AI Search Rewires Consumer Behavior From Discovery to Checkout

The front door to the internet has moved

For about twenty-five years, consumer technology ran on a simple pattern: people typed a few keywords, looked through a list of links and clicked one. Coverage across 2026 shows that pattern breaking down, and fast enough that it now drives decisions in product management, startup investment and even sustainable electronics.

Google made the biggest single move. At its I/O conference on 19 May 2026, it called the redesign of its search box the largest change to Search in more than 25 years, and it now powers that box with its Gemini 3.5 Flash model.9 Reports from the event say AI Mode, once an opt-in experiment limited to the US, became the default for users worldwide.2 Google also says AI Mode has passed 1 billion monthly users, about the same as ChatGPT's estimated monthly audience.3

The main point is that this is not only a new interface. People search differently once they are in it. AI Mode queries run about three times longer than traditional searches.45 Google says more than one in six US queries now come by voice or image.4 One analyst told Time that users in AI Mode rarely leave to check answers elsewhere, which means very few clicks.3 Searching is turning from a list of results into a conversation that stays inside one company's product.

Shoppers are already acting on AI answers

The effect on shopping is clear, though estimates of how large it is vary. A Q2 2026 survey of more than 1,000 US online shoppers by logistics firm Locus found that 45% use tools like ChatGPT and Claude somewhere in their shopping journey.21 Adobe's July 2026 survey of more than 5,000 US respondents put the share at 42%, up four points since March.24 Menlo Ventures offers an important counterpoint. It reports that the overall share of US adults using AI rose only from 61% to 64% this year, while consumer spending on AI more than tripled to an estimated $40 billion.25

These figures fit together once you read them side by side. Few new people are adopting AI. The growth comes from existing users relying on it for more decisions. Menlo found that half of shoppers who use AI used it to compare prices, 37% made a final purchase based on its recommendation, and 26% let it complete the purchase.25

The behavior of these shoppers matters more than their number. According to Locus, AI-assisted shoppers were more than twice as likely as the average consumer to try new brands (39% vs. 18%) and to buy more items per order (37% vs. 17%). They were also more than three times as likely to feel confident buying fewer items.23 That last finding cuts against the idea that AI simply makes people buy more. For some shoppers, the same tool speeds up spending. For others, it supports slower, more careful decisions.23

Traffic data supports the survey results. Adobe's July 2026 retail figures show AI-referred visits up 62% year over year. Those visits now bring in 53% more revenue per visit than non-AI traffic, a reversal from a year earlier.24 Other sources report even larger conversion gaps between AI referrals and traditional organic search, although the methods behind those numbers differ.1

Trust is high and also fragile

Trust is where the coverage disagrees most. Adobe reports that 77% of AI shoppers feel more confident after using AI and 69% say they are less likely to return what they bought. Yet 66% still visit a brand's own site to double-check a recommendation.24 A separate survey found that 78% of consumers believe AI shopping recommendations are already shaped by advertisers. It also found that 76% worry about how chatbots handle their data and 60% would not trust a chatbot with payment details.28

The suspicion about advertisers is turning out to be well-founded. Menlo notes that ChatGPT started showing ads in February 2026, and OpenAI says the business reached $1 billion in annualized revenue about 200 days later.25 Adobe's broader consumer report describes most people as only "somewhat" comfortable with AI. It also finds that trust drops sharply for high-stakes decisions, where consumers want a human involved.19 Some users are pushing back directly. After Google's I/O announcements, DuckDuckGo app installs reportedly rose 30% and visits to its "No AI" search page tripled.7

The pattern looks like convenience-first adoption with trust still uncertain. Consumers are giving AI the research work but keeping a habit of checking, and that habit could fade or harden depending on how ads change these systems.

Where You.com fits in a crowded search market

The changes at Google put smaller search players in a hard spot, and You.com shows how one company is responding. Reviews this year describe the company inconsistently. Some still treat it as a privacy-focused consumer search engine with model switching and a roughly $15-a-month Pro plan.2931 Others say it has moved its focus toward businesses and developers. One review says it now operates mainly as a web-search and research API platform for AI agents, after moving away from its earlier consumer chat product.32 Another describes a three-tier market: free AI search from ChatGPT and Google, paid consumer search led by Perplexity, and a developer and enterprise tier where You.com is placing its bet.30

The company's own materials support the infrastructure reading. Its site leads with search, extraction and synthesis APIs built for AI systems, along with case studies on agent latency and reliability.35 Profiles highlight its own search infrastructure and access to more than 40 language models.36 One review reports that 57% of Fortune 500 companies use the platform and lists usage-based pricing starting at a few dollars per thousand calls.32

One reading of this strategy is that it fits the consumer data above. If agents increasingly do the searching for people, the companies supplying fresh, cited web data to those agents may hold a sturdier position than any single consumer app. Google's own "information agents," which monitor topics in the background, point in the same direction.9 The risk is that reviewers still say You.com trails Perplexity on consumer polish and brand recognition.30

Startups are building the plumbing for agent-led shopping

Investors seem to agree that the infrastructure is where the money is. Rye's map of agentic commerce lays out seven layers, from AI platforms down to trust and security. It finds that payments and identity have drawn the most dedicated startup funding, with Basis Theory, Nekuda and Skyfire raising nearly $50 million combined.39 Visa Ventures and Amex Ventures both backed Nekuda, which suggests the card networks see standard agent checkout as necessary for their own transaction growth.45

New money keeps arriving. German startup ShopAgentic raised a €1.9 million pre-seed round to build commerce systems for AI shopping agents, though its product had not launched yet.41 Spangle, founded by former Amazon executives, raised a $15 million Series A at a $100 million valuation to generate storefronts that adapt to traffic coming from AI search and agents.47 Over two weeks in late July and early August, agentic AI companies raised about $676 million. Nearly half went to security firms that control what agents are allowed to do.42

That focus on security matches the consumer data. Shoppers' main concerns are data handling and payments, and the startups attracting the most capital are the ones working on exactly those problems.

Product managers now design for two kinds of users

Inside companies, these shifts are changing the product manager role. Userpilot reports that PM job postings rose 14% year over year by May 2026. It also cites MIT research finding that 95% of enterprise AI pilots fail to deliver measurable returns.50 Several analyses agree that the work is moving from managing feature lists to designing workflows in which AI agents do much of the task and humans set the limits.4953 Userpilot adds a point that ties directly to consumer behavior: products increasingly have to serve AI agents as well as people, and teams need visibility into how each group behaves.50

The sources mostly agree that AI makes routine work cheaper and makes judgment more valuable. One forward-looking piece warns that AI analysis can still be wrong about 20% of the time. It says pricing, legal disclosures and account deletion should still require strict human sign-off.56

Sustainability: people say green, then do something else

Sustainable technology shows the widest gap between what people say and what they do. In a 407-person study on electronics, most respondents showed strong environmental awareness. Yet only 29.48% knew where to dispose of old devices, and 53.04% kept them at home.15 The same research found growing interest in repairability and more confidence in professionally refurbished devices than in used ones.15 A four-country survey of Germany, Italy, India and South Africa found that limited availability, lack of incentives and the sense that buying new is easier are common barriers to circular buying.12 A Saudi study found that being open about circularity practices encourages responsible purchase intentions, but that effect weakens among consumers who are skeptical of greenwashing.18

AI could help close that gap. Coverage of retail trends notes that AI is making resale workable at scale, through automatic condition grading and pricing for pre-owned goods.22

The bottom line

Across search, shopping, startups and product teams, the common thread is that consumers are handing discovery to machines. The companies best placed to benefit are those that make those machines trustworthy, whether through cited data, secure payments for agents or clear sustainability claims. Trust is still the weak point. The next year will show whether AI's convenience can survive ads, agents that act on their own, and consumers who keep checking the answers.