Why AI Agents Haven't Caught On With Regular People

The gap between Silicon Valley's enthusiasm for AI agents and mainstream consumer adoption reflects a deeper misalignment in how the tech industry approaches product development. OpenAI and Anthropic have trained models capable of complex autonomous reasoning and task completion, investing billions to unlock capabilities beyond simple text generation. Yet at roughly 10 million weekly users, agent products represent what Miller calls "basically a rounding error" next to chatbots with a billion monthly active users. This disparity suggests that technical sophistication alone does not translate to consumer desire.

Miller's core argument—that AI agents are a frame invented by the industry rather than a consumer category people recognize or want—points to a structural problem in how builders approach the market. Rather than starting with the question "What problem do users want solved?", many are showcasing the most impressive things their models can do: website navigation, code writing, autonomous task completion. The result feels like demos rather than products designed for everyday use. His example of The Browser Company's morning briefing illustrates the alternative: a feature that delivers genuine utility (personalized information and task organization) happens to rely on agentic AI under the hood, but users experience it as a seamless, joyful interface, not as an "AI agent." The industry's next challenge is recognizing that hiding the technology and leading with human need, rather than showcasing technical capability, may be the path to the adoption milestone everyone is chasing.