Which fashion AI startups raised money in 2024 to 2026?

Confirmed rounds include Daydream ($50M seed, 2024), Raspberry AI ($24M Series A led by a16z, 2025), Doji ($14M, 2025), Alta ($11M led by Menlo, 2025), Gensmo ($60M+, 2025) and Phia ($8M seed in 2025 and a $35M Series A in 2026). Browzwear acquired Lalaland.ai in 2025.

What do investors say matters most for AI shopping?

Product data. a16z says AI shopping is bottlenecked by content, not compute; Bessemer says brands will win on clean data rather than the strongest ads; General Catalyst puts value in machine-readable catalogues and live stock and price APIs.

Do investors think fashion will be bought by AI agents?

They disagree. Bessemer says apparel shows the most early traction, while Activant and VMG argue that inspiration-led and luxury purchases close on brand channels, with agents in a support role for luxury fashion and fine jewellery.

Is there evidence that AI shopping agents drive sales?

Not much yet. Investor memos lead with users and engagement; a September 2026 Newcomer survey of investors called agentic commerce still a rounding error, and Forerunner's Kirsten Green said it was too early for hard data.

What should a jewellery brand take from the investor theses?

Keep the sale with your people and use agents for concierge, appointments and after-sales, as Activant recommends for luxury and fine jewellery, while making your catalogue machine-readable and owning first-party taste and size data.

Fashion AI investors: where the money points

Venture investors have made fashion one of the first consumer categories for AI shopping agents, but their own writing is more careful than the funding headlines. The money of 2024 to 2026 went to fashion search and styling agents such as Daydream, Phia, Alta and Gensmo, and to design and imagery tools such as Raspberry AI and Botika. The published theses, from a16z, Bessemer, Menlo, Activant, General Catalyst and M13, agree on three things: product data is the bottleneck, taste and fit data is the moat, and the rails of agentic commerce (payments, catalogues, trust) will mostly belong to incumbents. They also agree, in their own words, that the hard evidence of sales is not in yet.

This is part three of our fashion AI research series and the second of five deep dives. It covers what investors wrote and funded, how strong the evidence behind it is, and what a fashion or jewellery brand should take from it.

Where did the money go?

The rounds below are the fashion AI deals we could confirm from the investor's own post, the company's release or reputable trade press. Amounts are as reported; user and customer numbers are the companies' own.

Company	Date	Round	Led by	What it does
Daydream	Jun 2024	$50M seed	Forerunner, Index	Fashion search and shopping agent
Raspberry AI	Jan 2025	$24M Series A	a16z	Generative design for apparel, footwear and accessories
Botika	Jan 2025	$8M seed	Stardom, Secret Chord (per press)	AI fashion models and on-model imagery
Doji	May 2025	$14M seed	Thrive Capital	Avatar-based virtual try-on app
Alta	Jun 2025	$11M seed	Menlo Ventures	AI stylist, closet and avatar
Gensmo	Jun 2025	$60M+ early funding	Not named	Fashion agent with recommendations and try-on
Phia	Sep 2025, Jan 2026	$8M seed, $35M Series A	Kleiner Perkins, then Notable Capital	Shopping agent and price comparison, new and resale
Lalaland.ai	Jul 2025	Acquired	Browzwear	AI-generated models folded into 3D design software
Glance (InMobi)	Oct 2026	Strategic, undisclosed	L Catterton	AI shopping across phones and connected TVs

Two patterns stand out. On the consumer side, the leads are top-tier generalist funds betting that the way people find clothes is about to change. On the brand side, the money backs tools that brands pay for today: Raspberry AI reported 70 customers at its Series A, including Under Armour and MCM, and Browzwear bought Lalaland.ai to put AI models inside the 3D design tools brands already use.

The market around them is smaller than the noise suggests. CB Insights counted $5.4 billion of retail tech funding in Q3 2025, its lowest quarter in a year, with AI startups taking 36% of retail tech funding for the year to date.

In India the picture is different. Peak XV led a $26 million seed for Lyskraft, the Myntra founder's omnichannel premium fashion venture, and a $15.3 million Series A for ZILO, 60-minute fashion delivery in Mumbai. Neither was pitched on AI. The AI depth in Indian fashion sits with the platforms: Inc42 reports that Myntra gives AI size and fit recommendations to 85% of users with enough history.

What are the main investor theses?

a16z: content, not compute

Andreessen Horowitz's AI x Commerce essay (August 2025) sorts shopping by how much thought goes into it, from impulse buys to life purchases. Agents add little to impulse buys and most to routine buying and to researching "lifestyle" purchases such as handbags and skincare. Its sharpest line is that AI shopping is "bottlenecked by content, not compute": without structured product data, language models remain clever summarisers. Its example of the preference memory a good agent needs is a shopper who returns cotton sweaters but never wool ones.

In September 2026 a16z's growth team asked who gets paid when AI does the shopping. Their answer is that advertising profit is at risk: Amazon earned $69 billion from ads in 2025 against $34 billion of operating income outside AWS. Merchant software and payment companies gain a new channel; marketplaces that live on ads have the most to defend. The essay's core question is "who owns the customer", which is also the question for every brand.

On fashion design, a16z's Raspberry AI memo argues that general image models "lack the domain-specific detail" apparel needs, and pitches generative design as a route to faster ranges and less overstock.

Bessemer: the delegated buyer, not yet

Bessemer's agentic commerce roadmap (May 2026) is the clearest investor statement on fashion. It says apparel is "where we have seen the most early traction", naming Daydream, Phia and Alta, and in the same essay that the promise of seamless delegation "is not here yet". Its eight predictions include that shoppers become more loyal to their agent than to any brand, that brands win on clean data rather than the strongest ads, and that brands will need two storefronts: one for people and one for agents. It expects payment networks, Shopify, Stripe and the model companies to take much of the stack, leaving startups the neutral problems: trust, product data, measurement and creative governance.

Menlo: buy, don't build, and know the closet

Menlo Ventures' 2025 State of Generative AI in the Enterprise, a survey of about 500 US enterprise decision-makers, found enterprise spend reached $37 billion in 2025, up from $11.5 billion, with 76% of AI use cases bought rather than built. For brands deciding whether to build their own AI, that is the market voting. Menlo's memo on Alta makes the fashion case in one phrase: good personalisation has to know "what you already own, and what will look good on you". Alta reported 270,000 outfits made a day two months after launch, an engagement figure rather than a sales one.

Activant: a new channel, not a new world

Growth investor Activant's three-part series is the most sceptical serious work. Part one (June 2026) sets McKinsey's $3 trillion to $5 trillion global forecast and Morgan Stanley's $190 billion to $385 billion US forecast against its own scenario of about 4% of US retail by 2030. It cites a Bain survey in which 24% of consumers were comfortable buying with AI, 10% had done so and 89% double-checked what AI recommended. It calls taste-driven categories such as fashion technically the hardest, and says that in luxury fashion and fine jewellery agents play a support role at best.

Part three (July 2026) gives brands a playbook under the line "Market to People, Sell to Software": collect first-party data on taste and fit, treat fit and visualisation tools as conversion infrastructure, and, for luxury and fine jewellery, use agents for concierge, authentication and after-sales service rather than as the checkout.

The rails: General Catalyst, M13, Lightspeed and CB Insights

General Catalyst wrote in December 2025 that "the infrastructure enabling this shift doesn't exist yet", and put the value in machine-readable catalogues, live stock and price APIs, and programmable payment controls. M13's new rails of agentic commerce (February 2026) notes, citing Adobe, that traffic from AI tools to merchant sites rose about 700% in the 2025 holidays, while physical retail is still around 80% of global commerce. In an interview summary, Lightspeed's Aaron Frank called agent-led commerce "just inning one", with payment rails and merchant verification of agents unsolved. CB Insights' agentic commerce market map (November 2025) lists more than 90 companies across discovery, consideration, checkout and service.

Forerunner: "do it for me"

Forerunner, which co-led Daydream's seed, framed the consumer shift in 2024 as moving from "Access to Edit": shoppers tired of endless choice want services that filter and act for them, built on two things, magic and trust.

How strong is the evidence behind the theses?

Read the investor case for what it is. Most of it is argument, supported by third-party forecasts and company-reported engagement. Very little is measured revenue.

The big numbers are borrowed. McKinsey's $3 trillion to $5 trillion appears in Bessemer, CB Insights, M13 and Activant. It is a forecast, repeated, not a measurement.
Memos lead with engagement. Outfits made a day, monthly users and brand partners are what early consumer companies can show. They are not conversion, margin or returns.
Research use is high; delegated buying is low. a16z's Big Ideas for 2025 said 60% of US consumers had used a chatbot to research or decide on a purchase in the past 30 days, without naming the survey. Against that, Bain (via Activant) found 10% had bought with AI, and Modern Retail reported a Forrester survey in which only about a third of consumers would pay through an answer engine.
Conversion and returns figures are vendors' own. Claims such as Botika's 90% cut in shoot costs, or try-on vendors' conversion lifts quoted in investor essays, have not been independently audited.
Investors say so themselves. In a September 2026 Newcomer survey of investors, agentic commerce was "still a rounding error", and Forerunner's Kirsten Green said it was too early for hard data. The same piece reports that OpenAI pulled back its Instant Checkout in March 2026 to focus on discovery.

The best-funded consumer fashion agent is a useful signal. Daydream raised a $50 million seed in 2024 and launched a consumer beta in 2025; in July 2026 it launched "Powered by Daydream", putting its natural-language search on brands' own sites, with STAUD and Alice + Olivia piloting and Sandro and Maje among those signed. Whatever the reasons, the direction of travel was towards the brand's own channels.

The macro sceptics are still worth reading. Sequoia's David Cahn asked in 2024 how the industry would earn back its infrastructure spend in "AI's $600B Question". Menlo's spend figures show the money is now flowing into applications; whether it returns more than it costs is still being proven, use case by use case.

Where do investors disagree?

Question	One side	The other side
Will fashion go agentic first?	Bessemer: apparel shows the most early traction	Activant and VMG: inspiration-led and luxury purchases close on brand channels
How big is it by 2030?	$3 trillion to $5 trillion globally (McKinsey, widely cited)	About 4% of US retail (Activant); "a rounding error" today (Newcomer)
Who captures the value?	Startups win the application layer (Menlo: 63% share)	Incumbents own the rails: Shopify, Stripe, Google, OpenAI, card networks (Bessemer, M13)
Block or partner with agents?	Shopify and Instacart integrate (a16z)	Amazon blocks third-party agents; brands with loyal customers resist (a16z, VMG)

There is also an integrity question that affiliate-funded agents have not settled. Newcomer reports that Phia faces a cookie-stuffing allegation first reported by Bloomberg, which Phia disputes. Whether or not it holds, it shows why a brand should ask whose interest a third-party shopping agent serves.

What should a fashion or jewellery brand take from the investors?

Fix the catalogue first. Every thesis agrees that structured, attribute-rich product data decides whether agents can find and recommend you: fabric, silhouette, fit, occasion, metal, purity, stone, care. It is also what an in-store or on-site stylist runs on. We cover this in part eight.
Own the taste and fit signal. Activant and Menlo locate the moat in knowing what a shopper likes, owns and fits. A brand that captures this first-party, with consent, holds the asset rather than renting it from an agent. The store is the richest place to capture it, because shoppers try things on there; see the physical store data gap.
Plan for two storefronts. Persuade people; inform agents. Watch how much traffic AI tools already send you and what they say about your products.
In fine jewellery and luxury, keep the sale with your people. The investors most sceptical of agentic checkout agree that agents earn their place in concierge, appointments, authentication and after-sales service.
Put early money where brands already pay. Design-to-render, on-model imagery and product content have the clearest return, and are the easiest to test. Test them with control groups, because the conversion claims you will hear are vendors' own; part nine shows how.

Ateliq is built around the second and fourth points: its stylist answers from the store's own stock, sizes and prices, records what shoppers ask for and cannot find, and hands the sale to the store's staff, with opt-in follow-ups only for shoppers who confirm by code. That is one way to own the signal. The wider lesson from the investors stands on its own: whoever holds the richest taste and fit data in a vertical is best placed, and a brand with stores is already sitting on it.

Next in the series: what fashion houses, jewellers and retailers are actually doing.

Fashion AI and the investors: where the money and the theses point