Configure LLM to generate descriptions for conversational subflows
Choose which LLM you want to use by default to generate descriptions for inputs, outputs, and skills for conversational subflows.
Before you begin
Role required: admin
About this task
In the conversational subflow settings, you can use generative AI to generate descriptions for skills, inputs, and outputs of a conversational subflow. You can select a default LLM to generate these descriptions out of any of the following preconfigured LLMs:
- Anthropic Claude on AWS
- Azure OpenAI
- Google Gemini
- Now LLM
Note: This configuration affects only the capability to generate the descriptions.
Procedure
Select All.
In the Filter search box, enter
sys_one_extend_capability.list.In the OneExtend Capabilities list, search for the application
Conversational subflows and actions.
Image omitted: csa-llm-generate-description.png
OneExtend Capabilities page shows entries for input - output description generator and skill description generator.
OneExtend Capabilities page shows entries for input - output description generator and skill description generator.
Select the record for which you want to change the LLM:
- Input output description generator
- Skill description generator The OneExtend Definition Configs tab displays separate rows for each configured LLM. The following image shows one row each for Now LLM and Azure OpenAI.
Image omitted: csa-llm-list.png
Options to select the default LLM.
Options to select the default LLM.
- In the Default column of your selected LLM, set the value as true.
Image omitted: CSA-change-default-llm.png
Selecting the default LLM for conversational subflows description generation skill.
Selecting the default LLM for conversational subflows description generation skill.
Once you select the default LLM, ensure that the Default field value for any other LLM is set to **false**.
- Select Update.
Parent Topic:Building subflows