You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/ai_actions/work_with_ai_actions.md
+12-9Lines changed: 12 additions & 9 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -5,8 +5,14 @@ description: Create new AI actions or modify existing ones to work faster and in
5
5
# Work with AI actions
6
6
7
7
AI actions define what results are available to editors in AI-enabled areas, such as, for example, the AI Assistant.
8
-
If AI Actions have been configured in your application, and you have the required [permissions]([[= developer_doc =]]/permissions/policies/#ai-actions), including `Action configuration/Edit` and `Action configuration/Create`, you can reconfigure the existing AI actions, and create new ones.
8
+
If AI Actions, including at least one service connector, have been configured in your application, and you have the required [permissions]([[= developer_doc =]]/permissions/policies/#ai-actions), including `Action configuration/Edit` and `Action configuration/Create`, you can reconfigure the existing AI actions, and create new ones.
9
9
10
+
!!! note "Variations in AI action settings"
11
+
12
+
The following description outlines the UI options displayed when either the default OpenAI connector or the Anthropic Claude [LTS update]([[= developer_doc =]]/ibexa_products/editions/#lts-updates) is installed and configured.
13
+
14
+
Response generation settings may vary depending on the AI service, model implementation, and action type, so the settings shown below might differ from those available in your installation.
15
+
10
16
## View AI actions
11
17
12
18
With the right permissions, you can view all AI actions configured in the application by navigating to the Admin Panel and selecting **AI actions**.
@@ -19,7 +25,7 @@ Out of the box, there are two categories of AI actions present in the system:
19
25
-**Refine text** - used by default in [online editor](create_edit_content_items.md#ai-assistant) for refining text, for example: "Rewrite text in formal tone"
20
26
-**Generate alternative text** - used by default in the [image asset editing screen](upload_images.md#ai) to generate alternative text, for example: "Generate short alternative description of an image"
21
27
22
-
It may happen that a set of sample AI actions has been [installed with the AI actions package]([[= developer_doc =]]/ai_actions/install_ai_actions/#install-sample-ai-action-configurations-optional), and there is a number of existing AI actions that you can modify and clone.
28
+
It may happen that a set of sample AI actions has been [installed with the AI actions package]([[= developer_doc =]]/ai_actions/install_ai_actions/#install-sample-ai-action-configurations-optional), and there is already a number of existing AI actions that you can modify and clone.
23
29
24
30
!!! note "Custom action types"
25
31
@@ -52,22 +58,19 @@ You can modify the existing AI actions.
52
58
53
59
The default request can be seen at the top of the settings area, on a light blue background.
54
60
55
-
-**Max tokens** - sets a maximum number of "[words](https://help.openai.com/en/articles/4936856-what-are-tokens-and-how-to-count-them)" or tokens that can be used in a single request by both the request and the response
61
+
-**Model** - decides what AI service model is used to generate the response
62
+
63
+
-**Max tokens** - sets a maximum number of "[words](https://help.openai.com/en/articles/4936856-what-are-tokens-and-how-to-count-them)" or [tokens](https://docs.claude.com/en/docs/about-claude/glossary#tokens) that can be used in a single request by both the request and the response
56
64
57
65
-**Length of prompt output** - sets a maximum number of words of the generated result
58
66
59
67
-**Temperature** - controls the randomness of the response.
60
68
Takes a value between 0 and 2, but the usual range is between 0 and 1.
61
69
The output is more random at higher temperatures.
62
-
For more information, see the parameter description in [OpenAI API reference](https://platform.openai.com/docs/api-reference/chat/create#chat-create-temperature)
70
+
For more information, see the parameter's description in [OpenAI API reference](https://platform.openai.com/docs/api-reference/chat/create#chat-create-temperature) or [Anthropic's glossary](https://docs.claude.com/en/docs/about-claude/glossary#temperature).
63
71
64
72

65
73
66
-
!!! note "Action settings availability"
67
-
68
-
Action settings differ depending on the AI service used, model implementation and actual action type.
69
-
Therefore the settings visible in your installation may vary from the ones presented above.
70
-
71
74
5\. Click **Save and close** to apply the changes or **Discard** to discard them and close the window.
0 commit comments