"Nobody is using this": Why the public ignores AI agents
The artificial intelligence industry has invested billions in autonomous agents, but the general public is ignoring them. Josh Miller, co-founder of The Browser Company, explains to WIRED why this tech trend is failing.
In the corridors of Silicon Valley, it has been believed for months that AI agents are the greatest technological breakthrough since the launch of ChatGPT. Large companies are building clearing systems for agents, developing models that can write code independently, and even fighting rogue agents that try to hack into servers. However, outside the tech bubble, the general public is simply yawning.
A viral post uploaded this week by Josh Miller, CEO of The Browser Company (which was acquired last year by Atlassian for $610 million), exposed the industry's painful truth. In an interview with WIRED magazine, Miller said:
"Theoretically, the technology is ready to change the way we work and live... but in reality, the general public simply doesn't care."
The numbers speak for themselves. According to company reports, agents like OpenAI's Codex and ChatGPT Work, or Anthropic's Claude Code and Cowork, record only about 10 million weekly users. When more than a billion active users per month visit basic chatbots like ChatGPT and Gemini, this is a statistical rounding error.
"I haven't heard a single person outside the tech community talk about an AI agent they use," Miller tells the magazine. "As we in the industry get more excited about recursive self-improvement, it's worth stopping for a moment and asking: 'Wait, what?!'"
Silicon Valley's problem: technology instead of product
According to Miller, the central failure of AI labs lies in the perception: they treat "agents" as an off-the-shelf consumer product, while in reality, it is only technological infrastructure. Companies are rushing to release to the market the most impressive capabilities their models can perform — such as autonomous navigation of websites or writing code — instead of building products tailored to the desires and daily needs of ordinary users.
The bottom line shows a deep failure in the industry's product perception. Miller notes that during his meetings with the heads of leading AI labs, almost all of them presented the exact same vision for their product, basing it on the Hollywood movie "Her." The lack of diversity of opinion produces products that are similar to each other instead of providing solutions to real problems.
Users don't want to know how the "agent" works behind the scenes. For example, the most popular feature in his company's AI browser, Dia, is a personal morning summary. This feature is technically powered by an AI agent that collects tasks from email and the calendar, but the user receives only a summarized and calm daily schedule, without the need to understand the autonomous mechanism operating in the background.
While the current trend gives users the knowledge and ability to build personal automation software, the number of people who actually want to build basic productivity tools for themselves is tiny.
"Nobody wants AI agents, because AI agents are not a product — it's just the framework our industry invented to call it by a name," Miller concludes. "Let's create products that are useful, fun, and accessible. Who cares if it looks like an AI agent? Nobody is using AI agents."
