Subconscious
AI-generatedSummary
Execute a long-running AI agent task on the Subconscious API, waiting for completion and returning the final result.
Inputs
- engine (required) — Select the reasoning engine (TIM-GPT, TIM-Edge, or TIM-GPT-Heavy).
- instructions (required) — The natural language instructions for the agent to execute.
- platformTools — Select built-in tools (e.g., web_search, exa_search, parallel_extract) the agent may use.
- additionalOptions — A collection for advanced settings: answerFormat (JSON Schema), customTools (JSON), pollingInterval (ms), reasoningFormat (JSON Schema), timeout (seconds), and callbackUrl.
Output shape
A single execution item containing the final agent result, including status, reasoning steps, and output data.
The node polls the API for completion. If a timeout occurs or the run fails, the node throws an error unless 'Continue on Error' is enabled.
Examples
Example 1: Run a web search and summarize the top 3 stories.
Engine: TIM-GPT, Instructions: 'Search for the latest news about AI and summarize the top 3 stories', Platform Tools: Web Search
Example 2: Execute a task with a custom timeout and structured output.
Engine: TIM-Edge, Instructions: 'Find current stock prices', Timeout: 60, Answer Format: JSON Schema defining the stock object