Ron Johnson, the executive who built Apple’s retail operation, believes Silicon Valley is overestimating how much shopping consumers will delegate to AI. Agents can improve research and narrow choices, he argues, but they cannot let a buyer feel a laptop’s weight, judge its display or decide whether its size is comfortable.
His position is not a rejection of AI. Johnson describes himself as an optimist about the technology. The disagreement is over where automation ends and human experience continues, particularly for expensive or personal purchases.
Agentic Commerce Meets a Physical Purchase
Google and OpenAI are pushing AI further into shopping. Google’s Universal Commerce Protocol is designed to help agents move consumers from product discovery to checkout. OpenAI is developing ChatGPT as a destination where users can research, compare and sometimes buy goods without leaving the conversation.
Johnson doubts that consumers will routinely allow an agent to select and purchase a $1,000 or $2,000 laptop without visiting a store or website themselves. In his view, AI may produce a better-informed shopper, but the customer will still want direct experience before committing to certain products.
That distinction puts agents in a supporting role. They can reduce a large catalog to plausible options and prepare questions before a store visit. They do not recreate touch, scale or the confidence that comes from trying an item personally.
Johnson has heard previous predictions that online technology would eliminate stores. He joined Apple in 2000, as ecommerce was expanding, and helped create a physical network that became central to both sales and customer relationships.
Apple’s Retail Advantage Was Built Around People
Apple designed its stores as more than checkout locations. Customers could try products, learn how to use them and return for assistance when something failed. Johnson revisits those decisions in his book, “Shop Different: How Retail Revealed Apple’s Genius,” which describes building the operation with Steve Jobs.
Competitors copied visible features such as glass-heavy design, open floor plans and variations on the Genius Bar. Johnson says they often missed the human element. Apple Store employees were not paid on commission, removing pressure to sell a particular item and encouraging them to identify what the customer actually needed.
That model supports his skepticism about fully automated retail. The store’s value came partly from judgment, service and an ongoing relationship, not only from making product information available. An agent may improve information without reproducing the behavior of employees whose incentives and role were designed around customer help.
Johnson believes Jobs would have embraced AI while refusing to treat it as a substitute for intuition. He recalls Jobs bringing capable people together to debate problems and find new perspectives, a process that depends on human disagreement as much as computational output.
Johnson’s Record Adds Context to the Forecast
Johnson’s retail experience includes failure as well as Apple’s success. He became chief executive of J.C. Penney in 2011 with an ambitious transformation plan and was removed less than two years later after sales fell sharply. He now says he tried to change too much too quickly without bringing employees and customers with him.
He distinguishes Apple’s evolving startup environment from a department-store turnaround. The lesson in his account is that a model developed for one organization cannot simply be imposed on another with different customers and habits.
Johnson later founded Enjoy Technology, which delivered technology products and setup services to customers’ homes. The company filed for bankruptcy in 2022 and sold substantially all its assets to Asurion. Those outcomes do not prove his AI forecast, but they show that it comes from several attempts to connect technology, service and physical retail.
His expectation is measured rather than absolute: AI will improve online shopping, yet may not change the basic channel through which people choose to buy. The coming contest is not necessarily agents versus stores. It is whether retailers can use agents to inform customers while preserving the product experience and human judgment that still influence high-stakes purchases.
