AI agents may improve product research without replacing stores for expensive, personal purchases, according to former Apple retail chief Ron Johnson. He expects software to narrow options and make customers better informed, but doubts many people will let an agent independently choose a $1,000 or $2,000 laptop.
His argument rests on what online information cannot provide: the weight of a device, the feel of its controls, the appearance of its screen and confidence that help will be available later. Apple designed its stores not only as sales points but as places to try products, learn how to use them and return for support.
That view challenges current investment in “agentic commerce.” Google is building the Universal Commerce Protocol to move agents from discovery through checkout, while OpenAI is adding research, comparison and purchasing functions to ChatGPT. Johnson believes those systems will influence decisions but not determine how every category is bought.
The claim is an experienced retailer’s forecast, not evidence that automated shopping will fail. Routine replenishment and low-risk purchases may be easier to delegate than laptops or other tactile goods. The useful distinction for merchants is therefore not online versus stores, but which decisions customers are willing to surrender.