Also known as: agentic AI · autonomous agent · tool use
An AI agent is a system that uses a language model to plan and execute multi-step tasks by calling external tools — searching, running code, querying APIs — and iterating on the results, rather than producing a single response.
The essential shift from chat to agent is action. An agent decides which tool to invoke, interprets the result, and decides what to do next, potentially across many iterations, which introduces compounding error: a small mistake at step three propagates through everything after it.
The security implications are substantial. An agent with tool access and a prompt-injection vulnerability is an attacker-controllable actor inside your systems, which is why capability scoping, human approval for consequential actions, and audit logging matter far more here than in conversational applications.
Prompt Injection
Prompt injection is an attack in which malicious instructions are embedded in content an AI system processes — a web page, document, or email — causing the model to follow the attacker's instructions instead of the application's.
Retrieval-Augmented Generation
Retrieval-augmented generation (RAG) is a technique in which relevant documents are retrieved from an external knowledge source and supplied to a language model as context, so that its response is grounded in specific retrieved material rather than parametric memory alone.
Large Language Model
A large language model (LLM) is a neural network trained on vast quantities of text to predict likely continuations of a sequence, producing systems capable of generating coherent language, answering questions, and writing code.
Model Context Protocol
The Model Context Protocol (MCP) is an open standard that defines how AI applications connect to external data sources and tools, providing a common interface so that any compliant client can use any compliant server.
AI Evaluation
AI evaluation is the practice of systematically measuring model performance against defined tasks and criteria, using benchmarks, human review, or automated graders to quantify capability and detect regressions.