The Agentic Gap: Why AI Checkout Dreams Are Still Just Demos
AI-generated image · US National Wire
Silicon Valley is obsessed with the technical capabilities of AI agents, but until developers pivot from 'sci-fi visions' to solving actual consumer friction, mainstream adoption will remain a rounding error.
In the world of e-commerce and retail tech, we talk a lot about removing friction. But as the tech industry pushes toward a future of AI agents, there is a glaring disconnect between what the labs can build and what the consumer actually wants.
As Wired first reported, Silicon Valley is currently preoccupied with the infrastructure of agentic commerce—designing payment systems for agents and using them to automate corporate workflows. Yet, for the average person, AI agents remain virtually invisible. Josh Miller, CEO of The Browser Company, recently highlighted this gap in a viral X post, noting that while the technology is theoretically ready to transform daily life, the general public simply does not care.
***
**Opinion:** From an operator's perspective, this is a classic case of building for the demo rather than the journey. If developers continue to treat AI agents as a standalone category to be marketed rather than a backend mechanism to solve a specific pain point, they will fail at the checkout. The goal shouldn't be to sell an 'agent'; it should be to sell a frictionless experience where the agent is an invisible engine.
***
The scale of the problem is evident in the numbers. Wired reports that OpenAI's ChatGPT Work and Codex agents, along with Anthropic's Claude Code and Cowork agents, collectively see roughly 10 million weekly users. While that sounds significant, it is a rounding error compared to the billion monthly active users averaging across Gemini and ChatGPT.
Miller argues that the industry's failure stems from a lack of product thinking. He suggests that 'AI agent' is an invented industry frame rather than a product people actually desire. To illustrate this, Miller points to the most popular feature of Dia, the AI-powered browser from The Browser Company: a personalized morning briefing. The feature provides a greeting, a calendar-based to-do list, and curated art to spark joy. While an AI agent makes the feature possible, Miller notes that the user doesn't need to know that.
Instead of focusing on the 'harness' or the tools, Miller suggests the focus should be on creating products that make users feel calm and focused.
Currently, the industry is shipping capabilities—like writing code or navigating websites—because they are impressive technical milestones. Wired reports that many AI lab leaders are driven by a specific sci-fi vision, with Miller noting that almost every lab leader he met with referenced the 2013 film *Her* as their North Star.
Until the industry moves past these recursive self-improvement goals and the obsession with building personal productivity tools for a limited pool of enthusiasts, the 'ChatGPT moment' for agents will remain elusive. The tech is ready, but the product is not.