AI Agent Node
Executes AI-powered tasks using language models and tools.
Input Fields
Task Name: Descriptive name for this AI task.
Model: Select AI model (GPT-4, GPT-3.5, Claude, etc.). Different models have different capabilities and costs.
System Prompt: Instructions that define the agent’s role and behavior. Example: ‘You are a helpful customer service assistant.’
User Prompt: The actual task or question for the AI. Can include workflow variables: Summarize this invoice: ${workflow.variables.invoiceData}.
Temperature: Controls randomness (0.0 = deterministic, 1.0 = creative). Use lower values for factual tasks, higher for creative tasks.
Max Tokens: Maximum length of AI response. Higher values allow longer responses but cost more.
Tools: Select tools the agent can use (web search, calculator, database queries, etc.).
Tool Choice: Control when tools are used - auto (AI decides), required (must use a tool), or none (no tools).
Tips:
- Write clear, specific prompts for better results
- Provide relevant context in the prompt
- Test with different temperature settings
- Use system prompt to set consistent behavior across calls