CommerceClarity Acquires Katalogo.ai
Two deals crossed my radar recently. Different companies, different stages, different parts of the market. Same bet. That is why I am writing about them together.
CommerceClarity, an AI platform for retail catalog operations, acquired Barcelona-based Katalogo.ai in early June. CommerceClarity had raised €2.7 million six months earlier and is used by retailers and brands including Prada, STIGA, and Cisalfa Sport. Katalogo.ai had been building AI-powered styling agents that analyze product catalogs and generate outfit and product recommendations. Its founder, Luca Cozzolino, joined CommerceClarity as chief product officer after previous product leadership roles at Shopify and Zalando.
This is not a marketing acquisition. It is an infrastructure acquisition.
CommerceClarity’s core argument is that structured, consistent product data is the prerequisite for almost everything AI does downstream. What appears in an AI shopping search. What gets recommended beside a product. What feeds a digital lookbook. What a “complete the look” module can actually complete without creating a small visual crime scene.
Add styling logic on top of catalog intelligence and the system can understand more than what a product is. It can begin to understand where that product belongs, what it works with, and how it might live inside a wardrobe.
Vêtir Raises $5.5 Million at a $150 Million Valuation
Vêtir, founded by Kate Davidson Hudson, looks at first like a luxury consumer app. It uses AI for wardrobe management, styling, shopping, visual search, virtual try-on, and closet organization. But the company is also building tools for stylists and luxury clienteling teams, which is where the story gets more interesting for brands.
In May, Vêtir announced the first close of a $5.5 million Series A at a $150 million valuation. The company reported more than $2,500 in average order value, 200% month-over-month organic user expansion, nine-times year-over-year revenue growth, and more than 3,500% growth in B2B clients. Those are company-reported metrics, not independently audited figures, but they explain why investors are looking at this as more than a styling app.
| REPORTED METRIC | VÊTIR | WHY IT MATTERS |
|---|---|---|
| Average order value | $2,500+ | High-intent luxury commerce, not casual browsing. |
| Organic user expansion | 200% month over month | Reported acceleration in consumer adoption. |
| Revenue growth | 9× year over year | Commercial momentum beyond app engagement. |
| B2B client growth | 3,500%+ | The stylist and clienteling layer may be the larger strategic play. |
The B2B number is the one I keep coming back to. A wardrobe platform connected to stylists and clienteling teams can create a view of product performance that most brands do not have today. Not just what was clicked or purchased, but how a product fits into an existing closet, what it gets styled with, and what kind of next purchase makes sense.
That is not a campaign asset. That is relationship data.
Different Entry Points, Same Destination
CommerceClarity is approaching from the catalog side. Vêtir is approaching from the wardrobe and clienteling side. Both are building the layer between a product and the way a customer actually wears, combines, and understands it.
That layer is strategically valuable because it compounds. Better product data improves recommendations. Better recommendations create richer behavior signals. Better behavior signals improve styling and personalization. The system gets more useful because the data gets more useful.
That is a much harder advantage to copy than a prettier AI campaign image.
The Campaign Layer Matters. It Just Is Not the Whole Story.
Most fashion teams I speak with still meet AI through the campaign layer. Image generation. Caption writing. Brief building. Moodboards. Product copy. Those are legitimate tools and some of them save a ridiculous amount of time when used well.
They are also becoming easier to access. The same image model, writing assistant, or prompt structure can be used by your brand and six competitors before lunch. The tool can improve speed, but the tool itself is rarely the moat.
The advantage comes from what sits behind it: your standards, your product knowledge, your customer language, your creative judgment, your workflow, and the data only your company has.
This is also why I do not think brands should dismiss prompt systems or creative AI workflows as temporary. A strong prompt library can turn scattered taste into a repeatable team asset. It just needs to connect to the deeper operating system rather than float around in a folder called “AI stuff.” Our prompt packs are built around that exact problem, and the free Fashion AI Toolkit database is there to help teams find the right tools before buying another subscription they will forget exists.
Can an AI System Actually Understand Your Product?
Not in theory. Concretely.
There is a real difference between tagging a dress “maxi, blue” and describing it with structured attributes that reflect color, silhouette, fabric behavior, occasion, proportion, styling context, climate, price position, and what it pairs with. The first description is enough to store the product. The second gives a recommendation system something useful to work with.
But I would be careful about creating an enormous fantasy taxonomy just because AI is involved. More fields do not automatically create better intelligence. The goal is not to make your team complete 84 columns per SKU. The goal is to define the attributes that matter for discovery, styling, merchandising, customer confidence, and brand differentiation, then apply them consistently.
The brands most exposed are the ones with years of products tagged differently across PLM, PIM, Shopify, wholesale files, spreadsheets, and agency feeds. Often nobody owns the entire language system. Design calls the color one thing. Ecommerce calls it another. Paid media invents a third. Then everyone wonders why search, recommendations, and reporting feel slightly drunk.
Five Moves That Are Actually Worth Making
- Audit your current product attributes. Identify missing fields, duplicate terminology, inconsistent naming, and the places where useful product knowledge disappears during handoff.
- Choose the use cases before expanding the taxonomy. Search, styling, recommendations, AI shopping visibility, PDP confidence, and merchandising may require different levels of detail.
- Create one approved language system. Design, merchandising, ecommerce, marketing, and wholesale should not maintain separate dictionaries for the same product.
- Connect creative judgment to the data. A taxonomy built only by technical teams may be accurate and still miss how customers understand style, fit, mood, and occasion.
- Test on a contained assortment. Start with one category or seasonal capsule, then evaluate whether the richer data improves discovery, styling output, internal speed, or conversion.
This is the kind of problem that sits between creative direction, product storytelling, ecommerce, and AI workflow design. Flume and Field helps apparel teams identify the useful system, build the language and guardrails around it, and make it usable by the people who actually have to do the work. See the current ways to work together.
The window is still open. The point is not to panic-build an AI infrastructure project because two companies raised money. It is to notice where serious investment is going and make sure the foundation of your own brand is not held together by old spreadsheets, tribal knowledge, and one person who knows why “navy 3” is different from “dark navy final.”
That person deserves a vacation.