The dotAI feature adds four Workflow Sub-Actions, or actionlets, to dotCMS. Each is considered separately in its own subsection below.

For all asynchronous functions, it is recommended to save all user-generated content and exit the contentlet editor while the write action occurs in the background; the asynchronous process will save when complete, which can result in the loss of unsaved user-generated content.

AI Auto-Tag Content#

---

This actionlet will attempt to Auto-tag your contentlet based on its values — automatically addressing itself to a Content Type's Tag field, when present. It takes five parameters:

AI Content Prompt#

---

This actionlet will generate text content. You may specify both a prompt, which can itself contain Velocity code referencing content data via the $dotContentMap built-in content object.

If the prompt specifies that the AI should return a JSON object, be sure to specify the keys of the relevant object properties; their associated values will be sent to the corresponding fields by variable name.

AI Embeddings#

---

This actionlet gives more precise control over the "embeddings" of content. Embeddings are collections of numerical measurements that signify the relatedness of text strings, which are crucial for a variety of AI operations — in particular, for forming semantic search indexes.

If fields are not specified, dotCMS will attempt to guess which fields should be used for indexing operations through the following guidelines:

The AI Embeddings actionlet has three parameters:

AI Generate Image#

---

This actionlet allows generation of images based on a dynamically defined prompt. As with the prompt in AI Content Prompt, the prompt may contain references to content via the $dotContentMap built-in content object.

The actionlet's parameters are as follows:

AI Image Auto-Tagging and Descriptions#

---

The Image Auto-Tagging and Descriptions sub-action examines an image and provides metadata corresponding to its content — a prose description for use as alt text, and set of tags.

This actionlet has no parameters, but looks for two pre-defined field variables that determine which fields should receive the tags or description, and which image should be read as the source of those values.

The pattern is the same for both field variables:

1. In the content type editor, select the field that should receive the generated tags or alt text.
2. Set a field variable key equal to dotAITagSrc on a tag field, or dotAIDescriptionSrc on a text field.
3. Set the corresponding value to the variable name of the field that contains the image — e.g., in a standard dotAsset, the value would be asset.

In addition to descriptive tags, the tag generation will also include the dot:taggedbydotai tag, which prevents further tag population by this process.

Triggering#

There are two ways to trigger the actionlet:

1. Standard workflow approach: Firing a workflow action that includes the AI Auto-Tagging and Descriptions sub-action. This allows the usual range of uses, including firing an action from the content search menu.
2. On Publish: Even if the actionlet is not present, the Publish action will still trigger this actionlet if the appropriate field variables are defined within the content type.

The firing behavior can also be inhibited:

• Alt text will not be generated if there is already text in the target field.
• Similarly, tags will not be generated if the tag field contains the dot:taggedbydotai tag.

Configuration#

The vision model this actionlet uses is configured separately from the default dotAI configuration, via configuration properties.

AI Translate Content#

---

The AI Translate Content actionlet can use OpenAI to perform translations. You can add this actionlet to any workflow, though an asynchronous workflow is recommended, as the process can take time to complete.

By default, the actionlet will translate your Text, WYSIWYG, Textarea, and Block Editor fields. The actionlet's parameters allow you to specify which types of fields to include.

The default prompts are intended to prevent OpenAI's response from corrupting any HTML- or JSON-based tags found in WYSIWYG and StoryBlock fields.

The full list of parameters:

These parameters allow specifying field types or individual fields for auto-translation or omission.

You can also indicate language variables, in the final parameter. This effectively serves as a lookup table to assist in translation. This is useful when doing domain-specific translations that expect terms to be translated in an exact or non-standard ways. Specify a language key prefix to use; this will load the language property variables that begin with that string — e.g. translation.context. Only variables whose key starts with that prefix — e.g. translation.context.cms and translation.context.content.management — will be included in the lookup table.

Configuration#

The translation actionlet is configured separately from the default dotAI app configuration. Edit the following via configuration properties.

When configuring a new system or user prompt, there are some specific values that should be noted and included.

The system prompt has the following default value:

You are a domain specific translation engine that translates content from ${sourceLanguage} to ${targetLanguage}.#if($translationKeys)\n\nThe json object below contains a KEY/VALUEs of important domain specific terms and their translations and whenever the KEYs are found in the source, use these VALUEs as the translation in a way that makes sense:\n\n-----------------\n\n${translationKeys}\n\n-----------------\n\n#end

As you can see, it includes Velocity syntax, indicating variables. These variables refer to values specified in the workflow parameters. When multiple languages are specified, individual translations will occur in sequence, over multiple calls; as such, ${sourceLanguage} and ${targetLanguage} will refer to one language at a time.

The system prompt is where the lookup table is created via the Language Variable Prefix parameter, as indicated by the ${translationKeys} variable.

Similarly, this is the default user prompt:

Translate the values of the following source json object which is in ${sourceLanguage}.  Return your answer as a JSON object with the same KEYS as the source JSON below but with the values translated into ${targetLanguage}.  Rule 1.  If the VALUE in the source json object is html, only translate the text and not the tags and do NOT escape the html tags with back slashes.  \n Rule 2. if the VALUE is a nested json object - only translate the nested json keys called \`text\`.\n\nThis is the source JSON object to translate:\n\n-----------------\n\n${sourceJson}\n\n-----------------\n\n

Once again, note the use of variables:

• ${sourceLanguage}
• ${targetLanguage}
• ${sourceJson}

Be sure to include these if you specify an override prompt.