AI Data Based Prompt Activity
This activity is to do chat-like interactions with an LLM. You provide a User Prompt that is sent to the LLM and a response is sent back in the output.
You can optionally pass an AI Conversation to the activity to include conversation history.
Target parameters (in) and Activity options
Input (Entity) | The entity that should be sent to the LLM. |
Input Data Paths (List of String) | By default all properties directly on the input entity are sent. If you want to send only a subset of the properties, or if you want to send properties from a related entity, you can specify these here by adding property paths to the list. |
Output Data Paths | As with the input entity, only properties directly on the on output type are set in the response. If you also want the response to include related properties, or only a subset of direct properties, you can specify them here. |
Output Type (Entity Metadata) | The type of entity that the response should be sent as. |
Profile (AI Profile) | The AI Profile which will be used for the AI Chat. The AI Profile contains AI Service backend and System Prompt which is included in every message sent. |
User Prompt (String) | The User Prompt that is sent to the LLM along with the input entity. |
Source parameters (out)
Error Message (String) | Will contain any error message that was returned by the AI Service backend or that occurred during processing. |
Has Errors (Boolean) | Is true if an error occurred. |
Response (Entity) | The response entity, this is of the entity type specified in Output Type. Note that it will create a new entity for the response, both for persisted and virtual entities. |
Drop targets
Success, this branch will be taken if the operation was successful.
Error Occurred, this branch is taken if an error occurs.