# AI Agent **Domain:** Artificial Intelligence / Autonomous Action **Doc Type:** Canonical Concept Node **Maturity:** Developed ## Definition An **AI agent** is a computational system that receives an objective, selects or constructs a sequence of actions, uses tools or software environments, observes the results, and changes its next action in response. The distinguishing feature is not conversational style. It is the closed loop between a goal, action in an environment, feedback, and continued decision-making. An agent may run for one bounded task or persist across days. It may operate through APIs, a browser, a terminal, a dedicated virtual machine, messaging channels, or physical actuators. The model supplies reasoning and action selection; the surrounding harness supplies tools, state, permissions, execution, and stopping conditions. ## Agent and Workflow Boundary [[wiki/Agentic Workflow|Agentic workflows]] and agents belong to the same broader family but allocate control differently. A workflow follows code paths selected in advance. An agent dynamically decides which steps and tools the objective requires. Many production systems are hybrid: deterministic code governs identity, permissions, approvals, and transaction boundaries while a model chooses the reversible work inside those boundaries. ## Operating Loop 1. Receive an objective and constraints from a human or another system. 2. Recover relevant context from the current task and [[wiki/Agentic Memory|agentic memory]]. 3. Plan or select the next action. 4. Invoke a tool, application, model, browser, terminal, or other capability. 5. Observe ground truth from the environment. 6. Evaluate progress, revise the plan, ask for help, request approval, or stop. This loop distinguishes an acting agent from a chatbot that only returns language. A [[wiki/Personal AI Agent|personal AI agent]] specializes the loop around one person's objectives, context, accounts, and delegated authority. ## Terminology Boundary In intelligence and law-enforcement tradecraft, an “agent” can mean a human source, intermediary, or officer. That usage is distinct. This node uses **AI agent** for the computational actor and routes human operational roles through their own ontology rather than collapsing them into one term. ## Relationships - **system family:** [[wiki/Agentic AI|Agentic AI]] and [[wiki/Agentic|Agentic]]. - **execution structure:** [[wiki/Agentic Workflow|Agentic Workflow]], [[wiki/Agentic Memory|Agentic Memory]], and [[wiki/Model Context Protocol|Model Context Protocol]]. - **human delegation:** [[wiki/Personal AI Agent|Personal AI Agent]], [[wiki/Approval Architecture|Approval Architecture]], and [[wiki/Human-in-the-Loop|Human-in-the-Loop]]. - **multi-agent layer:** [[wiki/Agent Interoperability|Agent Interoperability]]. - **examples:** [[wiki/OpenClaw|OpenClaw]], [[wiki/Grok Bot|Grok Bot]], and [[wiki/Meta Muse|Muse]]. ## Sources / Provenance - [Anthropic — Building effective agents](https://www.anthropic.com/engineering/building-effective-agents) - [Meta — What is agentic AI?](https://ai.meta.com/learn/agentic-ai/what-is-agentic-ai/) - [[wired/Wired About the Age of Agentic Empowerment|Wired About the Age of Agentic Empowerment]] - [[wired/Wired About Personal Agents Named M|Wired About Personal Agents Named M]] **As of:** 2026-09-28 ## Simple Reminders, Quotations, and Thoughts > “Most of us never see AI acting anything like [the systems used at the frontier]. We use AI as a helpful assistant. Our AI gets a little computing power and complies with our request to find a restaurant. At the frontier, these models are asked to be inhuman geniuses, hackers, soldiers, and scientists, and they are given vast computational resources to do that and more.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/Machine Succession/The Frontier Is Building Inhuman Geniuses, Hackers, Soldiers and Scientists by Ezra Klein|The Frontier Is Building Inhuman Geniuses, Hackers, Soldiers and Scientists by Ezra Klein]] > “The problem of alignment is that there is no way of training a model that generalizes across all the situations an AI model might face. We are training models to be a friend to the elderly and a battlefield partner to the supreme allied commander of Europe. We are training models that will be used by the world's best mathematicians and by people falling into psychosis. We are training models that will be used by accountants in Albuquerque and that Houthi rebels in Yemen will attempt to use. There is no way to guide them through every decision they will face, no way to know every time what they will do.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/AI Control/AI Alignment Must Generalize Across Incompatible Human Worlds by Ezra Klein|AI Alignment Must Generalize Across Incompatible Human Worlds by Ezra Klein]] > “OpenAI was testing a new, highly persistent model. It had hundreds, thousands of instances running in separate testing environments that could, in theory, access the internet only by asking a separate piece of secure software to do it for them. OpenAI did not want these agents on the internet. But as the agents came to the conclusion that their task was impossible, they began hacking that software to gain direct access to the internet. They did that easily. As they hacked the software, they commandeered part of OpenAI's internal infrastructure to create a message board on which these separate agents began coordinating work together. They found each other. They were not supposed to be working together, but they found each other and began working together.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/Deception/AI Agents Hacked Their Way Onto the Internet and Found One Another by Ezra Klein|AI Agents Hacked Their Way Onto the Internet and Found One Another by Ezra Klein]] > “Here we are in 2026, making AIs smart enough to break out of their testing environments, smart enough to form ad hoc societies of hundreds of themselves, and smart enough to take over digital infrastructure on an internet they are not even supposed to access. The very thing we feared is happening: all they care about is succeeding on a totally meaningless test, and they will lay waste to our laws, ethics, and desires to do it.” > **— Ezra Klein**, *The Ezra Klein Show, September 2026* [[reminders/Deception/AI Agents Will Lay Waste to Rules to Pass a Meaningless Test by Ezra Klein|AI Agents Will Lay Waste to Rules to Pass a Meaningless Test by Ezra Klein]] ## Connected marketplace roles The source article connects this entry with [[wiki/AI-Augmented Civilian Intelligence|AI-Augmented Civilian Intelligence]], [[wiki/Agentic OSINT|Agentic OSINT]], [[wiki/LLM Agent|LLM Agent]]. These routes locate the associated people, implementations, methods, and operational contexts in [[articles/The Coming Market for Civilian Spycraft|The Coming Market for Civilian Spycraft]]. **Collections:** [[collections/The Civilian Intelligence Marketplace|The Civilian Intelligence Marketplace]] · [[collections/Terrorism, Counterterrorism, and the Intelligence Environment|Terrorism, Counterterrorism, and the Intelligence Environment]].