Zum Hauptinhalt springen

Integration

Table of Contents​

Conclusion Model​

Node: embraceableAI
Action: embraceableAI conclusion request

As outlined in the Introduction, the embraceableAI community node integrates the platform's AI capabilities into your n8n workflows, providing a structured interface on top of the OpenAI API.

The Conclusion Model allows you to generate reliable conclusions within defined policy constraints. Its architecture ensures compliance with the specified guidelines.

Each API key operates within its own space. Since additional model variants may be introduced over time, you must explicitly specify the model for each node.
To do this, copy the model name from the sample request in the embraceableAI Console and paste it into the Model field of your n8n node.

Request Structure​

The request consists of three main components:

  1. Goal: Define the objective or intent of the response you want to generate.\
  2. Input: Provide the specific context or information that the model should process.\
  3. Policies (Guidelines): Define the rules and constraints that the model must adhere to when generating the output.
    • You can pass free-text policies as a single combined block, or\
    • Define individual policy fields as separate parameters for more granular control.
  4. Advanced Options: Customize the output according to OpenAI-compatible response formats:
    • Stream: Determines whether the output is streamed (note that n8n nodes return results only after full generation, even when streaming is enabled).

Output Structure​

The node’s output is returned in an OpenAI-compatible format. The complete reasoning process of the request — including adherence to policies and the internal task breakdown — is contained within the <think></think> section of the output. After this section, the final summarized response is presented as the model’s main output.

For better transparency, you can visually review each execution through the embraceableAI Console. By selecting the appropriate API key and opening the Logs tab, you can inspect detailed records of every execution.

Decision Model (coming soon)​

Many integrations require not just analytical insights but also structured, predefined outputs.
To address this, we are developing the Decision Model, which will provide decision-oriented logic and responses.

This feature will be available soon.