What is an AI Agent?
An AI agent is a piece of software that takes a goal, decides what actions to take, carries them out using tools, and repeats until the goal is met. The language model is its reasoning engine, not the agent itself.
An AI agent is software that pursues a goal on your behalf. It sits around a language model and adds the parts the model does not have by itself: a way to take actions, a memory of what has happened so far, and a loop that keeps going until the job is done.
It helps to separate the agent from the model. A model turns input into output. An agent decides what input to give the model, what to do with the output, which tool to call next, and when to stop. Swapping the underlying model changes how well an agent reasons; it does not change what the agent is.
Most agents are assembled from four parts:
- A reasoning engine, normally a large language model, that decides what to do next.
- Tools, the actions it can take: search, an API call, a browser, a file system.
- Memory, so information from earlier steps survives into later ones.
- A control loop that runs the cycle and decides when the goal has been reached.
Agents vary enormously in how much freedom they have. Some follow a mostly fixed sequence with small decisions at each point. Others determine the entire path themselves. Both are called agents, which is why it is worth asking how much a given product actually decides on its own.
Why it matters
An agent is not a model
The model does the reasoning; the agent adds tools, memory, and a loop that turns reasoning into action.
Defined by its tools
What an agent can accomplish is bounded by the actions available to it, which is why tool design matters more than prompt wording.
Autonomy is a spectrum
Some agents follow a near-fixed path with small decisions; others choose every step. Both use the same word.
Needs a stopping condition
A well-built agent knows when the goal is met and when a task cannot be completed, rather than looping indefinitely.
Examples
- •A support agent that reads a ticket, looks up the account, and drafts a reply.
- •A browser agent that signs in to an application and completes a multi-step flow.
- •A research agent that runs several searches, reads the results, and writes a summary with sources.
Frequently asked questions
A chatbot exchanges messages with you. An agent takes actions in other systems to accomplish a goal, and may not involve conversation at all. A chatbot that can also call tools and complete tasks is functioning as an agent.
An AI agent is the thing; agentic AI is the property. The agent is the piece of software, and agentic describes the goal-directed, tool-using, multi-step way it behaves.
Not exactly. A script is better when the steps never change, because it is cheaper and completely predictable. An agent is better when the path varies or the interface changes, because it can adapt instead of breaking.
The main ones are compounding errors across steps, taking actions that are hard to undo, and access scope. Give an agent the narrowest permissions that let it finish the task, and prefer agents that report honestly when they could not complete something.
Vidocu's AI Recorder is an agent that operates a browser to record product workflows and produce documentation from them. Vidocu also provides an API and an MCP server so agents you build can generate subtitles, voiceover, and articles from video.
Related terms
Learn more
- AI Recorder — An agent performs your workflow in a browser and records it.
- Developers — Build video processing into your AI agents and apps.
- Video MCP servers (guide) — How AI agents can process video in 2026.