It has been revealed that daydream wants to become a personal stylist in your pocket. The AI-powered fashion discovery platform, launched in 2025 by former Stitch Fix executive Julie Bornstein, is introducing a 24/7 AI shopping agent that can learn consumers’ tastes, assemble outfits and purchase products on their behalf — even while they sleep.

In a recent development, the new agent, which shoppers can chat with through Daydream’s app or by texting it directly via SMS, marks an evolution from the conversational fashion search engine the company rolled out just over a year ago. It can remember previous exchanges, put together complete outfits, offer virtual try-ons using an uploaded photograph and handle the steps leading up to checkout.

During an exclusive preview, co-founder and chief brand officer Lisa Yamner asks the agent to resume shopping for an upcoming trip to Mexico. Remembering her aims, it asks a few questions before assembling looks for beachfront dinners, complete with dresses, shoes and bags. Yamner swaps between pieces, tries them on using a virtual version of herself and uploads an influencer’s Instagram image to find similar items.

“Once you express a need or an interest, it can shop while you sleep,” Yamner explains. Shoppers can leave the agent to monitor products for price drops or continue working on a shopping request in the background, rather than repeatedly checking themselves. Any purchase still requires their approval. “We firmly believe that shopping doesn’t have to be a task, but there are parts of it that become that.” The aim is to take the drudgery out of shopping without losing the pleasure of discovery, she says.

At launch, the agent can shop across Daydream’s existing network of roughly 350 brand and retail partners, spanning more than 10,000 brands and three million products. Shoppers can ask it to purchase an item on their behalf, or check out directly with the retailer.

It has been revealed that it comes as OpenAI, Instinct and Meta double down on shopping as a core use case and revenue stream for their newly introduced ‘always on’ agents, and fashion retailers are now racing to build their own agents. Last week, Zalando revealed plans for an AI agent that will purchase products on customers’ behalf from other retailers as well as its own catalog, from 2027. Meanwhile, brands including Ralph Lauren and J.Crew are building out their own branded AI shopping assistants, with J.Crew partnering with Anthropic on its upcoming AI “Style Advisor”. Last month, Prada became the first luxury brand to launch an official plugin within OpenAI’s ChatGPT.

Industry observers note that daydream's answer is to sell its fashion expertise. The firm has developed an internal “fashion knowledge graph” to enrich the information attached to products in its catalog, while human stylists help shape the agent’s recommendations. It recently appointed Francesca Sully as its first “stylist in residence”, tasked with teaching the technology to think more like a stylist and keep pace with trends.

Recent findings suggest that daydream recently appointed Francesca Sully as its first “stylist in residence”, tasked with teaching the AI agent to think more like a human stylist and keep pace with trends. Photo: Courtesy of Daydream.

Sources indicate that its founders believe a general-purpose AI agent, however sophisticated, will struggle to replicate the nuances of personal style. It comes back to the central question haunting fashion AI: can it mimic taste?

“We feel like a generalized agent is just not going to get fashion right,” Bornstein tells Vogue Business in an exclusive interview alongside co-founder Lisa Yamner. “Style is so nuanced. It relates to some very specific things that matter. So we need to know the right questions to ask, but also understand the product catalog of every brand out there — not just the price points, but also what the aesthetic of a brand world is, and how to connect that to other brands that align with that aesthetic and may be of interest.”.

It has been revealed that that expertise is also intended to help shoppers discover unfamiliar labels: approximately 97% of searches on Daydream contain no brand name, according to Bornstein. But in an increasingly crowded market, can that fashion expertise stand up as an independent AI shopping offering against the mass convenience of frontier AI agents and trusted retailers’ owned product catalogs?

Recent findings suggest that when Daydream debuted last year, its proposition was to solve one of fashion e-commerce’s enduring frustrations: finding the right product across a fragmented universe of brands and retailers. Its conversational search engine allowed shoppers to describe what they wanted in natural language, rather than relying on conventional keywords and filters, encouraging owners to “say more” about what they sought.

Although Daydream had already positioned itself around the idea of an AI agent, what it lacked was the ability to deliver the more involved, personalized service its founders envisaged — agents in their current form didn’t actually exist.

“We’ve been wanting to build from day one, but it took us a while to get there for a couple of reasons: one was that the agentic technology wasn’t yet there, and two is that the first version of the platform helped us really understand what consumers were looking for and what they wanted from an AI shopping assistant,” Bornstein notes, explaining that Daydream was finding its original product couldn’t deliver the level of assistance shoppers wanted. Shoppers were asking very specific questions to find products for a need that included an occasion and their body type, and Daydream was showing them a page of search results, “when what they really wanted was more tailored help,” Bornstein adds. Conversational search might make finding fashion products easier, but it doesn’t necessarily solve the more subjective problem of deciding what to wear.

Bornstein recalls a similar challenge at her previous firm, The Yes, where shoppers struggled to move beyond the conventions of traditional search. “We tried to get people to search by telling us a need and they would say, ‘Oh, I need a red dress for a Valentine’s Day party. I want to look sexy, but age-appropriate.’ And then you’d say, ‘Great, go ahead and type it into the search part.’ But then they would type in ‘red dress’, because their trust in search working was so low, and there was no mental model of being able to give a more detailed prompt.”.

ChatGPT helped change that behavior, she says, making consumers more comfortable describing their needs in detail — and helping people understand how to use a product like Daydream. “Consumer expectations have started to change and people now understand what an agent is in a way they definitely didn’t two years ago when we started and no one understood what we were talking about,” Bornstein says.

In a recent development, the platform indicates it has amassed more than 2 million customers since its June 2025 launch, with millennial women its largest audience, followed by Gen X and Gen Z. Bornstein describes two overlapping groups: fashion-savvy shoppers who appreciate having thousands of brands in one place, and time-poor consumers seeking more direction. But Daydream won’t disclose how many of its 2 million shoppers return or make purchases, making it difficult to gauge whether initial curiosity is translating into increased sales.

According to recent reports, it comes just a week after Zalando revealed an adjacent offering — a fresh AI agent feature, Zalando Extended, that will allow shoppers to discover and buy products from other third-party retailers without leaving the Zalando app, available from 2027. The European multibrand platform has 63 million active customers, extensive brand relationships and years of purchase and sizing data — assets that a much younger shopping platform like Daydream can’t easily replicate. Yet Bornstein argues the two companies are targeting different markets — Zalando solely operates in Europe, while Daydream is focused solely on the US. She also says the US retail landscape’s department stores and multibrand retailers generally have “more curated assortments”.

“The fact Zalando is going outside its own catalog makes a lot of sense because they’ve always tried to be extremely broad and like the higher-end ‘Amazon of fashion’ in Europe,” Bornstein argues.

But geography offers only a partial defence in a market where frontier AI models like ChatGPT and Claude are developing shopping capabilities that could eventually reach consumers across multiple countries, and gain more sophisticated fashion knowledge, via direct brand plugins like ChatGPT’s recently launched Prada plugin.

Asked what Daydream can offer that a much larger retailer can’t, Bornstein points to its first-mover advantage as the first cross-brand conversational search tool that’s “properly entrenched in fashion”, and the two years of data and prompts it’s collected from fashion-savvy consumers in that time. “I would also say we are solely focused on this,” she adds. “Having worked for other larger companies before in my past, I know how hard it is to build something like this on the side.”.

Recent findings suggest that there is also competition from fashion brands themselves. Ralph Lauren’s Ask Ralph and the launch of J.Crew’s fresh AI stylist last week suggest that brands increasingly want to provide personalized guidance within their own shopping environments.

Recent findings suggest that yet Bornstein says these brand-owned stylists can sit alongside Daydream’s ambitions. “When we launched originally, many brands asked us if we could build this AI agent for their sites,” she says. “And what we said to them is our initial goal is to have one place for consumers to go and see all brands — without having 10 tabs open.”.

Daydream has also expanded into providing its technology to brands. Around eight brand sites are currently using its conversational search technology, Yamner says, and the company plans to bring some of its newly introduced agentic capabilities to those partners.

Recent findings suggest that “Our feeling is if you know the brand and you go straight to the brand, there’s no reason why you wouldn’t want tools to help you shop that brand as effectively as you can,” Bornstein says. “So we’ll evolve the agentic product that we provide to brand sites to map to our core product, as it’s very portable to single brands too. But we are fundamentally focused on the B2C experience — it just also happens to work for B2B.”.

One of Daydream’s potentially most valuable assets is the information it gathers from the way consumers describe their shopping needs. Traditional e-commerce search data might tell a brand that shoppers are searching for dresses or shoes. Conversational search can reveal the occasion, aesthetic and practical requirements behind those searches, as well as details like whether shoppers are searching for particular past collaborations or archive pieces — valuable data for brands’ product development and merchandising decisions, Bornstein indicates.

But the next stage of its personalization strategy requires shoppers to share considerably more. Daydream is developing integrations with Gmail and Google Calendar so its new AI agent can access this customer data and allow it to identify upcoming occasions and learn about previous purchases and marketing email subscriptions. Shoppers will soon also be able to share Pinterest boards with the agent to communicate their style preferences.

It has been revealed that users can set Daydream’s upcoming AI agent tasks, like price comparison and stock checks, that it can carry out 24/7, even while they sleep. Photo: Courtesy of Daydream.

The pitch is that Daydream’s AI stylist would therefore know what’s already in a shopper's wardrobe and make more relevant recommendations. But granting a commercial shopping platform access to emails and calendar information is a significant privacy decision. Consumers will need to understand exactly what Daydream can access, how that information is used and what control they retain over it.

In a recent development, daydream also operates on a commission model, yet Bornstein insists that commercial relationships won’t influence what the agent recommends — and it’s not pursuing advertising within the agent, at least for now. “There are no commercial incentives that go into what highlights up for the customer,” she says. “Whatever they buy, we receive the same level of commission. So we are completely agnostic in what we’re showing and it’s about the customer’s preferences.” Brands will continue to receive customer information when purchases are completed through Daydream.

Recent findings suggest that while Daydream is investing in a more complex and personalized experience, it’s not disclosed the conversion or retention figures needed to assess how effectively that experience translates into repeat purchasing. But Bornstein and Yamner remain convinced that fashion requires a different approach from the general-purpose agents increasingly being developed to manage everyday tasks.

“We’re watching the direction that the tech and customers are moving in, and we’re trying to get out ahead of it,” Yamner indicates. “We’re trying to say we get it. We get that this is what the customer’s expectations are going to become.”.

→ Read original article