Build AI That Takes Actions

Guide the user through building AI that reasons and takes actions — calling APIs, using tools, and completing multi-step tasks. Uses DSPy's ReAct and CodeAct agent modules.

Step 1: Understand the use case

Ask the user:

1. What should the AI do? (answer questions, call APIs, perform calculations, search, etc.)
2. What tools does it need? (calculator, search, database, APIs, file system, etc.)
3. How many steps might it take? (simple tool call vs. multi-step reasoning)

Step 2: Define tools

Tools are Python functions with type hints and docstrings. DSPy uses these to tell the AI what's available: