A normal chatbot replies and waits. An agent can plan, use tools (search, a browser, a spreadsheet), check the result and try again. You give it a goal, and it works through the steps.
Example
You ask an agent to “find three AI video tools with a free plan and compare them”. It searches, opens the websites, and writes the comparison.
Why it matters to you
Agents move AI from “helpful answers” to “work done for you”, which also means you need to check what they did.
Related terms
ChatbotA program you talk to in a conversation, by typing or speaking./knowledge-base/chatbot/AutomationLetting software do repetitive tasks for you, automatically./knowledge-base/automation/LLM (Large Language Model)An AI model trained on huge amounts of text so it can understand and write language./knowledge-base/llm/WorkflowThe series of steps and tools you use to go from idea to finished result./knowledge-base/workflow/
