Agents & tools
AI Agent
Software that uses an LLM to plan and act — picking tools, taking actions, and adapting based on results to complete a user’s goal.
01 ——
In plain English
An AI agent is software that wraps a large language model with the ability to take action in the world. Where a chatbot replies, an agent does — it calls APIs, runs code, browses the web, or edits files in pursuit of a user-defined goal.
Anatomy of an AI agent:
- Brain — the underlying LLM (Claude, GPT, Gemini, Llama)
- Tools — the things it can call (search, code execution, your CRM)
- Memory — short-term context and longer-term recall
- Loop — the cycle of "think → act → observe → think again"
Common agent categories on the market:
- Coding agents (Cursor, Claude Code, Aider)
- Research agents (ChatGPT Deep Research, Perplexity, Elicit)
- Voice agents (Vapi, Retell)
- Browser agents (Operator, Browser Use, Comet)
- Operations agents (Lindy, Relevance AI, Gumloop)
Agent quality depends as much on the harness (how well it uses tools, recovers from errors, knows when to stop) as on the underlying model.
02 ——
Related terms
Agent
An AI system that can take actions, use tools, and make decisions autonomously to complete a goal.
Agentic AI
AI systems designed to act, not just respond — they plan, use tools, and make decisions across multiple steps to complete a goal.
Agent Harness
The scaffolding around an LLM — tools, memory, loops, and orchestration — that turns a model into an agent.
Tool Use
The ability of an AI model to call external tools — like a calculator, search engine, or API — to help answer a question.
Coding Agent
An AI agent specialised for writing, editing, and debugging code — usually with the ability to read your repo, run tests, and open pull requests.
Browser Agent
An AI agent that controls a real web browser — clicking, typing, and reading pages — to complete tasks on websites that lack APIs.
Back to glossaryLast reviewed May 2026