content directly from URLs is not currently available. Reading images is also not available currently.
   4. At **Field to save the optimized text**, select the project field where you want to save the generated text. We recommend [creating a new field](/get-started/getting-started/fields-in-channable/map-and-add-fields.md) to avoid overwriting your existing title and description data.
   5. At **Output language**, select the language if you want your texts to be translated to. If you leave this at **Do not translate**, the generator will reference the language in the text fields or a prompt when generating.

{% hint style="info" %}
**Note:** Choosing **Do not translate** increases the chance of generating inaccurate information. We recommend selecting an output language for more precise control over your generated texts.
{% endhint %}

2. Click **Save**.

<figure><img src="https://501928696-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fu8whzIfip72Kvn3M1bsc%2Fuploads%2FHPLWjoPOFhBu4IG68HUk%2FScreenshot%202026-09-21%20at%2015.57.32.png?alt=media&amp;token=2c82c56e-809c-4624-b2bc-5e2bef07ebe3" alt=""><figcaption></figcaption></figure>

{% hint style="warning" %}
**Important:** Trial connections can only be used for one generator per project.
{% endhint %}
{% endstep %}

{% step %}

### Select and filter items

1. In the I**tem filtering** step, set up a filter to select which items to generate text for.
   1. To generate text for all items, leave the filter as **If** `all` .
   2. To generate text for specific items, set up the condition part of a rule to filter them out. (For example, **If** `title` contains `t-shirt` .)
2. Click **Save**. Once finished filtering, click **Selected items (%)** to see which items your prompt in the following step will apply to.
3. Click **Save & continue**.

{% hint style="success" %}
**Tip:** This filter functions similarly to [Rule conditions.](https://helpcenter.channable.com/hc/en-us/articles/360019983459-How-to-use-conditions-IF-in-rules)
{% endhint %}

{% hint style="warning" %}
**Important:** To save on token costs, filter your items and use specific prompts for each group. Avoid generating texts for thousands of items at once to prevent high costs and the need for revisions.
{% endhint %}

<figure><img src="https://501928696-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fu8whzIfip72Kvn3M1bsc%2Fuploads%2FZs3SFeVoC1s9mCT9Z8Dh%2FScreenshot%202026-09-21%20at%2015.58.32.png?alt=media&amp;token=b306b965-8adf-42b8-9b97-3dd781893825" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### Apply a prompt

1. At **Prompt**, write your custom prompt in the text field or adjust the template based on your goals or requirements.

{% hint style="info" %}
**Example**

You could change the maximum number of characters based on your channel limitations, or describe the style of the text you want to have.
{% endhint %}

<figure><img src="https://501928696-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fu8whzIfip72Kvn3M1bsc%2Fuploads%2FLycegobmCEhOJnHuHoaB%2FScreenshot%202026-09-21%20at%2016.01.00.png?alt=media&amp;token=a436640b-0451-4cd1-91c9-e7d91a50fc93" alt=""><figcaption></figcaption></figure>

2. Click <i class="fa-sparkles">:sparkles:</i> **Generate 5 previews**. A preview of the items will appear in the **5 previews** section.

<figure><img src="https://501928696-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fu8whzIfip72Kvn3M1bsc%2Fuploads%2FohML6vZTy7KUw11Tbp5c%2FScreenshot%202026-09-21%20at%2016.01.19.png?alt=media&amp;token=271bc841-ba2c-433d-9e86-060ff3bd6e3f" alt=""><figcaption></figcaption></figure>

3. Review the previews and adjust your prompt until you're happy with the results. Once you're satisfied, click <i class="fa-sparkles">:sparkles:</i> **Generate all** to apply the same logic across the items selected in the **Filter items** step.

{% hint style="info" %}
**Note:** Generating previews and regenerating values still contributes to token usage.
{% endhint %}
{% endstep %}

{% step %}

### Review values (approve & regenerate)

1. To review one generated text at a time select either **Approve** or **Regenerate**. You can also edit each text manually if needed.
2. To regenerate all generated texts at once, click **Regenerate all**. Already approved texts won’t be regenerated.
3. To approve all texts at once, click **Approve all**.
4. To unapprove all texts at once, click **Unapprove all**.

The chosen field for the optimized text now contains the approved values and can be sent to any channel. For example, via **Rules** or **Mapping** steps for Google Merchant Center or Amazon.

{% hint style="success" %}
**Tip:** Enable **Approve new items automatically** to accept the generated values for new items as you import them into Channable.
{% endhint %}

<figure><img src="https://501928696-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fu8whzIfip72Kvn3M1bsc%2Fuploads%2FSrn8se9OCz6pQQtaTTE5%2FScreenshot%202026-09-21%20at%2016.02.10.png?alt=media&amp;token=0fff17f7-a1ec-4143-94f5-1ca513039d04" alt=""><figcaption></figcaption></figure>

After generating the texts and before using generated texts in your channels, make sure that there are no empty fields. This could cause errors in the product list with generated values but this could also be due to an issue with Gemini/OpenAI.
{% endstep %}
{% endstepper %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://helpcenter.channable.com/manage-improve-product-data/generate-with-ai/ai-generated-text/how-to-create-and-use-ai-generated-text.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.