Agentic AI

With the Agentic AI, you can define a goal and let the AI autonomously plan and execute tools to achieve it using the Embabel Agent framework with GOAP planning.

Categories: Artificial Intelligence

Type: agenticAi/v1

Run the agentic AI to autonomously achieve a goal. The GOAP planner selects and orders configured actions based on their input/output bindings, choosing any valid path from the seeded input binding to the goal output binding.

Properties

Name	Label	Type	Description	Required
goalDescription	Goal Description	STRING	Describe the goal the agentic AI should achieve using the configured actions.	true
goalOutputBinding	Goal Output Binding	STRING	The output binding that, once produced on the blackboard, satisfies the goal. Must match the output binding of at least one configured action.	true
goalMode	Goal Mode	STRING Options STRUCTURAL, SMART	Structural: the goal is satisfied as soon as the goal output binding is produced. Smart: additionally asks an LLM to judge whether the produced value actually satisfies the goal description; the planner may backtrack and try alternative action paths if not. Smart mode adds an LLM call per evaluation.	true
systemPrompt	System Prompt	STRING	System prompt to the model.	false
response	Response	OBJECT Properties {STRING(responseFormat), STRING(responseSchema)}	The response from the API.	true

Example JSON Structure

{
  "label" : "Run",
  "name" : "run",
  "parameters" : {
    "goalDescription" : "",
    "goalOutputBinding" : "",
    "goalMode" : "",
    "systemPrompt" : "",
    "response" : {
      "responseFormat" : "",
      "responseSchema" : ""
    }
  },
  "type" : "agenticAi/v1/run"
}

Output

The output for this action is dynamic and may vary depending on the input parameters. To determine the exact structure of the output, you need to execute the action.

Actions

Run

Properties

Example JSON Structure

Output

On this page