AI Agents

AI agents are systems that use LLMs not just to generate text, but to make decisions, call tools, execute actions, and iterate toward a goal. Unlike a chatbot that answers questions, an agent can navigate systems, edit files, make API calls, and coordinate multiple steps of a workflow.

Agent washing is a term coined by Gartner to describe products sold as "AI agents" that are in practice just conditional logic (if/else) with a superficial LLM layer. The risk for PMs is investing in "agents" that lack real autonomous decision-making and iteration capabilities.

Agents make sense when the workflow has high variability (you cannot codify every path), requires multiple integrated tools, and tolerates iteration with human verification. If the process is deterministic and repetitive, traditional automation is cheaper and more reliable.