Daydream.

It's a Tuesday night and someone in Brooklyn is typing a sentence into a website. The sentence is, "I'm going to a wedding in Tulum in August, the bride hates white, I have $400 to spend, please help." Eight seconds later, a strapless silk midi from a brand she has never heard of arrives at the top of her screen. She clicks. She buys. She tells nobody. That, more or less, is Daydream.

For two decades, online shopping has worked roughly the same way: type two words into a search bar, scroll past 11,000 results, give up, refine, give up again. Daydream's bet is that none of this is necessary anymore - that the limiting factor in fashion ecommerce was never inventory, but the act of asking for what you want.

The company calls this conversational commerce. Its founder, Julie Bornstein, has a less polite name for the thing it replaces. She has called it, in interviews, "the keyword search tax."

A search bar that never quite listened

Bornstein has spent her career staring at the same problem from different chairs. She built ecommerce at Nordstrom. She ran digital at Sephora. She was COO at Stitch Fix. In 2018, she co-founded The Yes, an AI shopping app that Pinterest acquired in 2022 and quietly absorbed. Each of those companies, in its own way, tried to answer the same question: what would shopping look like if the software actually understood you?

The honest answer, until very recently, was: it can't. The tech was not there. Recommendation engines could match you to products you had already bought. They could not parse "wedding in Tulum, the bride hates white." That requires understanding occasion, color theory, climate, and the social politics of beachwear in one breath - a job for a person, not a search index.

Then, somewhere between 2022 and 2024, large language models got embarrassingly good at understanding sentences. Bornstein, watching this from the sidelines, saw a door that had been locked for fifteen years swing open. She walked through it.

Fifty million dollars, no product, two months.

In June 2024, Daydream announced a $50 million seed. Forerunner Ventures and Index Ventures co-led. GV and True Ventures joined. Karlie Kloss, the supermodel and founder of Kode With Klossy, wrote a check too. The kicker: there was no consumer product to point to. There was a thesis, a team, and a track record.

The team is the part most people skip past. Bornstein recruited four co-founders - Matt Fisher, Dan Cary, Lisa Green, Richard Kim - and later brought on Maria Belousova, formerly CTO of GrubHub, to run engineering. The headcount has since grown to roughly sixty, working out of New York and quietly stacking infrastructure: Vertex AI, Delta Lake, Elasticsearch, Kubernetes, a heavy lean on Anthropic's Claude. None of which the shopper will ever see, which is rather the point.

A box, a button, and a lot of math.

Open daydream.ing and the interface is almost insulting in its simplicity. A chat box. A camera icon. A prompt that reads, more or less, "Say more." Type a sentence. Drop in an image. Both. The system asks a few onboarding questions - sizing, favorite brands, budget - and then it gets out of your way.

Under the hood, an LLM rewrites your sentence into structured intent (color, silhouette, season, price band, formality), runs it against a vector index of two million product descriptions and images, and surfaces matches across roughly 8,000 brands. Click an item and Daydream sends you to the merchant's own checkout. It holds no inventory. It processes no payments. It runs no ads. It takes a commission. That is the whole business.

It is, deliberately, the opposite of Amazon. Amazon owns the warehouse, the recommendations, and the credit card flow. Daydream owns none of those things. What it owns is the conversation - which, if Bornstein is right, is the only piece worth owning.

Numbers, names, and the cap table.

A pitch is a pitch until people show up. By launch day, Daydream had 200+ retail and brand partners signed: Chloé, Acne Studios, Khaite, Markarian on the luxury side; Madewell, J.Crew, Levi's, Alo Yoga, Nike for the rest of us; Net-A-Porter and LoveShackFancy holding the middle. The catalog is large enough that the product can actually do what the marketing claims.

The brand identity, designed by Red Antler, leans hard into the conversational metaphor. The logo is two overlapping chat bubbles. The wordmark is "Say More." Whether this aged into self-aware charm or marketing-deck cliché is a fight for another decade. For now, it does the job: you look at it, you understand what the product wants you to do.

Make the search bar feel like a friend.

Stripped of its marketing, the Daydream mission is small and specific: replace the keyword box with a sentence. Most of the company's choices flow from there. No checkout, because checkout would distract. No advertising, because ads would corrupt the recommendations. No inventory, because inventory would slow the catalog. The product is the conversation. Everything else is plumbing.

This is also why Daydream is, on paper, fragile. Apple, Google, OpenAI, Amazon - any of them could, in theory, drop a similar product on top of an existing distribution channel. Daydream's bet is that they will not do it as well, because they do not care about fashion the way fashion-tech veterans do. Whether that bet holds is the question that will define the next twenty-four months.

If this works, the search bar is finished.

Daydream is one of the first consumer-scale products that treats a chat box as the front door to a transaction, not a customer-support afterthought. If the model is right - if shoppers really do prefer sentences over filters - then a category of software older than the iPhone is about to start dying. That category is keyword search, in the narrow ecommerce sense. Filters, facets, drop-downs, sort-by-price-low-to-high. All of it.

If the model is wrong, Daydream becomes a footnote: a beautifully built, well-funded, slightly-too-early experiment. Bornstein has had one of those before. She built a company, sold it to Pinterest, and walked away with both money and lessons. She is not, by any measure, afraid of trying again.

Back to that Tuesday night in Brooklyn. The shopper has paid. The dress will arrive Thursday. She does not write a review. She does not screenshot the chat. She closes the tab and goes back to whatever she was doing. The remarkable thing is how unremarkable the whole experience felt. Which is, of course, the whole point.