What is AI agent?

An AI agent is a system that uses an LLM to plan and execute multi-step tasks autonomously — calling tools (search, browser, APIs, file IO), interpreting results, and iterating until a goal is met. Agents are distinct from single-turn LLM calls (one prompt → one answer) and from copilots (LLM-as-suggester with human-in-the-loop on every step). The 2025 generation of agents (Anthropic Claude with Computer Use, OpenAI Operator, Google Gemini Agent Mode, ChatGPT Agent Mode, Microsoft Copilot Studio agents) operate browsers, compose multi-file research reports, run code, and orchestrate other AI systems. For marketing teams, agents have started to replace ad-ops tasks: pulling Meta and Google performance data, drafting creative briefs from low-CTR cohorts, generating ad variants, and pushing them back into ad accounts — all without per-step prompting. The same agentic pattern shows up inside AI search: Perplexity's deep research agent runs 30+ web queries before composing an answer, and Google's AI Mode runs multi-step skill chains for complex commerce queries.