What is an AI Agent ?
An AI Agent, is an LLM Tool calling process that calls AI Tools in a loop to achieve a given goal.
Put simply an AI Agent is simply an AI that keeps calling AI Tools, either locally or through MCP, until is "thinks" its achieved the goal you gave it.
As an example, imagine an AI that schedules calendar invites. Although that task sounds like an atomic, singular thing to do, it's actually made up of a bunch of steps, which the AI Agent can decide to take.
If we asked the scheduling AI Agent "please schedule a meeting with Julia on Monday when she's free to discuss recruiting", the AI Agent would call a bunch of tools, before finally calling the tool to actually create the calendar invite.
- It'll use tools to figure out who "Julia" is,
- Check both our schedules for Monday, and choose a good time,
- Formulate an event description, title and duration,
- Actually create the invite,
- Validate the invite has been correctly added.
Each of those things might need multiple tools, and some might need human input, for example what happens when it finds multiple Julia's, or there are no free slots on Monday ? It'll come back to the user, ask for clarification, and then continue using tools to try achieve the goal of creating an invite.
That is what an AI Agent is.
An AI Agent, is an LLM Tool calling process that calls AI Tools in a loop to achieve a given goal.
There are also AI Agents, who's purpose is to orchestrate other Agents. For example if you have a marketing agent, and a content agent, you could have a orchestrating agent, which decides which agent should handle your request. If you ask it marketing related questions it'll pass the request to the marketing agent, and any content asks will be passed to the content agent.
