AI Agent vs. Chatbot vs. AI Workflow
'Built an AI agent' is now on many resumes, and it covers three different things. A chatbot answers questions in a conversation. An AI workflow runs fixed steps a developer designed, with an AI model used at some of them. A true agent decides its own steps — which tools to call, what to look up, when it is done. Agents are the hardest to make reliable, so real agent experience is the rarest of the three.
How to tell them apart on a resume
Chatbot
Customer-support bot, FAQ assistant, RAG over documents, Dialogflow, a chat interface on an LLM API — question in, answer out.
AI workflow
n8n, Zapier, Make, Power Automate, LangChain chains, 'automated invoice processing' — fixed steps with an LLM in some of them.
AI agent
Tool calling, MCP, LangGraph, agent SDKs, planning loops, evaluations, guardrails — the model chooses what to do next.
The question that settles it
“In what you built, who decided the next step — a fixed sequence you designed, or the AI model itself choosing which tool to use?”
Read the full definitions
Open the full tool for the other look-alike pairs, role profiles, and the JD decoder.