Agents· 3 min
How do AI agents work?
Agents are LLMs that plan, call tools, observe results and loop until a goal is reached.
Key idea
An agent turns your goal into steps, uses tools (search, code, APIs), reads the outputs, and updates its plan — often for many iterations.
Example
A research agent given 'find 5 recent papers on RAG' searches the web, opens PDFs, extracts titles, and returns a formatted list.
What to remember
- Core loop: think → act → observe → repeat
- Reliability depends on tool design, memory and guardrails
- Multi-agent systems split roles: planner, coder, critic