AI shopping assistants now pick products for people. Is yours chosen?

AI shopping assistants are starting to buy on people's behalf, and being mentioned by one is not the same as being chosen. A marketing contributor to Entrepreneur ran 320 simulated shopper chats. Nearly 3 in 10 brand mentions never turned into a recommendation. The brand came up, then the assistant picked someone else.

What the test found

The simulated shoppers were looking for headphones under $150. The drop-off came once a budget entered the conversation. One premium brand was mentioned in 82 answers and recommended in none. A value brand went from never being recommended to almost always being recommended after the price cap appeared.

The author calls that a price-fit story, not a trust story. The assistant knew the premium brands. They just did not fit the shopper's limits, so it had no reason to justify picking them. This is one team's simulation, not a public study. The author does say independent research shows assistants get worse as chats get longer.

Meet the new shopper's helper

Meta's Muse launched Sept. 8 and quickly climbed to the top of the App Store. It is less a chatbot than a personal agent. It runs on its own cloud machine, opens its own browser, scans retail sites, compares prices and checks out for you. It also keeps a shopping profile on each person, and it asks for approval before it buys.

In one early test, most of Muse's product picks came from live websites, not Meta's own catalog. So a brand's site, its product feeds and every catalog that lists it all matter.

Where brands lose without knowing

• Blocked agents: Amazon has already blocked Muse, and other brands may block agents by accident through bot protection.
• Missing details: if price and fit are unclear, an agent has little to justify a pick.
• Memory: a profile note like wide feet and bad knees can shape every later recommendation.

The author also pushes back on the idea, voiced by Forrester among others, that the agent is the new customer. The customer is still the person. The agent is a new gatekeeper and a new audience.

What to do

Check whether your bot protection turns away shopping agents. Then read your product pages the way an agent would. Are the price, size, use and who the product suits stated plainly? If a shopper says under $150 or good for bad knees, your page should make the match obvious. The author says most brands only measure whether they are mentioned. Tracking whether they are recommended is the next step.