AI Agents
AI that plans and takes actions without you guiding every step.
An AI agent is an LLM combined with a loop that lets it observe, plan, and act — autonomously — toward a goal. Instead of a single prompt → response, an agent can decompose a goal into subtasks, call tools (search, code execution, APIs), evaluate results, and iterate.
A coding agent might: read the failing test → examine relevant files → write a fix → run tests → check output → repeat until green.
Key components: a capable LLM (the brain), tools (capabilities it can invoke), a memory mechanism (context + vector store), and an orchestration loop that feeds observations back as new context.
In plain terms
The difference between a consultant who gives a one-time answer vs. an employee who takes initiative, manages their own workflow, and reports back when done.
Related concepts
Large Language Model
AI trained on vast text to understand and generate language.
Function Calling
Letting an AI invoke real code and APIs mid-reasoning.
Skills
Pre-built capabilities you attach to an AI to extend what it can do.
Model Context Protocol
A universal standard for AI to connect to any tool or data source.