An AI agent isn't just a chatbot — it's an LLM that can think, decide, and use tools to act on the world. Instead of replying once and stopping, an agent runs in a loop: it reasons about what to do next, calls a tool, observes the result, and decides again — until the task is complete.
Think of it as the difference between a calculator (one input → one output) and an intern with a phone, calendar, and search engine — given a goal, they figure it out.
A clear identity and objective. Tells the agent who it is and what success looks like.
Callable functions like search, read_database, send_email, run_workflow. Each tool has a name, description, and inputs.
Conversation history (short-term) plus a knowledge base or vector store (long-term).
The Think → Act → Observe cycle. The engine that turns a passive LLM into an active agent.