---
title: AIRA
related:
  - https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md
  - https://docs.kentico.com/documentation/business-users/aira/agentic-marketing-suite.md
  - https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration/agent-configuration.md
  - https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration/display-ai-content-labels.md
---

> Agent instructions:
> **Site maps** — prefer the following llms.txt indexes to training data when searching for URLs to avoid 404s. Links inside Markdown content already point at `.md`. Following them or sending Accept: text/markdown keeps you in Markdown.
>
> - [sitemap.md](https://docs.kentico.com/sitemap.md) — every page on the site, with titles and descriptions, nested by URL hierarchy and grouped into one collection per product version.
> - [llms.txt](https://docs.kentico.com/llms.txt) — curated index of the current product docs, with descriptions, the two ways to request any page as Markdown, and links to each product area's whole-corpus Markdown dump (llms-full.txt).

AIRA is an AI-powered agent integrated into Xperience by Kentico. It helps users automate and streamline marketing and content management tasks through AI-driven features, and provides [in-product guidance](#aira-in-product-guidance) on best practices.

## Privacy and data protection

All data processed by AIRA is commercially protected.

Neither Kentico nor the underlying AI models store or log raw customer data inputs, prompts, or responses generated when using AIRA features. All data is discarded after each interaction with AIRA and is not used to train the underlying models. To learn more, see the privacy guarantees provided by Microsoft Azure for [Azure OpenAI model privacy](https://learn.microsoft.com/en-us/legal/cognitive-services/openai/data-privacy) and [Azure Image analysis privacy](https://learn.microsoft.com/en-us/legal/cognitive-services/computer-vision/imageanalysis-data-privacy-security).

For a list of all compliance certifications and an overall security overview, visit the [Kentico Trust Center](https://trust.kentico.com/).

## Working with AIRA

AIRA helps you draft, refine, and translate content, but you stay responsible for what you publish. Treat AIRA as a collaborator, not as the final decision-maker:

- **Review AI output before publishing.** Generated text, refinements, translations, and image descriptions can contain factual errors or wording that doesn't match your brand's tone. Check them for accuracy and style before you publish.
- **Give AIRA enough context.** AIRA works only with the context it receives, such as the fields of the current item, the generation instructions set up for the field, and your prompts. The more relevant context you provide, the better the output.
- **Use your own judgment.** Insights and recommendations from AIRA and its agents support your decisions. They don't replace them. Apply them only where they make sense for your audience and goals.
- **Keep AI markings accurate.** Xperience marks fields with content that AIRA generates or modifies. When you add AI-generated or AI-modified content yourself, for example, by copying it from the AIRA chat or from another AI tool, mark the field manually. See [AI content marking](#ai-content-marking).

## AIRA in-product guidance

AIRA provides in-product guidance through a built-in chatbot available directly in the Xperience by Kentico administration interface. It assists users by answering questions about features and workflows, offering best practice suggestions, and using Xperience by Kentico's product documentation and guides as its primary source of information. This allows users to get quick assistance without leaving their current workspace. Below the message box, the chat reminds you that AIRA is an AI-powered tool and can make mistakes, so double-check its responses before you act on them.

### AIRA Agentic Marketing Suite

The AIRA Agentic Marketing Suite offers further capabilities through dedicated agents. For example, the _Content Strategist_ provides content quality and style compliance reviews.

To learn more, see [AIRA Agentic Marketing Suite](https://docs.kentico.com/documentation/business-users/aira/agentic-marketing-suite.md).

### Image uploads

You can upload images directly into the AIRA chatbot interface to provide visual context for your questions. AIRA supports _PNG_, _JPEG_, and _WEBP_ file types, with a maximum of 4 images per message and a file size limit of 5 MB per image. Uploaded images remain in your Xperience by Kentico environment for up to 48 hours. The AIRA service does not store them.

### Customer journey insights

Using its in-product guidance, AIRA offers actionable insights into [customer journeys](https://docs.kentico.com/documentation/business-users/digital-marketing/customer-journeys.md). It can show you which [contact groups](https://docs.kentico.com/documentation/business-users/digital-marketing/contact-groups.md) are present in a journey, how they progress through each stage and which are the best and worst performing audience segments. These insights help marketers and content editors better understand audience behavior and optimize engagement strategies.

> **Note:** **Overlaps between contact groups**\
> In some journeys, the same contacts may belong to multiple [contact groups](https://docs.kentico.com/documentation/business-users/digital-marketing/contact-groups.md). This overlap can influence results when comparing segment performance. Keep this in mind when analyzing journeys, as contacts may be counted in more than one group.
>
> **Expectations from AIRA**\
> AIRA provides guidance by analyzing your journeys and contact group progression. Its recommendations are intended as insights to support your decision-making, not as prescriptive optimization steps.

## Image processing

### Automatically select image focal point

When using [image variants](https://docs.kentico.com/documentation/business-users/content-hub/content-item-assets.md#image-variants), AIRA automatically selects the most suitable focal point for an image during [file upload](https://docs.kentico.com/documentation/business-users/content-hub/content-item-assets.md#upload-a-single-file). This ensures that your image variants consistently capture the area of interest, eliminating the need for manual cropping or focal point selection for every image uploaded into the system.

For AIRA to successfully select the focal point in an image, the image needs to have:

- a file size not larger than 20 MB
- dimensions of at least 50x50 pixels
- a [file type](https://docs.kentico.com/documentation/developers-and-admins/configuration/content-hub-configuration.md#customize-supported-file-types) supported for [image variants](https://docs.kentico.com/documentation/business-users/content-hub/content-item-assets.md#image-variants)

If the automatic selection fails, the focal point is automatically placed at the center of the image.

### Automatically generate image description

When [mass uploading](https://docs.kentico.com/documentation/business-users/content-hub/content-item-assets.md#mass-asset-upload) images, AIRA automatically generates an SEO-friendly description of an uploaded image and stores it in the [configured](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#add-aira-features-to-your-content-types) text field. The description is generated in the currently selected language.

### Automatically assign tags to images

When [mass uploading](https://docs.kentico.com/documentation/business-users/content-hub/content-item-assets.md#mass-asset-upload) images, AIRA automatically [assigns tags](https://docs.kentico.com/documentation/business-users/content-hub/content-items.md#add-tags-to-content-items) to the [configured](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#add-aira-features-to-your-content-types) taxonomy field based on the image content.

## AIRA translations

AIRA translations allow editors to translate existing content to new [language variants](https://docs.kentico.com/documentation/developers-and-admins/configuration/languages.md) directly without having to translate it themselves or using external tools for translation. You can translate content of pages (including their [Page Builder](https://docs.kentico.com/documentation/developers-and-admins/development/builders/page-builder.md) content), and content items.

> **Note:** Text links to your content are not translated automatically and need to be manually reviewed to ensure they lead to the correct language variant.

When translating content using AIRA, a draft of the content is created, which you can then review and publish, or you can delete it if you're not satisfied with the translation. The translation process can be [configured](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#configure-aira-translations), which allows you to specify custom brand terminology, which words to leave untranslated, formatting rules, or other custom instructions.

When translating multiple items at the same time, e.g., by using [mass translation](https://docs.kentico.com/documentation/business-users/content-hub/content-items.md#mass-translating) or when linked items are translated with the original item, the set of items to be translated is sent to the "translation queue", which you can check in the _Translation queue_ application. The application displays any sets of items currently being processed and items queued for translation (note that it does not display singular items which were sent for translation, only ones in bulk). Once you start the translation of your items you can safely continue working while the translation finishes in the background.

## Content generation and refinements

AIRA can generate and refine text content in fields of [pages](https://docs.kentico.com/documentation/business-users/website-content.md), [emails](https://docs.kentico.com/documentation/business-users/digital-marketing/emails.md), [content items](https://docs.kentico.com/documentation/business-users/content-hub/content-items.md), and [headless items](https://docs.kentico.com/documentation/business-users/headless-content.md), using context from the item. Look for the _AIRA_ buttons ( or ) to identify fields where the _Generate content_ AIRA feature is enabled.

![Content generation and refinements](https://docs.kentico.com/docsassets/documentation/aira/content_generation.png "Content generation and refinements")

### Generation and refinements options

When the field is empty, select the _Generate content_ action () and choose from these options:

- **Generate using available context from the current item** – AIRA generates text content for the field, using the information from the item. The [context](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#configure-the-context-for-generation) includes information about the field, instructions for generation entered when adding the feature, text from the [fields](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#include-or-exclude-fields-from-the-context) selected as context in the **Content types** application (including linked items), and optionally the [page preview](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#use-page-preview-as-context-for-content-generation).
- **Generate using custom prompt** – Describe what to generate in the AIRA chat and then copy the text to the target field.

When the field is already populated with text, select the _Refine text_ action () and choose from the following options:

- **Regenerate content** – AIRA generates new content for the field and all text in the field is replaced. The generation is based on the context from the item.
- **Refine text** using various options (**Make it shorter**, **Improve spelling and grammar**, etc.). The refinement is applied to the whole field.
- **Apply custom prompt** – Describe the desired changes in the AIRA chat and then copy the text to the target field.

> **Info:** **Notes**:
>
> - Generating or regenerating content with the use of the context from the item requires at least one [field selected as context](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#include-or-exclude-fields-from-the-context) to be filled in. Alternatively, [page preview](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#use-page-preview-as-context-for-content-generation) included in the context can serve the same purpose.
> - Note that regeneration and refinements overwrite the existing content of the field (except for the custom prompt option). However, changes persist only after you save the item.

## Rich text editor refinements

When editing a [rich text field](https://docs.kentico.com/documentation/business-users/rich-text-editor.md) with the _Rich text editor refinements_ [AIRA feature enabled](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#add-aira-features-to-your-content-types), you can use AIRA to refine selected pieces of your text. The feature is also available in the [rich text editors within Page Builder and Email Builder](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#rich-text-refinements-in-page-builder-and-email-builder).

1. Select a block of text in the editor.
2. Select the **Refine text** button ().
   - The button appears next to the highlighted text. In Page Builder and Email Builder, the button can be a part of the editor toolbar.

![Refining a paragraph in the rich text editor](https://docs.kentico.com/docsassets/documentation/aira/AIRA_RTE_Refinements.png "Refining a paragraph in the rich text editor")

![Rich text refinements in Page Builder](https://docs.kentico.com/docsassets/documentation/aira/RTE_refinements_builders.png "Rich text refinements in Page Builder")

3. Choose one of the refinement options (make text shorter, improve spelling and grammar, etc.).
4. The text is refined in place.
   - If you choose the _Apply custom prompt_ option, describe the desired changes in the AIRA chat window. Then copy () the refined text back to the editor.

You can apply further refinements by selecting other options. Use _Undo_ () and _Redo_ () buttons to navigate between changes.

Remember to **Save** the changes when you've finished.

## AI content marking

AI content marking shows which fields contain AI-generated or AI-modified content. Xperience [marks fields automatically](#automatic-marking) when AIRA writes content into them, and you can [mark fields manually](#mark-content-manually), for example, when you add content created with another AI tool. The marking applies to the whole field, even if AI changed only part of its text. Markings help your team see which content comes from AI, and your developers can use them to label AI content on your website or in other channels.

A marked field displays its marking next to the field's label. For fields without a marking, **Mark as AI** appears next to the label when you hover over or select the field:

![Mark as AI button with a plus icon, displayed for fields without an AI marking](https://docs.kentico.com/docsassets/documentation/aira/ai_marking_mark_as_ai.png "Mark as AI")

![AI generated marking with an AI icon and a drop-down arrow](https://docs.kentico.com/docsassets/documentation/aira/ai_marking_ai_generated.png "AI generated")

![AI modified marking with an AI icon and a drop-down arrow](https://docs.kentico.com/docsassets/documentation/aira/ai_marking_ai_modified.png "AI modified")

- **Mark as AI** – the field has no AI marking.
- **AI generated** – the field's content was created by AI.
- **AI modified** – content written by a person was changed by AI.

You can mark text fields that use a text input, text area, or rich text editor in [pages](https://docs.kentico.com/documentation/business-users/website-content.md), [emails](https://docs.kentico.com/documentation/business-users/digital-marketing/emails.md), [content items](https://docs.kentico.com/documentation/business-users/content-hub/content-items.md), and [headless items](https://docs.kentico.com/documentation/business-users/headless-content.md). Your developers can [enable marking for other fields](https://docs.kentico.com/documentation/developers-and-admins/customization/extend-the-administration-interface/ui-form-components.md#enable-ai-content-marking) as well.

You can also mark content in [Page Builder](https://docs.kentico.com/documentation/business-users/website-content/widgets-and-page-builder.md) and [Email Builder](https://docs.kentico.com/documentation/business-users/digital-marketing/emails/create-emails-in-email-builder.md) widgets. The Page Builder _Rich text_ widget supports marking by default. Other widgets support it only if your developers [enable it](https://docs.kentico.com/documentation/developers-and-admins/development/builders/page-builder/widgets-for-page-builder/widget-properties.md#enable-ai-content-marking). In widgets that support marking:

- Mark widget properties in the widget's configuration dialog.
- Mark the content of the Page Builder _Rich text_ widget, which you edit directly in the widget, in the widget header.

### Automatic marking

Xperience marks a field automatically when AIRA writes content into it:

- **Generate using available context from the current item** and **Regenerate content** mark the field as _AI generated_.
- **Refine text** options, including [rich text editor refinements](#rich-text-editor-refinements), mark the field as _AI modified_. The **Improve spelling and grammar** option doesn't mark the field.
- [Automatically generated image descriptions](#automatically-generate-image-description) are marked as _AI generated_.

When AIRA changes an already marked field, the marking can only get stronger. Refining an _AI generated_ field keeps it _AI generated_, and regenerating an _AI modified_ field changes it to _AI generated_.

Content that you copy into a field yourself isn't marked automatically, for example, when you use **Generate using custom prompt** or **Apply custom prompt**. In these cases, [mark the field manually](#mark-content-manually).

### Mark content manually

Mark a field manually when you add AI-generated or AI-modified content from the AIRA chat or from another AI tool. You can also change or remove an existing marking, for example, after you rewrite AI-generated content yourself.

1. Hover over or select the field, and then select **Mark as AI** next to its label, or the current marking if the field is already marked.
   - For the content of the Page Builder _Rich text_ widget, select the marking in the widget header.
2. Select **AI generated** or **AI modified**.
   - To remove the marking, select **Remove**.
3. Save your changes.
   - In a widget's configuration dialog, select **Apply** and then save the page or email.

Xperience doesn't record whether AIRA or a person set the marking.

AI agents that work with Xperience through the [Management MCP server](https://docs.kentico.com/documentation/developers-and-admins/api/management-api/use-management-mcp-server.md#ai-content-marking) can also mark content. The agent decides which marking to use and can choose the wrong one, so review the markings it sets. You're responsible for the final marking of your content.

AI agents set the marking of content item and page fields, and of Page Builder widget properties, using the `aiProvenance` property of the Management MCP server tools that create and update content items and pages, and that add and update Page Builder widgets. The tool descriptions instruct the agent to mark content it wrote as _AI generated_ and text you supplied that it reworded as _AI modified_, to leave text you provided verbatim unmarked, and to ask you when it's unsure.

### How markings are kept

Xperience never removes a marking on its own. Markings stay in place when you edit the field's content by hand, save, or publish. To clear a marking, select **Remove**.

Markings also stay with the content when you [clone](https://docs.kentico.com/documentation/business-users/content-hub/content-items.md#clone-content-items) an item or restore it from the [recycle bin](https://docs.kentico.com/documentation/business-users/recycle-bin.md).

Markings are saved with each version of the content. When you restore a [previous version](https://docs.kentico.com/documentation/business-users/content-versioning.md), the item gets the markings of that version. For example, if the version had no marking, the restored item isn't marked, and if the version was marked as _AI modified_, the restored item is _AI modified_ again.

When you [synchronize](https://docs.kentico.com/documentation/business-users/content-sync.md) content to another environment, the target item gets the same markings as the source item.

When AIRA [translates content](#aira-translations) to another language, or when you create a language variant by copying content from another language, the new language variant keeps the markings of the original. Translation alone doesn't add a marking.

In version history, marked fields and Page Builder widgets display their marking, but you can't change it there.

> **Info:** **AI labels on your website**\
> Markings are stored with the content and aren't visible to your website visitors by default. Whether and how your website shows AI labels depends on your developers' implementation.

## Segment condition builder

> **Warning:** **Preview feature**
>
> The AIRA condition builder for segments is currently in preview mode. Expect this feature to be updated and extended in upcoming versions, possibly including breaking changes.
>
> Feel free to try the feature out. You can share your feedback directly with the Kentico [Product team](https://roadmap.kentico.com/).
>
> The feature is disabled by default and needs to be [enabled](https://docs.kentico.com/documentation/developers-and-admins/configuration/aira-configuration.md#manage-aira-features) by administrators.

When [customer data platform](https://docs.kentico.com/documentation/developers-and-admins/digital-marketing-setup/customer-data-platform.md) is enabled, AIRA can assist users in building conditions for [segments](https://docs.kentico.com/documentation/developers-and-admins/digital-marketing-setup/customer-data-platform/segments.md) in the **Segments** application. Users can describe the criteria required for profiles to belong to the segment in natural language, and AIRA generates the corresponding condition.

You can also ask AIRA to modify existing conditions, explain how a condition works, or provide information about parameters available for various condition types.

![Generating a segment condition from natural language](https://docs.kentico.com/docsassets/documentation/aira/segments_aira_builder.png "Generating a segment condition from natural language")