The past few weeks have seen an explosion of talk about AI agents in the workplace, with everyone talking about how we’re moving past chatbots and rigid workflows towards true autonomous agents which do actual work. Google Next was all about agents, OpenAI announced workspace agents, Anthropic introduced enterprise controls for Cowork, and Microsoft dropped a slew of posts about agents in the enterprise. They’re all basically telling the same story: that autonomous AI is about to do your job, your team’s job, and your company’s entire back office operations.
Meanwhile in userland, most knowledge workers are still just using AI as a fancy answer engine.



