When Julie Bornstein was an e-commerce chief at Nordstrom within the early 2000s, she envisioned a world the place on-line buying would sometime stay as much as the promise of enabling a real customized expertise for every shopper.
“I simply keep in mind considering sometime I want to construct a search and discovery platform that simply helps you discover the correct stuff,” she instructed Fortune in an interview earlier this month.
Over the following few many years, in government roles at Nordstrom, City Outfitters, Sephora, and Sew Repair, and as a founding father of trend startup The Sure, which she offered to Pinterest in 2022, Bornstein received nearer to her objective.
However on Wednesday, 52-year-old Bornstein is unveiling what she considers to be probably the most full manifestation of her quarter-century imaginative and prescient to this point. It comes within the type of an AI-powered trend discovery and buying market known as Daydream, which is launching in beta one yr after Bornstein first introduced a $50 million seed funding.
“That is very a lot the end result of every thing I’ve seen and accomplished,” Bornstein stated. “I’ve simply been ready for the expertise to catch up.”
For Bornstein, the primary seeds for the thought of Daydream got here in 2022 when ChatGPT burst onto the scene and began to coach web customers that written queries, phrased in conversational vernacular, might produce nuanced and detailed solutions and proposals powered by massive language fashions, or LLMs.
Now, Bornstein’s Daydream is making an attempt to beat many present trend manufacturers and retailers, in addition to ask-me-anything chatbots like ChatGPT and Perplexity, by creating an internet expertise to find new trend attire in a lower-friction and extremely customized manner. Whoever cracks this primary might use a primary mover’s benefit to construct a hard-to-beat flywheel of highly-engaged, big-spending customers that entice associate manufacturers and retailers that don’t wish to miss out a brand new development channel.
At launch, prospects who navigate to Daydream.ing (there is no such thing as a app but) are first greeted with a sequence of biographical questions (title, date of beginning, gender), adopted by queries regarding their sizing, favourite manufacturers, and budgets (The truth that the the most cost effective price range vary extends as much as $150 says one thing concerning the high-end buyer that Daydream seems to be focusing on.)Prospects are then prompted for a question: “Inform me, what’s the occasion, temper, or product that you simply’re purchasing for at present?” (New customers can skip the quick introductory survey if they need, and skip forward.)

This query is accompanied by a big field to sort in, that appears deliberately designed to elicit queries which might be longer than only a easy key phrase search. On this reporter’s transient testing, the system rapidly produced fairly good outcomes to some queries—in addition to follow-up instructions when prompted: “On the lookout for sneakers I can put on to children sports activities recreation but in addition get away with at a reasonably fancy dinner.”
When you sort one question, the system rapidly produces a flood of product photographs, every displayed in rectangular tiles, which take over two-thirds of the web page.
The Daydream agent—or chat expertise—strikes to the left hand facet of the display and asks inquiries to attempt to immediate suggestions to additional slim down the outcomes (Customers can even faucet on a product they like, and ask for one thing comparable in a special value level, or present other forms of suggestions comparable to, “I like this model however wish to see totally different materials.) Present extra suggestions and the system will slim or alter the outcomes till you come throughout the correct match.
To buy the merchandise, you click on by from Daydream onto the retailer or model’s personal web site. (Sooner or later, Bornstein plans for an AI agent to finish the transaction on the associate’s buying web site on behalf of the client.) The truth that the transaction takes place on the model or retailer’s web site, and never inside Daydream, is one motive why Bornstein believes so many manufacturers and multi-brand retailers are partnering from launch, however not the one one.
“You possibly can see us as a menace to them, however they’re not,” Bornstein stated. “Everybody’s so bored with being so depending on Google and Fb for brand new customers.”
Daydream has signed up round 200 companions to this point, nearly all of that are attire manufacturers—from Nike to Madewell, and from luxurious labels like Chloe to Khaite—alongside round 15 multi-brand retailers, together with Nordstrom and Web-a-Porter. The startup fees them a referral payment when a Daydream consumer goes on to make a purchase order from them, although Bornstein wouldn’t present the particular minimize. She did enable that their price is increased than affiliate community referral charges—which might vary from 5% to 10% in trend—however decrease than among the charges charged by luxurious marketplaces, which might surpass 30%.
These associate manufacturers and retailers present the startup with their product catalogue knowledge, which Daydream then enriches with further attributes to assist its AI agent make high quality suggestions in response to nuanced questions. At launch, Daydream’s outcomes are influenced by round a dozen small language fashions, every of which deal with sure attributes of merchandise, like coloration or material.
Bornstein believes that whereas the do-everything AI chat and search experiences, like these of ChatGPT or Perplexity or Gemini, would possibly very nicely develop into common for product-related searches in classes comparable to TVs or hairdryers, the place prompts and solutions typically heart on specs quite than issues of fashion and style. However she additionally believes that generative and Agentic AI’s impression on buying is shifting so quick that it isn’t but clear what breakthroughs would possibly come subsequent, nor how quickly.
“I feel the best way e-commerce goes to evolve is difficult to think about,” Bornstein stated. “What we’ve constructed at present is type of a bridge between the place it’s been…and the place it’s going.”