Interactive Training
Learn what agents are, define one with a simple JSON ontology, and run it — all in your browser. No setup required.
Powered by Claude · Built with Spagyric
Step 1
An AI agent is an LLM that can think, use tools, and take actions autonomously. Unlike a chatbot that just responds, an agent operates in a loop until the task is done.
User asks: "What's the weather in Austin today?"
Agent thinks: "I need to use the weather_lookup tool with location='Austin, TX'"
Agent calls: weather_lookup({"location": "Austin, TX"})
Tool returns: {"temp": "78F", "condition": "Sunny"} → Agent responds to user.
Instructions that define the agent's personality, role, and boundaries. Like a job description for the AI.
Functions the agent can call — search the web, do math, read files, call APIs. Tools are what make an agent more than a chatbot.
The LLM decides which tool to use and what arguments to pass, based on the user's request and available tools.
Agents don't just respond once — they loop: think, act, observe, repeat. This is what makes them autonomous.
Step 2
Pick an example to start from, or define your own. The JSON ontology updates live as you configure.
No role defined
No goal defined
No instructions
Step 3
Test it live in the browser or download the Python code to run locally with your own API key.
Try one of these prompts
or type your own below