OpenAI Assistant Input
Last updated
Last updated
This action will forward any user text to an existing OpenAI assistant. The conversation will hold in position, waiting for the user reply.
The action uses the following settings:
OpenAI API Key
The API Key that you can generate in your OpenAI account
✅
Assistant Id
The unique id of the trained assistant you wish to use
✅
Auto-Generate Content
If enabled, Xenioo will generate the bubble with the content of the assistant reply
⛔
Split Paragraphs
If enabled, Xenioo will follow your assistant line break to generate multiple bubbles
⛔
The assistant reply is returned to the chat (or to a variable) directly without any intervention of Xenioo on the content.
Conversation with the assistant will stay in thread, based on the openai_assistant_thread variale. If you wish to start a new thread just empty the variable before making a new request.
Depending on the OpenAI model, request complexity and OpenAI account level, replies might take more than 20 seconds to be generated.
On Reply Received. This action will trigger any child operation as soon as the result is available. This trigger will not be executed if the "Auto-Generate Content" flag is enabled.
This action produces the following variables after being triggered:
Variable
Description
openai_assistant_thread
The thread id of the current conversation
openai_assistant_model
The LLM model used by the assitant
openai_assistant_raw_reply
The raw JSON reply of the last assistant request
openai_assistant_prompt_tokens
The number of prompt tokens used by the last request
openai_assistant_completion_tokens
The number of completion tokens used by the last request
Channel specific availability follows the table below:
Channel
Availability
Fully Available.
Fully Available.
Telegram
Fully Available.
Web
Fully Available.
Alexa
Not Available. Out of AI Detection is not supported inside skills.
Google Assistant
Fully Available.
Slack
Fully Available.
API
Fully Available.
This action can be used only at Bot Global level or Behavior level.
