Agenda-setting intelligence, analysis and advice for the global fashion community.
A year after it announced $50 million in seed funding, Daydream — the AI-shopping platform co-founded by e-commerce veteran Julie Bornstein — launched to the US public Wednesday.
The site is the most significant effort to date by a fashion retailer to reimagine online shopping through the capabilities of the large language models underlying AI tools such as ChatGPT. Where other retailers have typically introduced AI chatbots as add-ons to their sites, which still rely primarily on the usual navigation dropdowns and search bar, Daydream’s experience centres primarily on its chat window, which lets visitors find products by typing prompts in natural language — as opposed to keywords — or by uploading a picture.
The AI is able to process open-ended queries and take context into account when providing product matches. Bornstein said in a demo for press Tuesday that one of the team’s favourite examples of a search by an early user in testing was “revenge dress for a wedding in Paris with ‘Saltburn’ vibes,” referring to the 2023 movie.
Shoppers can further refine the results with text prompts, or by clicking a “Say More” button that appears on each product to see similar items, which can also be tailored further with written prompts. Clicking “Say More” on a black T-shirt and typing in a field provided to show versions in another colour will return listings of the same style shirt in that colour.
ADVERTISEMENT
As with Bornstein’s previous start-up, The Yes, which Pinterest acquired in 2022, personalisation will play a key role on Daydream. First-time visitors complete a style profile including their sizes in different products, brand preferences and preferred price range. The homepage features a “daily fashion edit” of products tailored to the user’s tastes, and as the site learns their preferences, it will increasingly personalise the products they’re shown, including when they search.
Daydream believes a high degree of customisation is necessary to make sure users are seeing products that are relevant to them. The platform already has more than 200 retail and brand partners, with a catalogue of nearly 2 million products spanning more than 8,000 brands. It carries no inventory itself. Instead, when shoppers find a product they want to buy, they’re taken to the retailer’s site to complete the purchase, with Daydream taking a commission on the sale.
Like numerous other AI start-ups, Daydream relies on open-source AI models released by some of the major players in the space, such as OpenAI and Google. The company fine tunes its AI specifically for fashion retail, however, and plans to keep upgrading its capabilities to ensure it’s differentiated from competitors using off-the-shelf models with minimal fine tuning, according to Daydream’s chief technology officer, Maria Belousova. It’s already readying a new AI architecture that will see it using a number of different models in concert, each one specialised for a different purpose, which Belousova said will make Daydream’s search both more accurate and faster.
Learn more:
Fashion’s New Era of Product Discovery
AI-powered curation across content and search is helping shoppers overwhelmed by choice to find the items they want, with benefits for brands and retailers, according to the BoF-McKinsey State of Fashion 2025.
