dotAI Tool
With the [dotAI feature] active, this tool will appear in the dotCMS back end under Dev Tools -> dotAI. It allows the immediate use of AI operations in context, as well as additional management and configuration options.
Search#
The Search tab performs semantic search against your content indexes: type a question or phrase in plain language and dotAI returns the closest matching content items using vector similarity.
Results display the content item title, Content Type, date, match score, and a snippet of the best-matching passage within that item. The result count and query time appear above the list.
The right panel controls the search parameters:
| Setting | Description |
|---|---|
| Index | The embedding index to search. Leave empty to search all available indexes. |
| Threshold | Maximum allowed distance between the query vector and a result. Higher values return more, looser matches. Default: 0.75. |
| Distance measure | The vector similarity function: Cosine (default), Distance, or Inner Product. |
| Model | The chat model used to generate an answer from the matching passages. |
| Temperature | Sampling temperature for the answer model. 0.0 produces the most deterministic output. |
| Response length | Maximum tokens in the generated response. Minimum: 128. |
| Site | Restrict the search to a specific site. Leave empty to search all sites. |
| Content types | Comma-separated list of Content Type variables to restrict the search. Leave empty to search all types. |
Chat#
The Chat tab provides a conversational interface grounded in your indexed content. dotAI streams its reply as it is written, drawing on the content in your selected index rather than its general training.
Type a message in the input field and press Enter to send (or Shift+Enter for a new line). The same right-panel parameters as Search apply — Index, Threshold, Distance measure, Model, Temperature, Response length, Site, and Content types.
Image#
The Image tab generates images from a text prompt using your configured image provider.
Enter a prompt in the field at the bottom of the pane, select an aspect ratio from the dropdown, and click Generate. The generated image fills the main area. Two icons appear in the top-right corner of the image: a save icon to capture the image as a dotAsset in your instance, and a download icon to save it locally.
Embeddings#
The Embeddings tab manages the content indexes that power semantic search and chat. Each index is listed with its name, chunk count, number of indexed content items, total tokens consumed, and current status.
Use the Filter indexes field to narrow the list by name. The toolbar provides three actions:
| Action | Description |
|---|---|
| New Index | Creates a new embedding index. Configure which Content Types and sites it covers. |
| Rebuild DB | Drops and rebuilds the pgvector index tables. Use this if the index becomes corrupted or after a schema change. |
| Remove content | Removes selected content items from the index without deleting the index itself. |
To delete an index entirely, click the trash icon at the end of its row.
Config Values#
The Config Values tab is a read-only browser showing all resolved dotAI configuration values for the current site, including any values inherited from the System Host. The header identifies which site's settings are being shown and whether they are inherited.
Use the Filter settings field to locate a specific key. API keys and other secrets are masked. All values shown here can be changed via the App configuration by editing the providerConfig JSON — either in the provider sections (chat, embeddings, image) or in the settings block.