How Consumers Are Using AI to Discover and Buy

Consumers are using AI to frame purchase decisions before they search, compare, or visit a brand site. AI defines the category, assembles the shortlist, and recommends which brands deserve consideration. That makes brand familiarity a weaker predictor of inclusion than it has been for decades.

In the Agentic Divide study, 94% of U.S. AI users said they would consider a brand they had never heard of if an AI assistant recommended it. Only 6% said they would stick with a brand they already use. Being known is no longer the same as being recommended.

Do consumers buy from AI recommendations?

Yes, and a majority already have. Fifty-one percent of U.S. AI users reported having purchased from a brand an AI assistant recommended, up from 36% in the first wave fielded in March 2026. Another 68% said they have asked an AI assistant to build them a bundle or a plan.

Purchase behavior is not the same as delegation. Comfort drops sharply as the AI takes more control: Seventy-four percent are comfortable with an AI suggesting options, 50% with an AI filling a cart they review, and 40% with an AI completing the purchase outright.

How many people start a purchase with AI today?

Financial services is furthest along. In wave two, 21% of home lending buyers said they already begin the category through AI, the highest of the 12 categories tested.

Growth runs the other direction. The categories with the lowest adoption today are expanding fastest in expected use, and the gap between where a category sits now and where it is headed differs enough that a single national number will mislead you about your own. The full leaderboard, with today and two-year figures for all 12 categories, is in the report.

Do consumers trust AI product recommendations?

Partly. AI ranks fifth of seven information sources on trust at 59%, behind search engine results at 76%, friends and family at 74%, brand websites at 73%, and retailer sites at 70%. It ranks ahead of store associates and influencers.

Consumers also named four specific conditions they need met before they will act on an AI recommendation, and those demands intensify with age. The report covers all four and what each one requires you to build.

How does AI buying differ by generation?

The generational split is the widest gap in the study. Fifty percent of Gen Z respondents said they are comfortable letting an AI complete a purchase, compared with 13% of Boomers and older. Gen Z also leads on having bought from an AI recommendation, 61% to 25%.

Older buyers lead on the things that slow a purchase down. Eighty-two percent of Boomers and older prioritize total cost, against 58% of Gen Z, and 79% want the final say, against 58%.

One finding runs against expectation: Trust in AI recommendations is higher among Boomers and older (57%) than among Gen Z (37%), while trust in influencers runs the opposite way. AI recommendations reach the buyers creator marketing was never going to.

What should marketers do?

The report includes a six-step playbook covering how to audit the way AI assistants currently describe your brand, how to make your value story legible to a model, how to read your own category's timeline, and how to prepare for the visit that follows a recommendation.

One number sets up the last of those. Fifty-one percent of consumers go straight to a brand's own site after an AI recommendation, so being recommended is wasted if that page cannot finish the job.

Methodology

The study, in partnership with GWI, ran in two independent waves among U.S. AI users. Wave one fielded in March 2026 with 2,000 respondents across five categories: travel, vehicles, insurance, mobile devices, and home appliances. Wave two fielded in April 2026 with 1,500 respondents across seven new categories: checking accounts, credit cards, home lending, investing, consumer electronics, fashion, and healthcare. Results are survey weighted. Generations are defined as Gen Z 28 and under, Millennial 29 to 44, Gen X 45 to 60, and Boomer and older 61 and up.