Why Regular Users Aren't Adopting AI Agents
Original: Why Normal People Aren’t Using AI Agents
Why This Matters
Agent adoption is critical for AI labs to justify massive model training costs beyond basic chatbot use.
Despite Silicon Valley's enthusiasm for AI agents, mainstream consumers remain largely unengaged. OpenAI reports only 10 million weekly agent users versus roughly 1 billion monthly ChatGPT users. Industry insiders acknowledge no killer consumer agent product exists yet.
Silicon Valley is heavily invested in AI agents, yet mainstream adoption remains negligible. OpenAI's Codex and ChatGPT Work agents combined reach about 10 million weekly users, while ChatGPT and Gemini each boast around 1 billion monthly active users — making agent usage statistically marginal. Anthropic's Claude Code and Cowork agents show similar limited uptake, according to sources close to the company.
Josh Miller, CEO of The Browser Company — whose previous startup Arc was acquired by Atlassian in an undisclosed deal reported at $610 million — sparked debate with a viral X post questioning why the general public 'dgaf' about agents. Miller argues the core issue is definitional: 'No one wants AI agents, because AI agents aren't a thing. It is an invented frame made up by our industry.' He believes the industry must shift focus from showcasing model capabilities to building products people genuinely want.
The disconnect matters financially. AI labs have invested billions training models capable of complex, multi-step tasks, and agents represent their primary strategy for monetizing that investment beyond simple chatbot interactions. Industry insiders note that while resources have poured into agent development, no consumer-facing breakthrough product has emerged. The open question remains: when, if ever, will a 'ChatGPT moment' arrive for AI agents?