When AI assistants are the front door: the future of brands

Where does shopping start now?

More and more, in an assistant. In a survey of US consumers by L.E.K. Consulting, reported by MarTech in July 2026, 46% of people who use AI started their purchase research on a standalone AI platform, up from 25% in 2024. Traditional search fell from 43% to 24% as a starting point.

The assistants are big enough for that to count. Google says AI Mode passed a billion monthly users within a year, and the Gemini app passed a billion monthly users in August 2026. Amazon says more than 300 million customers used its Rufus assistant in 2025, before folding it into Alexa for Shopping in May 2026.

The traffic that does reach a site is worth more. Adobe measured AI referral traffic to US retail sites up 62% year on year in July 2026, converting 60% better than other traffic. It expects AI traffic to US retail sites to rise 130% this holiday season.

Who is building the front door?

Everyone with an assistant. Where each stood as of 31 October 2026:

1. ChatGPT.Stopped checking out. OpenAI ended Instant Checkout in March 2026, and shoppers now pay on the retailer's site. Brands build apps there instead: Louis Vuitton showed one in June 2026, Prada launched a pilot on 28 October.
2. Google.Checks out natively in the US through UCP, and launched a Universal Cart across Search and Gemini on 15 October 2026. The brand stays the merchant of record.
3. Apple.Said in January 2026 that its next foundation models would be based on Google's Gemini. Its rebuilt Siri, shown on 8 June 2026, answers from the web and acts across apps, but won't be on the iPhone in the EU at first. Apple has said nothing about shopping.
4. Meta.Launched Muse in the US on 8 September 2026, an agent that completes purchases with a one time card from Stripe's Link and can ask the shopper to approve first.
5. Copilot and Perplexity.Both check out in the US only: Copilot Checkout, announced in January 2026, and Perplexity's Instant Buy, through PayPal.
6. Amazon.Keeps the others out. In September 2026 its robots.txt shut 100 of 101 named crawlers out of the whole site, Google's shopping agent, GPTBot, PerplexityBot and ClaudeBot among them, and on 20 September it blocked Meta's Muse. It won a court order against Perplexity's shopping agent in March, which an appeals court threw out on 4 August.

Two patterns stand out. Checkout inside the assistant is staying with the companies that already hold a card: Google, Meta through Stripe, Amazon. And the biggest store online is building its own assistant while locking out everyone else's. Amazon can afford that. Most brands can't.

What changes for brands?

Our read, in four parts.

1. The website becomes the checkout and the proof.Fewer shoppers browse the homepage. They land on a product page having already decided: 31% of AI users told L.E.K. their decision was largely made before they reached a brand or retailer site, up from 26% two years earlier. The site's job is to confirm the choice, with reviews, policies, delivery dates and a checkout that works.
2. Product data becomes the brand.An assistant meets your brand through your feed and your markup before it sees a single photograph. A wrong size chart or a missing color is now a brand problem.
3. If you're not in the answer, you don't exist for that shopper.An assistant names three or four brands, not ten pages of results. There's no page two to scroll to.
4. Loyalty drifts to the assistant.If the shopper remembers that Gemini found the jacket, Gemini gets the credit, the repeat visit and the next question. Unless the brand owns the relationship somewhere the assistant isn't.

What doesn't change?

People still buy from brands they trust. Assistants recommend, shoppers check, and trust is still earned the slow way: products that are as described, deliveries that turn up, returns that don't need a fight.

And the experience after purchase is still the moat. An assistant can find a pair of trainers. It can't make the person who bought them feel part of something. That's the part no agent can copy, and the part most brands spend least on.

That's why we build both sides. Storefront is the agent side: it gives a brand that isn't on Shopify its own agent address and UCP profile from its product feed, so assistants can read the range and send the shopper to the brand's own checkout. Community is the human side: a brand's own app, with events, a feed and chat for the people who buy from it, hosted by us and run by the brand.

What should a brand do this year?

Five things, in order.

1. Find out what the assistants say about you.Ask the questions your shoppers ask, in each market you sell in, across every assistant. You can't fix an answer you haven't read.
2. Treat the product feed as brand work.Every title, variant, price and stock level, kept current, with descriptions that answer questions. Give it an owner with the same standing as whoever owns the brand book.
3. Give agents an address.A UCP profile and an MCP address let any agent search your range and get live price and stock instead of scraping your pages. In our research, every site that had both ran on Shopify.
4. Make the site the proof.Fast product pages, readable policies, real reviews and guest checkout. The shopper who arrives from an assistant came to buy, not to browse.
5. Own the relationship after the sale.Accounts, email and a community the customer chooses to join. That's where loyalty lives when the first visit belongs to an assistant.

So what is a brand for?

What it was always for: being the one people come back to. The difference is that the first meeting now happens between a shopper and a machine reading your data. Get the data right and you're in the answer. Get the experience right and they come back without asking anyone.

The free score shows where you stand today. Your next customer is AI. Your best is still human.