Which AI shopping startups are making the most money today?
In our AI shopping market deck, you will find everything you need to understand the market
SUMMARY
Bloomreach is making the most money among private AI shopping companies today, with more than $260 million in ARR; Algolia is close at roughly $250 million if we use a broader definition that includes substantial non-commerce search.
The leaderboard is dominated by retailer infrastructure, not the consumer AI shopping apps getting most of the attention. Search, personalization, merchandising and product-discovery software already support hundreds of large retail contracts, while consumer agents are still trying to turn fast user growth into dependable revenue.
Constructor may be the most important company we cannot rank precisely. It does not disclose current revenue, but 82% customer growth, 96% gross revenue retention and 322 billion annual shopping interactions make it hard to treat the company as anything other than a top-tier pure ecommerce AI business.
The next group is already substantial. Stylitics and Syte appear to be tens-of-millions businesses, while Nosto and Lily AI are commercially meaningful but harder to place because the clean revenue disclosures are thinner.
The biggest analytical trap is confusing retailer economics with vendor revenue. Stylitics processes more than $60 billion in transaction data, Constructor has documented tens of millions of dollars in added customer sales, and Lily AI has published roughly $80 million of incremental retailer revenue from experiments; none of those figures is the software company's own revenue.
Phia gives us the clearest view into consumer-agent monetization. Its observed daily revenue has ranged from roughly $10,000 to $80,000 across different periods, which would annualize to about $3.7 million to $29.2 million if sustained — meaningful, but still far below Bloomreach.
Consumer AI shopping economics are harder than the user-growth headlines suggest. At a 5% effective commission, an agent needs about $2 billion of attributable merchandise sales to generate $100 million of revenue, whereas an enterprise vendor can reach the same scale with a few hundred large recurring contracts.
That is why the best-known consumer names are not yet the financial leaders. Daydream has broad fashion coverage and Phia has passed one million users, while Onton has reached more than two million monthly active users, but those adoption figures still do not translate cleanly into company revenue.
The category is also broader than generative-AI shopping assistants. Bloomreach, Constructor, Stylitics, Nosto and Syte were building machine-learning-driven commerce products before the current agent boom, and that head start is exactly why so much of the revenue sits with companies consumers rarely see.
The market could still flip. AI-referred retail traffic is converting far better than it did a year earlier, and consumer agents can improve their economics through checkout, payments, ads, loyalty or retailer software. For now, though, the money is still concentrated in the infrastructure quietly deciding what shoppers see and buy.
This market map, featured in our AI shopping market deck, highlights top companies and startups in the AI shopping market
Which AI shopping startups are making the most money today?
Why is AI shopping suddenly worth taking seriously?
AI shopping is becoming a real ecommerce channel because people arriving from AI assistants are increasingly turning into buyers, not just curious visitors.
Adobe's latest retail data makes the change hard to dismiss. By July 2026, traffic from AI sources to U.S. retail websites was still 62% higher than a year earlier. More importantly, those visitors converted 60% better than non-AI traffic. It was the 11th straight month in which AI-referred visitors converted better.
The trajectory is more revealing than either percentage on its own. Adobe had found that AI traffic converted 38% worse than normal traffic in March 2025. By March 2026 it converted 42% better, and the advantage had widened to 60% by July. Over roughly a year, AI referrals went from unusually weak ecommerce traffic to unusually valuable traffic.
We still should not confuse rapid growth with market dominance. Google, Amazon, direct traffic, social platforms and traditional performance marketing remain much larger sources of ecommerce demand. What has changed is that AI shopping can now be measured in purchases. Retailers have a much stronger reason to pay companies that improve how products appear, rank and convert inside AI-driven shopping journeys.
What actually counts as an AI shopping startup?
For this comparison, an AI shopping startup needs to make a meaningful part of its business from helping people discover, compare, personalize or buy products.
That definition includes two groups that are easy to mix up.
Companies such as Bloomreach, Constructor, Stylitics, Syte and Nosto sell AI search, recommendations, merchandising and personalization directly to retailers. Consumers may never recognize their names, but their software decides which products millions of shoppers see.
Then there are consumer-facing companies such as Phia, Daydream and Onton. These businesses want shoppers to start with an AI assistant, describe what they want and then move toward a purchase.
We do not count every large AI company that happens to support shopping. Perplexity, OpenAI and Google all matter to AI commerce, but shopping represents only one part of much larger businesses. Including them would answer a different question.
Public companies also stay outside the main ranking. Coveo, for example, generated $148.3 million of revenue in fiscal 2026 and has a large AI-commerce business, but it is already publicly traded. It works better here as a useful benchmark for private companies.
As this chart shows, and as featured in our AI shopping market deck, search interest in AI shopping has grown significantly
Why is ranking AI shopping startups by revenue so messy?
AI shopping companies regularly publish huge commerce numbers that have almost nothing to do with the revenue they actually keep.
GMV, sales influenced, transaction data processed, retailer revenue lift and ARR are completely different measurements.
Stylitics says its platform processes more than $60 billion of annual transaction data. That does not mean Stylitics earns $60 billion.
Constructor says one optimization program added $35 million of revenue for Belk. Constructor did not collect that $35 million either.
Phia said the retailers connected to its platform represented billions of dollars in annual GMV. Again, that describes merchandise moving through those retailers, not Phia's own sales.
The distinction gets especially important with affiliate-funded shopping agents. If an AI shopping app influences $100 million of purchases and earns an effective 5% commission, its revenue is closer to $5 million than $100 million.
For this article, reported revenue and ARR therefore carry the most weight. External estimates come next, while GMV and customer sales are useful mainly for understanding commercial reach.
Is Bloomreach the biggest private AI shopping company by revenue today?
Bloomreach is the clearest revenue leader we found among private companies whose core business is closely tied to AI-powered commerce and personalization.
Bloomreach said in February 2026 that it had passed $260 million in annual recurring revenue during 2025. The company also finished the year with positive free cash flow and record net-new ARR, which gives the number more weight than a vague growth claim.
Its scale extends well beyond one AI feature. More than 1,400 brands use Bloomreach, and its Loomi platform now covers ecommerce search, personalization, marketing automation and AI agents. Nearly half of Bloomreach customers had adopted at least one of its newer agent or next-generation AI products when the company announced the revenue milestone.
The newer products appear to be producing measurable ecommerce results too. At Bloomreach's 2026 product event, the company said Bash had increased conversion by 35% and revenue per visitor by 40% using its conversational agent. Its adaptive-search beta was averaging a 4.5% increase in revenue per visitor.
Bloomreach stretches the meaning of “AI shopping startup” slightly because it predates the generative-AI boom and operates beyond product discovery alone. Yet if the question is which private AI company heavily involved in shopping already makes the most money, $260 million-plus ARR gives Bloomreach the strongest documented case.
If you want more recent data on this point, please see our latest AI shopping market report.
This chart, featured in our AI shopping market deck, illustrates yearly VC funding for AI shopping startups
Is Algolia making almost as much money as Bloomreach?
Algolia appears to be roughly the same size as Bloomreach today, although we have less confidence in the exact number and its business is less purely focused on shopping.
Sacra estimates that Algolia reached about $250 million in ARR in July 2026, compared with $230 million at the end of 2025 and $210 million in 2024.
That progression implies roughly 19% ARR growth over about a year and a half. For a company already above $200 million, that is meaningful growth rather than a small-company percentage jump.
Algolia also operates at enormous technical scale. More than 18,000 businesses use the platform, and the company says its infrastructure handles more than 1.75 trillion queries annually. Retailers use Algolia for search, merchandising, recommendations and AI product discovery.
The caveat is simple: Algolia also powers search inside software applications, media sites and other non-retail products. We cannot treat all $250 million as AI-shopping revenue.
So Algolia belongs near the top if we ask which private AI businesses participating in shopping make the most money. Bloomreach remains the cleaner number-one answer for a commerce-focused ranking because its $260 million figure comes directly from the company and a larger share of the business revolves around personalization and commerce.
| Company | Best current revenue evidence | How clean is the AI-shopping fit? |
|---|---|---|
| Bloomreach | $260M+ reported ARR | Very strong, though broader than product discovery |
| Algolia | ~$250M estimated ARR | Strong commerce business, but significant non-commerce usage |
| Constructor | Exact current revenue undisclosed | Extremely strong pure ecommerce fit |
How big is Constructor now?
Constructor looks like one of the biggest pure-play AI ecommerce companies in the market, even though the company still does not give us an exact current revenue figure.
The company has disclosed enough growth data to show that this is already a serious business. Constructor doubled revenue and ARR for three consecutive fiscal years through FY23 and said revenue nearly doubled again in FY24.
More recently, Constructor reported 82% customer growth during FY26, while EMEA revenue rose 116%. Gross revenue retention was 96%, meaning very little existing revenue disappeared during the year.
Its platform also handled 322 billion product-discovery interactions, up 266% in two years. That works out to more than 10,000 personalized shopping interactions every second.
Constructor works with hundreds of ecommerce companies across apparel, grocery, beauty, furniture, general retail and B2B distribution. Its customer list includes names such as Sephora, Petco, Under Armour and Target Australia.
The customer results explain why retailers keep paying. Constructor has published cases showing a $35 million revenue gain, a 47% increase in revenue per visitor and returns above 21 times the retailer's investment. Its AI Shopping Agent has also produced reported improvements including a 52% increase in add-to-cart rate and a 56% increase in purchase rate in individual deployments.
We would rather leave Constructor's revenue blank than manufacture precision from an old database estimate. The useful conclusion is still strong: Constructor belongs near the top of the private AI-shopping market and is probably the most important company in the ranking whose business is almost entirely built around ecommerce product discovery.
If you want more recent data on this point, please see our latest AI shopping market report.
This chart, featured in our AI shopping market deck, shows why Constructor is winning in AI shopping
How much money are Stylitics and Syte making from AI fashion shopping?
Stylitics and Syte appear to have built genuine tens-of-millions businesses from AI-powered fashion and retail discovery, although the financial disclosure is much thinner than at Bloomreach.
Stylitics currently works with more than 150 enterprise retailers and says its systems process over $60 billion of annual transaction data and roughly 30 billion product-page views a year. More than 5,000 brands sit inside its data ecosystem.
Outside databases have generally placed Stylitics around the $40 million-to-$55 million annual-revenue range. We would treat that range as directional because Stylitics has not publicly confirmed a current figure.
The operating scale nevertheless makes an eight-figure business believable. Stylitics has spent more than a decade building automated outfitting, visual content, catalog intelligence and personalization products. It says retailers using its outfitting technology see an average 21% lift in order value.
Syte gives us a more explicit clue. Its own company profile says it generates “tens of millions of dollars in annual revenue,” employs about 150 people and powers product discovery for more than 100 major brands.
That disclosure is unusually useful because it tells us the right order of magnitude without pretending we know whether Syte generated $25 million, $35 million or $45 million.
Both companies remain well below Bloomreach's scale, but they are already much bigger revenue businesses than most of the consumer AI shopping agents receiving more attention today.
| Company | Revenue indication | Commercial footprint |
|---|---|---|
| Stylitics | Roughly $40M–$55M from outside estimates | 150+ enterprise retailers, $60B+ transaction data processed yearly |
| Syte | “Tens of millions” reported by company | 100+ major brands |
| Nosto | Roughly ~$20M from outside estimates | 1,500+ brands across 100+ countries |
Is Nosto actually making meaningful money from AI shopping?
Nosto is already a meaningful AI-commerce software business, although its revenue appears to sit well below the market leaders.
Outside estimates put Nosto around $20 million of annual revenue. The exact number deserves caution, so the more interesting evidence comes from how rapidly its AI search product has spread inside its existing customer base.
Nosto reported that searches powered by its personalized-search technology increased 323% during 2024. The number of brands using Nosto for onsite search rose 85% over the same period.
Revenue generated for retailers through that search product jumped 1,024%. That last percentage describes customer sales rather than Nosto's own revenue, but the combination is useful: more retailers adopted the product, shoppers used it much more frequently, and the sales associated with those searches rose even faster.
Nosto now supports more than 1,500 brands in over 100 countries. That customer footprint makes the company considerably more established than the newer AI shopping apps, even if consumers rarely see the Nosto name.
The best way to place Nosto today is in the second tier of AI-shopping companies: a real eight-figure software business, but nowhere close to Bloomreach or Algolia in absolute revenue.
This chart, featured in our AI shopping market deck, illustrates yearly funding for AI shopping startups
Is Lily AI's huge retail revenue number actually Lily AI revenue?
No. Lily AI can generate very large sales gains for retailers while collecting only a much smaller software fee itself.
This is one of the easiest places to overstate an AI-shopping company's financial size.
Lily AI published results from four retail experiments that together produced roughly $80 million in incremental retailer revenue over 28 days. One luxury retailer attributed about $22 million of additional sales to an 8% lift. Another experiment linked Lily's optimization to approximately $50 million of incremental sales after ROAS improved 15%.
Those are impressive commercial results. They still belong on the customers' income statements.
Lily AI has also reported a 28% revenue increase in a controlled Google Shopping experiment and a 28.3% increase in onsite-search revenue for another retailer.
Public company databases generally place Lily somewhere in the eight-figure annual-revenue range, but their estimates vary too much for us to rank Lily confidently against Nosto, Syte or Stylitics.
So Lily clearly belongs in the commercial part of the AI-shopping market. We just do not have enough reliable financial disclosure to pretend we know its exact place.
Are consumer AI shopping agents making serious revenue yet?
Consumer AI shopping agents are attracting users much faster than they are catching the established B2B companies in revenue.
The gap is large.
Phia crossed one million users within ten months of launching. Onton went from roughly 50,000 monthly active users to more than two million. Daydream launched with more than 8,000 fashion brands accessible through its search product.
Those are impressive distribution numbers for young companies.
Their monetization model is tougher. A consumer shopping agent generally helps someone find a product, sends the shopper to a retailer and keeps a percentage of the resulting sale. That means the app needs enormous purchase volume before its own revenue becomes large.
A 5% effective commission illustrates the problem. To earn $10 million, the agent needs roughly $200 million of attributable sales. Getting to $100 million of revenue would require about $2 billion.
Enterprise AI-search companies have a shorter route. A few hundred large retail clients paying sizeable annual contracts can support tens of millions of recurring revenue without the vendor having to capture billions of dollars of consumer purchases itself.
That difference still explains most of the revenue gap between the older commerce-AI platforms and today's consumer shopping agents.
If you want more recent data on this point, please see our latest AI shopping market report.
This chart, featured in our AI shopping market deck, compares the main business model options for AI shopping assistants
How much money is Phia actually making now?
Phia is probably the consumer AI shopping startup with the clearest evidence of meaningful monetization, but recent reporting also shows how far it remains from the biggest enterprise players.
Phia raised $35 million at a $185 million valuation after passing one million users and partnering with more than 6,200 retail brands. At the time, Phia said its revenue had increased elevenfold since launch and that the platform was sending millions of dollars of sales to retailers every month.
The company did not disclose its own revenue in dollars.
A much more revealing number emerged later. Bloomberg reviewed an internal Phia revenue chart showing average daily revenue of around $80,000 before the company disabled affiliate-monetization features that had become controversial.
After those changes, the internal chart showed daily revenue falling to roughly $10,000 to $28,000. Phia said the decline was partly caused by shutting down most monetization during that period and disputed an internal calculation of how much merchandise value had been attributed to problematic cookie behavior.
The range still helps us understand scale. Sustained revenue of $80,000 a day would annualize to about $29 million. A sustained $10,000-to-$28,000 daily range would imply roughly $3.7 million to $10.2 million.
We should not call any of those figures Phia's annual revenue because daily monetization was changing rapidly. But they are far more informative than an “11x” growth rate with no starting number.
As seen above, Bloomreach is above $260 million ARR. Even Phia's earlier $80,000-a-day level would annualize to barely one-ninth of that amount. The consumer-agent market is commercial now, but Phia's numbers show how wide the gap still is.
| Phia revenue evidence | Annualized equivalent if sustained |
|---|---|
| ~$80K average daily revenue before monetization changes | ~$29.2M |
| ~$28K daily revenue after changes | ~$10.2M |
| ~$10K daily revenue after changes | ~$3.7M |
Is Daydream making much money from AI fashion shopping yet?
Daydream is monetizing AI fashion shopping, but we still do not have solid evidence that it has become a large revenue business.
The company raised a huge $50 million seed round before opening its shopping product broadly. When Daydream launched publicly, shoppers could search products from more than 8,000 brands using natural language and images.
Daydream does not complete the transaction itself. When shoppers choose a product, they move to the retailer's website, and Daydream takes a percentage of the sale.
That model can scale, but the company has never disclosed a reliable current revenue figure. Some commercial databases put Daydream in the low-single-digit millions annually, yet we could not verify those estimates from the company.
Daydream has begun broadening its model by taking its search technology directly to fashion retailers. That could eventually give it recurring enterprise revenue alongside consumer commissions, which would make the economics more attractive.
Today, though, Daydream belongs in the emerging category. Its funding, brand coverage and technology are much easier to verify than its revenue.
This chart, featured in our AI shopping market deck, shows how market revenue is split across customer segments in the AI shopping market
Does Onton's two million monthly users mean it is already a big business?
Onton's two million monthly active users make it a significant AI shopping product, but they do not tell us whether Onton is making significant money.
The growth itself is striking. Onton, previously called Deft, said monthly active users increased from roughly 50,000 to more than two million while the company was still focused heavily on furniture shopping. That is roughly 40 times the original user base.
Onton then raised another $7.5 million and started expanding into categories such as apparel and consumer electronics.
What we still lack is the information needed for a serious revenue calculation: purchase volume, conversion, commission rates and the proportion of transactions from which Onton actually earns money.
Two million shoppers who browse extensively but rarely purchase through monetized links could support a surprisingly small business. A smaller audience with high purchase intent and valuable baskets could be much more lucrative.
So we would rather leave Onton financially unranked for now. Its usage deserves attention, while its revenue remains unknown.
Why are the biggest AI shopping companies mostly names consumers never hear?
Retailers have been paying for AI product discovery for years, which gave B2B companies a huge head start over consumer shopping agents.
Look at the volume running through the underlying software.
Constructor powered 322 billion shopping interactions during its latest fiscal year.
Stylitics says its technology sees about 30 billion product-page views annually and processes more than $60 billion in transaction data.
Algolia handles more than 1.75 trillion queries a year across its overall platform.
A company serving large retailers can make substantial recurring revenue from a few hundred contracts. The software only needs to improve conversion or average order value slightly for the economics to work.
Constructor's published $35 million increase in Belk revenue shows the scale of the opportunity. Stylitics reports average order-value improvements around 21% from its outfitting technology. Bloomreach recently reported a 35% conversion gain and a 40% rise in revenue per visitor in one conversational-agent deployment.
When retailers can measure improvements that large, paying six or seven figures for the underlying software becomes much easier to justify.
Consumer shopping apps have a more visible brand but a harder revenue equation. They usually need millions of people to browse, click and eventually buy before affiliate commissions add up to the same amount.
This chart, featured in our AI shopping market deck, shows how AI shopping assistant technology has evolved over time
Are Bloomreach, Constructor and Stylitics really AI companies?
Yes. Their companies may be older than the current AI-agent boom, but AI already sits deep inside the products retailers are paying for.
Constructor uses machine learning, reinforcement learning, behavioral data and large language models to rank products and handle conversational shopping queries. Its AI Shopping Agent now lets shoppers ask questions such as what to wear to a particular event and then receives personalized, in-stock suggestions.
Stylitics spent more than a decade building automated outfitting and product-matching systems. Its current platform combines styling, image generation, catalog enrichment, personalization and retail intelligence.
Bloomreach has also shifted more of its product around Loomi AI. Nearly half its customers had adopted at least one newer agent or next-generation AI capability by the time Bloomreach announced its latest ARR milestone, and the number using at least four such products had quadrupled in a year.
The founding year tells us less than the current product and revenue mix. Excluding every company created before generative AI would remove many of the places where AI shopping has already reached commercial scale.
Does all the funding going into AI shopping agents mean they will soon lead in revenue?
No. Funding shows how aggressively investors are betting on future AI shopping revenue, and the current financial rankings still look very different.
Daydream raised $50 million before its consumer product was broadly available. Phia reached a $185 million valuation less than a year after launch.
Compare those numbers with the companies already collecting software revenue. Syte says it makes tens of millions of dollars annually. Constructor has spent years compounding revenue and now serves hundreds of major ecommerce businesses. Bloomreach is above $260 million ARR and already free-cash-flow positive.
Phia makes the disconnect particularly easy to see. Investors valued the company at $185 million while its later internal daily-revenue figures implied anything from several million dollars to roughly $29 million on an annualized basis, depending on which monetization period we examine.
That valuation could still prove sensible if Phia becomes a major shopping layer. It tells us very little about who makes the most money today.
Funding is a bet. Revenue is the score we can already see.
If you want more recent data on this point, please see our latest AI shopping market report.
In our AI shopping market deck, we identify pain points entrepreneurs should prioritize
Can consumer AI shopping agents eventually become bigger than Bloomreach and Constructor?
Consumer AI shopping agents could eventually become much bigger businesses, but they will probably need to own more of the transaction than most of them do today.
The demand side is becoming more convincing. Adobe's latest data found that AI-referred retail visitors converted 60% better than non-AI visitors by July 2026. That advantage had persisted for 11 consecutive months.
People are clearly becoming comfortable asking AI what to buy.
The bigger question is who captures the economics. Today, Daydream redirects users to merchant websites. Phia earns through commerce and affiliate relationships. Many other shopping assistants follow similar models.
Commissions can produce large companies, but the required merchandise volume becomes enormous. A shopping agent earning 5% effectively needs $2 billion of attributable purchases to make $100 million of revenue.
Integrated checkout, payments, advertising, loyalty programs or merchant software could improve that equation considerably. Daydream's move into retailer-facing search is one example of how consumer companies might add recurring B2B revenue instead of relying entirely on affiliate commissions.
The consumer-agent opportunity looks real. It just has not caught the enterprise AI-commerce businesses financially yet.
Which AI shopping startups are actually making the most money today?
Bloomreach currently has the strongest documented claim to being the highest-revenue private AI shopping company, while Algolia is close enough to matter if we use a broader definition of the category.
Bloomreach has reported more than $260 million in ARR. Sacra's latest estimate puts Algolia around $250 million ARR, although part of Algolia's business comes from non-commerce search.
Constructor belongs immediately behind that pair in commercial importance, but we cannot responsibly give it an exact current revenue number. Its repeated revenue growth, 82% customer expansion, 96% gross revenue retention and 322 billion annual shopping interactions clearly place it far beyond the experimental stage.
Stylitics and Syte form another serious group measured in tens of millions of dollars. Nosto and Lily AI appear to sit lower in the eight-figure range, with less precise financial disclosure.
Then comes the consumer-agent wave. Phia has the best hard monetization evidence we found, yet its recently observed daily revenue levels still imply a business much smaller than Bloomreach. Daydream is monetizing but has disclosed little. Onton has millions of users without enough financial data to rank.
The current leaderboard therefore looks very different from the companies getting the most consumer attention.
| Rank / tier | Company | Best revenue evidence today | Our read |
|---|---|---|---|
| 1 | Bloomreach | $260M+ reported ARR | Strongest documented private AI-commerce leader |
| 2 if broadly included | Algolia | ~$250M estimated ARR | Similar scale, but not entirely commerce |
| Top tier | Constructor | Current revenue undisclosed | Likely one of the biggest pure ecommerce-AI companies |
| Next tier | Stylitics | Roughly $40M–$55M estimated | Large vertical retail-AI business |
| Next tier | Syte | “Tens of millions” reported | Proven commercial scale |
| Next tier | Nosto | Roughly ~$20M estimated | Established personalization and AI-search business |
| Next tier | Lily AI | Likely eight figures | Commercially meaningful, exact ranking uncertain |
| Consumer leader | Phia | Recent daily figures imply roughly $3.7M–$29.2M annualized depending on period | Most concrete consumer-agent monetization evidence |
| Emerging | Daydream | Undisclosed | Early commission and B2B monetization |
| Emerging | Onton | Undisclosed | Large usage, financial scale unclear |
This chart, featured in our AI shopping market deck, shows how market revenue is split across Europe, Asia, North America, Africa, and South America in the AI shopping market
So who is making the most money from AI shopping right now?
Bloomreach is our clearest answer today, with more than $260 million in ARR, followed closely by Algolia if we allow a broader AI-search company into the category.
Constructor looks like the strongest pure ecommerce-AI challenger below them. Stylitics, Syte, Nosto and Lily AI have already built meaningful software businesses, generally in the tens-of-millions range rather than the hundreds of millions.
The new consumer shopping agents are much earlier financially. Phia is the most interesting because we have unusually concrete revenue evidence, but even its strongest observed daily-revenue level would annualize to around $29 million. Its lower post-change range works out to only about $4 million to $10 million.
So the money in AI shopping today still sits mostly inside retailer infrastructure: search engines, recommendation systems, personalization tools, merchandising software and product intelligence.
Phia, Daydream, Onton and other consumer agents may eventually flip that ranking. They have the user growth and AI shopping behavior is moving quickly in their favor. For now, though, the companies quietly deciding which products appear on retailers' websites are making far more money than the apps trying to become everyone's personal AI shopper.
If you want more recent data on this point, please see our latest AI shopping market report.
OUR METHODOLOGY
This analysis ranks private AI shopping companies by the strongest recent evidence of revenue scale. We compare reported revenue or ARR, credible outside estimates, commercial adoption, monetization economics and how directly each company's core business is tied to shopping.
Directly reported revenue and ARR carry the most weight. When a company does not disclose a current figure, we use outside estimates and operating data to establish an order of magnitude without turning a rough estimate into a precise ranking.
We keep company revenue separate from GMV, transaction data, retailer sales influenced and customer revenue lift. Those figures can show reach or economic impact, but they are not interchangeable with the revenue the AI shopping company actually keeps.
For fast-changing monetization figures such as Phia's reported daily revenue, we show annualized equivalents only to make the scale easier to compare. We do not treat those annualized figures as reported annual revenue.
The main ranking excludes public companies and large general-purpose AI businesses where shopping is only one part of a much broader product. Companies such as Coveo are useful benchmarks, while Perplexity, OpenAI and Google sit outside the private AI-shopping startup ranking.
Key sources include Adobe Digital Insights on AI-referred retail traffic and conversion, Bloomreach on $260 million-plus ARR and AI-product adoption, Bloomreach on results from Bash and adaptive search, Algolia on platform scale and agentic commerce, Constructor on FY26 growth and 322 billion shopping interactions, Constructor's Belk case study, Stylitics on retailer footprint and transaction-data scale, Syte on its annual-revenue range and brand footprint, Nosto on personalized-search adoption, Lily AI on incremental retailer-revenue experiments, Phia on its funding, valuation, users and retailer network, Fortune on Phia's reported daily-revenue data and affiliate changes, TechCrunch on Daydream's launch and commission model, Daydream on its retailer-facing search product, and TechCrunch on Onton's user growth and funding.
The final ranking aggregates those signals rather than pretending the market offers one clean financial leaderboard. Where the evidence supports a direct ranking, we use one. Where it does not, we keep the company in a broader tier.
This chart, featured in our AI shopping market deck, illustrates yearly VC funding for AI shopping startups
Related blog posts
