Michelle Jones

ChatGPT now has a “Try On” button. Upload a selfie or a full-body photo, and the chatbot will show you wearing a jacket, a dress or a pair of sunglasses from its shopping results. OpenAI rolled the feature out globally in early October, along with a Favorites library for saving products. It’s the clearest sign yet that AI style analysis is moving out of niche styling apps and into a general-purpose assistant that many shoppers already use for everything else. That makes it worth being clear about what a picture of you in a new outfit can tell you, and what it can’t.

What OpenAI Actually Launched

According to TechCrunch’s report, the update adds two things to ChatGPT’s shopping experience:

• Try On: A button now appears in shopping results. You supply a photo of yourself, and ChatGPT generates an image of you wearing the clothing item or accessory. You can also upload your own picture of an item, such as a screenshot from a store’s website, and ask ChatGPT to put it on you.
• Favorites: Products you find can be saved to a Library in the app, stored alongside the try-on images you’ve made.

Both run on ChatGPT Images 2.5, OpenAI’s newest image model. OpenAI says it delivers more natural lighting, richer textures, more reliable instruction-following when editing, and faster generation. The new features build on shopping tools ChatGPT already had: you can describe a style and ask it to find the pieces that complete the look, or upload a photo of a celebrity’s outfit and ask for buyable matches.

TechCrunch also notes the context. OpenAI previously moved away from an instant-checkout feature that didn’t perform well, and Google launched its own virtual try-on in July 2025. Whether people will start their fashion shopping in ChatGPT is still an open question.

Why This Matters for AI Style Analysis

Until now, most AI style analysis came from dedicated tools: color-season apps, body-shape scanners, and try-on widgets built into individual retailers’ sites. We’ve covered several, from PointAI’s one-second try-on to DRESSX’s conversion data. ChatGPT changes the scale. A general assistant that can read your style from a photo, find matching products and show them on you puts every step of the process in one chat window.

The catch is that a generated image is a picture, not a measurement. Try On produces a convincing image of you in a garment. TechCrunch’s report says nothing about sizing, fabric drape or fit data, and a model praised for “richer textures” is optimized to look right, which is not the same as being right about how a size M will sit on your shoulders.

Recent testing points the same way. In September we covered a skin-tone benchmark in which no general-purpose model classified skin tone correctly more than about half the time. On October 27, skin-analysis company Haut.AI published SkinBench, which compared six general and medical AI models, including Gemini 3.5 Flash, GPT-5.6 Sol and Claude Fable 5, against expert assessments. The models frequently reported skin concerns that weren’t there; eye puffiness, for example, was flagged in roughly 95% to 100% of images where it was absent. Haut.AI sells its own competing skin AI and ran the test itself, so treat the rankings with caution. The broader pattern still matters for anyone using a chatbot for style advice: a model that can describe a photo fluently is not necessarily measuring what’s in it.

How to Use AI Try-On Without Getting Burned

• Use it for direction, not decisions. Try-on is good at answering “does this color or silhouette suit my overall look?” It isn’t built to answer “will this fit?” Check the retailer’s size chart and measurements before you buy.
• Give it a good photo. Even, natural light and a straight-on, full-body shot give any image model less to guess about. Heavy filters and mixed lighting change how your skin tone and proportions come across.
• Cross-check color advice. If ChatGPT tells you a shade “works with your undertone,” try the color against your face in daylight or in a mirror. Recent benchmarks suggest general models still misread skin.
• Think before you upload. Try-on needs a photo of you, and the images are saved in your Library next to your saved products. Decide which photos you’re comfortable sharing, and review ChatGPT’s privacy settings before you start.

FAQ

How does ChatGPT’s virtual try-on work?

Upload a selfie or full-body photo, then tap the Try On button on a clothing or accessory result in ChatGPT’s shopping results. ChatGPT generates an image of you wearing the item. You can also upload a screenshot of an item from another site and ask it to try that on you.

Can ChatGPT’s Try On tell me what size to buy?

Not based on what OpenAI has described. TechCrunch’s report covers image generation and saving products, not sizing or fit measurements. Use the retailer’s size guide for fit.

Is AI style analysis accurate?

It depends on the task. Recent benchmarks found general-purpose AI models often misread skin tone and report skin features that aren’t there, while specialized tools score better on narrow tasks. Use AI style analysis as a starting point and confirm important choices in person.

Sources & Further Reading: TechCrunch — “ChatGPT can now virtually try on clothes for you” (Sarah Perez, October 1, 2026), https://techcrunch.com/2026/10/01/chatgpt-can-now-virtually-try-on-clothes-for-you/ · Cosmetics Business (promoted content from Haut.AI) — “SkinBench: six AI models saw skin problems that were not there” (October 27, 2026), https://www.cosmeticsbusiness.com/skinbench-six-ai-models-saw-skin-problems-that

By Michelle Jones, Fashion News GF

Gap Inc. is changing how people search its stores online. Instead of typing “black jeans” into a search box and working through filters, Old Navy and Banana Republic shoppers can now describe what they want in plain language and get suggestions back. The company announced the tools at the start of San Francisco Tech Week in early October, about 30 years after Gap launched its first e-commerce site in 1996. Gap is also opening its catalog to two outside AI styling apps, Alta and Daydream. Together the moves show how a mass-market fashion group is betting on AI shopping without throwing out the old way of browsing.

What Gap Actually Launched

According to Retail Dive’s report, there are three parts to the rollout:

• Conversational shopping agents on the Old Navy and Banana Republic websites and apps. Shoppers type a request the way they’d say it to a sales associate, rather than choosing from category menus.
• Alta, a styling app founded by CEO Jenny Wang, builds outfit suggestions around a user’s schedule, budget and the weather forecast. It can mix pieces the user already owns with items from brands’ ranges.
• Daydream, co-founded by CEO Julie Bornstein, answers everyday-language fashion requests. It will be able to pull from every Gap Inc. brand: Gap, Old Navy, Banana Republic and Athleta.

Gap is presenting the tools as an addition to search boxes, filters and product grids, not a replacement for them. Chief technology officer Sven Gerjets told Retail Dive that “AI should make shopping easier and more intuitive, not add complexity or take choice away from the customer.” He also said Gap would learn and adjust if the tools don’t deliver.

Why This Matters More Than Another Chatbot

Most of the AI shopping launches we’ve covered this year came from startups or luxury boutiques. This one is different because of who’s doing it. Old Navy and Banana Republic sell everyday basics to a very broad customer base, and a company that size doesn’t run experiments on its main websites lightly.

The bigger story is distribution. Gap isn’t only building its own assistant. It’s also letting third-party AI stylists show its products. That means a shopper may find an Old Navy sweater through an app that never sends them to Old Navy’s homepage. Daydream, which we wrote about in August when it pitched a “shop with AI” button for fashion brands, now has one of the biggest US apparel groups on board.

The caution is just as telling. Retail Dive points to a SmartCustomer survey in which 45% of purchases consumers regretted making with AI help were apparel and footwear. It also cites research from Synchrony and Oxford Economics: shoppers are more willing to use AI at retailers they already trust, but they worry about how their data is used and want to keep some control. Gerjets’s comments about relevance and not taking choice away seem written with exactly those concerns in mind.

What Shoppers Should Keep in Mind

• Be specific. These tools work from your description. “A warm layer for a rainy office commute, under $60” gives the AI more to go on than “jacket.”
• Still check fit and fabric yourself. In that survey, apparel and footwear made up 45% of the AI-assisted purchases people regretted. Read the size notes and fabric details before you check out.
• Know what you’re sharing. Apps that plan outfits around your calendar and the weather need that information to work. Decide how much you’re comfortable handing over before you connect your accounts.
• The old filters are still there. If the assistant isn’t helping, Gap describes it as an addition to the usual search and filters, not a replacement, so you can still browse the familiar way.

FAQ

What AI shopping tools has Gap launched?

Gap Inc. has added conversational shopping agents to the Old Navy and Banana Republic websites and apps, where shoppers describe what they want in natural language. It is also making its products available through the outside AI styling apps Alta and Daydream.

Which Gap brands work with Daydream?

According to Retail Dive, Daydream will have access to all Gap Inc. brands: Gap, Old Navy, Banana Republic and Athleta.

Is Gap replacing its normal search and filters with AI?

No. Gap describes the AI tools as an addition to its existing search boxes, filters and product grids, not a replacement for them.

Sources & Further Reading: Retail Dive / CX Dive — “Gap unveils AI-assisted shopping, three decades after its first website” (Daphne Howland, October 5, 2026), https://www.customerexperiencedive.com/news/gap-ai-assisted-shopping-three-decades-after-website/832188/

By Michelle Jones, Fashion News GF

Is luxury fashion a good investment? Fashionphile’s new resale data gives a more useful answer than a simple yes or no: the classics are holding up, and the biggest jumps belong to bags that were out of fashion until very recently. The resale platform’s 2026 Ultra-Luxury Resale Report, covered this week by TheIndustry.fashion, compares January to July 2026 against the same months of 2025. The headline is nostalgia. Searches for vintage pieces rose 112% year over year, and a few revived archive bags saw their average sale prices climb by triple digits. Read more closely, though, and the data splits luxury into two very different kinds of “investment”.

The Classics: Steady, Single-Digit Gains

The bags people usually mean when they talk about investment pieces grew at a modest pace. According to the report, the second-hand value of the Chanel Double Flap rose 8.89%, the Hermès Kelly 7.67%, and both the Hermès Birkin and the Cartier LOVE bracelet 6.14% each.

Those are gains, not windfalls. They also come from a single platform over a seven-month comparison, so they tell you about one marketplace’s sales this year rather than a long-run return. Still, they line up with how these pieces behave: dependable demand, predictable pricing, and a resale value that tends to hold rather than spike.

A small number of current styles did better than that. Fashionphile says it paid sellers above the original retail price for a handful of pieces, including the Urban Essential (15% above retail), the Souplissimo and the Archetype Tote (13% above each) and the Preppy Coco (1% above). Note what that figure measures: what Fashionphile paid the person selling, not what the next buyer paid.

Popularity and appreciation aren’t the same thing either. Louis Vuitton was the most-shopped and most-sold brand, and the Speedy topped the most-shopped list for a second year running, ahead of the Chanel Double Flap and the Prada Re-Edition 2005, which climbed from eighth place to third. Goyard’s Saint Louis entered the top ten at seventh. Those rankings show what sells quickly, which matters if you ever want to sell. They don’t, by themselves, show rising prices.

The Revivals: Big Numbers, Bigger Risk

The eye-catching figures belong to archive styles. The report says the average sale price of Saint Laurent’s Mombasa, the horn-handled bag Tom Ford designed in 2001, rose 291%, and the number sold tripled. Fashionphile ties the revival to a recent runway nod from creative director Anthony Vaccarello. The Fendi Spy’s average resale value rose 127%. The Bottega Veneta Veneta Hobo saw searches jump 146% after the brand re-released it in February.

Search interest ran even hotter than prices in places. Searches for the Fendi Baguette rose 157%, but its average sale price rose under 10%. Chanel’s Cambon line saw interest up 148% and prices up 5%. Searches for monogram denim rose more than 1,000%. In other words, a lot of the nostalgia wave is people looking, and only some of it has turned into higher prices so far.

That is the risk for anyone buying a revival as an investment. The Mombasa’s jump followed a single designer’s runway reference. What a runway reference starts, the next season’s references can just as easily end, and the report can’t tell you which of today’s archive favourites will still be wanted in three years.

What This Means for Buyers

Treat luxury as a purchase that may hold its value, not as a portfolio. The data supports a few practical rules:

• Buy classics to use, not to flip. Single-digit gains on a Kelly or a Double Flap are reassuring if you’ll wear the bag for years. They’re thin once you account for selling fees, condition and authentication.
• Buy revivals because you love them. Triple-digit moves on the Mombasa or the Spy show how fast demand can return. They also show how much that demand depends on what’s on the runway right now.
• Watch searches before prices. In this report, interest often ran well ahead of sale prices. A style with soaring searches and flat prices hasn’t proven itself yet.

That last point is where resale data meets the trend-tracking tools we cover here. Search growth, runway references and sell-through are exactly the kinds of signals AI trend forecasting draws on, and as we noted last week, most of the industry still checks those forecasts before acting on them. Shoppers should do the same: a spike in interest is a signal to look closer, not proof of lasting value.

Resale value is also shaping how people shop new. TheIndustry.fashion points to Vinted’s 2025 Impact Report, in which 35% of surveyed UK buyers said they consider an item’s potential resale value before buying new, and to eBay’s Recommerce Report, in which 46% of UK respondents said they buy second-hand at least once a month.

FAQ

Is luxury fashion a good investment in 2026?

For a few classic pieces, it can hold or slightly increase its value. In Fashionphile’s 2026 report, the Chanel Double Flap’s second-hand value rose 8.89% and the Hermès Kelly’s 7.67% over January to July compared with a year earlier. Those are modest gains from one platform over seven months, so it’s safer to think of luxury as a purchase that may keep its value than as an investment.

Which luxury bags saw the biggest resale price increases?

In this report, revived archive styles led. The Saint Laurent Mombasa’s average sale price rose 291%, and the Fendi Spy’s average resale value rose 127%, both driven by renewed nostalgic interest.

Do vintage revivals hold their value?

Sometimes, but they carry more risk than classics. Demand for revived styles often follows runway references and social buzz, which can fade quickly. Rising searches don’t always mean rising prices: the Fendi Baguette’s searches rose 157% while its average sale price rose under 10%.

Sources & Further Reading: TheIndustry.fashion — “FASHIONPHILE resale report finds Y2K revivals driving luxury bag prices higher” (Catherine Rowe-Kosary, October 14, 2026), https://www.theindustry.fashion/fashionphile-resale-report-finds-y2k-revivals-driving-luxury-bag-prices-higher/

By Michelle Jones, Fashion News GF

Nearly everyone in fashion is using AI now. Hardly anyone trusts it with the big calls. That gap is the most useful lens for reading any AI trend forecast in 2026. In the space of two days, The Interline published two pieces from its new AI Report 2026 that point at the same tension. A survey for the report found that roughly nine in ten fashion professionals use AI every day, but only about a quarter trust its output enough to base important decisions on it. In a separate essay, WGSN chief executive Carla Buzasi argued that AI makes trend forecasting faster but can’t do the part that matters most: telling a passing spike from a lasting shift.

The Trust Gap, in Numbers

The headline figures come from The Interline’s AI Report 2026, which drew on around 100 survey respondents across different levels of the industry. The sample is small, so treat the exact percentages as a snapshot rather than a census. Still, the shape is striking: AI has become routine at work, yet most people using it every day still check its answers before acting on them.

Speaking on The Interline’s podcast, Rupert Schiessl, chief strategy and AI officer at sourcing and product-lifecycle software company Bamboo Rose, offered a practical reason for the gap. People trust what they can check, he argued, and a bare confidence score doesn’t give them anything to check. “Providing a chain you can check is much more convincing than a confidence score,” he said. The report also suggests the industry sees AI as most mature in concept development and marketing, and least trustworthy in technical design, sourcing and production, where a wrong answer costs real money.

Trend forecasting sits awkwardly between those two groups. A forecast feels creative, but it drives very concrete buying decisions: how many units, in which colors, landing when.

What a Forecaster Says AI Can’t Do

Buzasi’s essay makes the case from the vendor side, so read it with that in mind: WGSN sells human expertise alongside its AI products. Her core argument is still worth taking seriously. Social platforms produce endless micro-trends, and in her words, “for every emerging behaviour on TikTok that becomes a product shift, there are a hundred that flame out in three weeks.” Spotting what’s viral is easy. Knowing what will still sell next season is the hard part.

She also argues that the earliest signals show up in real life before they show up in any dataset, which is why WGSN describes its network of more than 250 trend experts as the “input layer” of its AI systems rather than an add-on. Then comes the step she calls craft: turning a trend into a product that actually works, which depends on people who know how a fabric behaves after fifty washes.

What changes, in her telling, is speed and access. Forecasts that used to arrive as reports and slide decks now live as structured data that a merchandiser can query directly and get an answer in seconds. That matches what we saw last month when WGSN opened up its runway AI data to clients. Buzasi says those answers are grounded in “25 years of forecast accuracy.” Neither piece explains how that accuracy is measured, and that’s still the question buyers should be asking.

How to Read an AI Trend Forecast

Put the two pieces together and you get a fair working rule for 2026: use AI trend forecasting to see more signals, faster, and keep a person accountable for which ones you act on. Before you lean on any AI forecast, whether it comes from a big subscription service or a free tool, ask three things:

• Where did the signal come from? Runway tagging, social posts, search data and sales data tell you very different things. A forecast that can’t show its sources is the confidence-score problem Schiessl describes.
• How long has it lasted? A spike that’s a few weeks old is exactly the kind Buzasi warns about. Look for signs that it’s persisting across markets or seasons.
• Can it be made and sold? A trend that holds up on social media can still fail on cost, fit or fabric. That’s still a human call.

The takeaway isn’t that AI forecasts are unreliable. It’s that the industry hasn’t settled how to check them yet, and the people using them most are, sensibly, still checking.

FAQ

How many fashion professionals trust AI for important decisions?

According to The Interline’s AI Report 2026, roughly nine in ten fashion professionals surveyed use AI daily, but only about a quarter trust its output enough to base important decisions on it. The survey had around 100 respondents.

Can AI predict fashion trends on its own?

AI is good at spotting patterns across huge volumes of runway, social and search data. Forecasters like WGSN argue it still needs human experts to catch early real-world signals, separate lasting trends from short-lived viral moments, and judge whether a trend can become a product that sells.

What should I look for in an AI trend forecast?

Check where the data came from, how long the trend has persisted, and whether the forecast explains its reasoning rather than just giving a score. Forecasts that show their working are easier to trust and easier to challenge.

Sources & Further Reading: The Interline — “Trend Forecasting In The Age Of AI: Amplification, Impact & Decision Success” (Carla Buzasi, September 30, 2026), https://theinterline.com/2026/09/30/trend-forecasting-in-the-age-of-ai-amplification-impact-decision-success; The Interline — “How Much Will Fashion Let AI Decide?” (Ben Hanson, October 1, 2026), https://theinterline.com/2026/10/01/how-much-will-fashion-let-ai-decide

By Michelle Jones, Fashion News GF

One of the most striking uses of AI at Paris Fashion Week so far didn’t generate a model, a campaign, or a single image. It routed wires. For his Spring/Summer 2027 collection, Anrealage designer Kunihiko Morinaga sent out six sculpted evening dresses covered in up to 2,000 individually controlled E Ink scales that shift color across the body, and AI was used to map the circuitry that makes it work. The show, Anrealage’s 25th in Paris, was held at the Maison de la Radio et de la Musique. It’s a useful reset on what “AI fashion” means in 2026, because the most impressive use of the technology here is one you can’t actually see.

What Anrealage Put on the Runway

The collection is called SKIN, and Morinaga built it around the idea of an outer layer that can be shed and renewed while the person underneath stays the same. “Living things shed their skin, again and again,” he said. “They slip out of the old skin and live on in the new.” His references ranged from chromatophores (the pigment cells that let some animals change color) to video-game avatar “skins” and Osamu Tezuka’s manga Phoenix.

The show started simply, with second-skin bodysuits printed with psychedelic scales and trompe l’oeil patchwork styled to look like denim and varsity jackets. It built up to the six armor-like cocktail and evening dresses, each carrying between 1,000 and 2,000 electronic-paper scales. Unlike LEDs, the scales give off no light of their own. The material itself changes color, so patterns ripple across the dress, fade to ash gray, and come back in new arrangements. Other looks used 3D-printed bustiers that change color depending on the viewing angle, Japanese washi paper with sequin details, and Toray Ultrasuede scales. The finale was a glowing look based on the Phoenix rising.

Why the AI Part Matters More Than It Looks

Reports describe AI’s role narrowly: it mapped the circuitry needed to control the system, a network of wires connecting each scale. That sounds like a footnote. It isn’t. Getting a thousand or more separately addressable elements to work on a curved garment that has to move on a walking body is a hard engineering problem. It’s the kind of layout-and-routing work that would take a human team a long time to solve by hand.

That’s the pattern we keep seeing this year, and it’s one of the clearest AI fashion trends of 2026. Last week we covered Google’s custom AI tools for two New York Fashion Week designers, which handled fittings and set planning rather than making anything the audience saw. Anrealage takes that a step further: AI as engineering infrastructure inside the garment itself. Meanwhile, AI-generated campaign imagery keeps getting called “AI slop.” The uses holding up best are the ones where the creative vision stays clearly human and the AI solves a problem people couldn’t practically solve alone.

E Ink is also a smart pairing. Electronic paper generally uses power when it switches state rather than to keep an image on screen, and it doesn’t glow. That’s why Anrealage’s dresses look like a living surface rather than a wearable screen, and it fits Morinaga’s chromatophore idea.

What We Still Don’t Know

The coverage so far leaves real gaps. None of it names the AI system used, says how the dresses are powered, how heavy they are, or how long the effect lasts. None of it says whether any version will ever be sold. These are runway showpieces, and Anrealage has a long record of tech-driven collections that stay closer to art than to retail. Morinaga’s own framing points to a longer horizon: “Perhaps those clothes will be called a third skin.”

The practical takeaway: when you see an “AI fashion” headline, ask what the AI actually did. Generating a finished image is the easy, contested part. Solving the engineering or planning problem behind a human designer’s idea is quieter, harder to fake, and increasingly where the real work is happening.

FAQ

What did AI actually do in Anrealage’s Spring 2027 collection?

According to reports from the show, AI was used to map the circuitry that controls the E Ink scales, meaning the wiring network that lets up to 2,000 scales per dress change color individually. The collection itself was designed by Kunihiko Morinaga, not generated by AI.

How is E Ink different from LED clothing?

LEDs emit light. E Ink electronic paper doesn’t. The material itself changes color, the same way an e-reader screen does. That gives Anrealage’s dresses a matte, skin-like color shift rather than a glowing screen effect.

Can you buy the color-changing dresses?

No retail plans have been reported. The six E Ink dresses were presented as runway pieces, and details like power source, weight, and durability haven’t been disclosed.

Sources & Further Reading: WWD (via Yahoo) — “Anrealage Spring 2027: Under the Skin” (Alex Wynne, September 30, 2026), https://www.yahoo.com/entertainment/articles/anrealage-spring-2027-under-skin-094924414.html; Designscene — “ANREALAGE Spring Summer 2027 Rethinks Clothing as Skin” (Jana Kostic, September 30, 2026), https://www.designscene.net/2026/09/anrealage-spring-summer-2027.html; Design and Art Magazine — “Anrealage SS27 Moves Beyond Skin Deep,” https://www.designartmagazine.com/2026/09/anrealage-ss27-moves-beyond-skin-deep.html

By Michelle Jones, Fashion News GF

At New York Fashion Week this month, the most interesting AI in the room wasn’t generating models or campaign images — it was working backstage, before a single extra garment got cut. Google says it built custom tools in Google Flow, its AI creative studio, alongside designers Jane Wade and Sergio Hudson to help them plan their shows. Google published details on September 18, and the project is worth a look for one reason above all: it points AI at the expensive, slow decisions that happen before a runway show, not at the creative work audiences actually came to see.

What the Two Designers Actually Got

The tools were built by Google’s Envisioning Studio with support from Google Labs, and each one was built around a specific problem in that designer’s studio rather than handed over as a general-purpose app.

For Jane Wade, the target was casting and fittings. Google says in-person casting and fittings can take a design team as long as three full days. Her styling suite let the team swap hair, makeup, accessories, shoes, and garments on models digitally. That meant they could judge whether a look felt balanced, or whether something was missing from the collection, before anyone made a physical piece. According to Google, that cut out cutting and sewing extra garments that would never have made it to the runway.

Sergio Hudson’s tool dealt with money. He was staging a show on a tight budget. In past seasons, asking his production crew to change lighting or props added cost, because every design revision needed a fresh 3D rendering. His runway visualization tool simulated the show environment instead. He could adjust the venue setup, lighting, and props within budget limits he set himself, and refine the paths models would walk, without paying for another round of renders each time.

Why “Backstage AI” Is the Smarter Bet

This year has been rough for AI’s image in fashion. When Gucci promoted its February Milan show with images labeled as AI-created, critics online dismissed the campaign as “AI slop.” Tools that promise to replace the photoshoot outright, which we’ve covered here before, run into the same question every time: why pay luxury prices for something no human made?

The Google project avoids that argument. Nothing the tools produced was meant for the audience. The collections were still designed, sewn, and walked by people. The AI helped with the decisions that come before all that: which styling works, which piece isn’t needed, what the lighting should look like. When those decisions go wrong, they cost fabric, labor, and production budget. That’s why pre-production is a better use of AI than making the finished product. It also answers the complaint about creative homogenization: if the output is a planning decision rather than a published image, there’s no generic “AI look” to criticize.

The waste angle matters too. Every sample that doesn’t get cut is fabric and labor saved. For small labels, where one wasted sample can take a real bite out of a season’s budget, that could matter as much as the time saved.

What to Keep in Perspective

This is a case study published by Google about its own product, so read it that way. Google hasn’t released hard numbers such as hours saved, samples avoided, or dollars cut from production. It also hasn’t said which underlying AI models powered the tools. Two designers working closely with a Google team is also very different from an independent label building the same thing alone.

The practical takeaway is Google’s pitch that anyone can describe a tool or workflow in plain language and build it in Flow without writing code. If that holds up for small studios, the useful question for a designer isn’t “should AI design my collection?” It’s “which of my slow, costly pre-show decisions could I test digitally first?” Fittings, colorway comparisons, and set planning are the obvious places to start.

FAQ

What is Google Flow?

Google Flow is Google’s AI creative studio. According to Google, users can build their own custom tools in it by describing the tool or workflow they want in natural language, with no coding experience required.

Did AI design the collections shown at New York Fashion Week?

No. According to Google’s account, the tools helped Jane Wade and Sergio Hudson with show preparation: styling and fitting decisions for Wade, and runway staging within budget for Hudson. The clothes themselves were still designed and made by the designers’ teams.

Can smaller fashion brands use the same approach?

Google says Flow’s tool-building doesn’t require coding, which in principle opens it up to small studios. But the NYFW tools were built with direct help from Google’s own team, and Google hasn’t published results data. Brands should start with a single, well-defined pre-production task and measure whether it actually saves time or samples.

Source: Google, The Keyword blog — “Reimagining New York Fashion Week prep with Google Flow” (Yeawon Choi, UX Designer, Envisioning Studio, September 18, 2026), https://blog.google/innovation-and-ai/technology/ai/google-flow-fashion-week/ — with additional context from Outlook Luxe’s February 2026 report on reaction to Gucci’s AI-generated Milan Fashion Week imagery.

By Michelle Jones, Fashion News GF

WGSN has stopped keeping its runway trend data behind analyst reports and handed the query box straight to clients: the trend forecasting giant’s new Catwalks AI Dashboard lets designers, buyers, and merchandisers filter live fashion week data themselves — by market, season, brand, and city — instead of waiting for a human-written wrap-up. The rollout landed this September, timed to New York, London, Milan, and Paris fashion month, and it says more about where AI trend tools are heading than the dashboard’s feature list alone lets on.

What the Catwalks AI Dashboard Actually Does

The dashboard runs on Fashion Vision, WGSN’s AI image recognition model, which tags runway looks by category and color as shows happen and refreshes the data daily through fashion month. Users can pull up a city, a season, or an individual brand and compare product, color, and style shifts side by side — the kind of cross-referencing that used to mean digging through separate PDF reports or waiting for WGSN’s analysts to publish a seasonal recap.

That’s a real shift in who does the interpreting. WGSN’s existing Catwalks service still pairs live show coverage with commentary from its analyst teams, and that layer isn’t going away. The dashboard sits alongside it as a self-serve option for teams that want to build their own read on a season against their own design criteria, rather than starting from someone else’s conclusions.

Why WGSN Built a Self-Serve Tool

WGSN Group CEO Carla Buzasi has said the dashboard’s build came from specific requests by some of the company’s largest clients — brands that wanted direct access to the underlying data rather than only the finished analysis. That’s a notable admission from a company whose business has historically been selling interpretation, not raw data.

It also reads as a defensive move. Runway-data competitors like Heuritech, Tagwalk, and Stylumia have built businesses around letting fashion teams query catwalk and social data directly, and WGSN’s dashboard closes a gap those companies had been filling. For a fashion brand deciding where to put its trend-intelligence budget, the practical effect is more choice: the same underlying runway season is now available as either a polished analyst narrative or a raw, filterable dataset, often from the same vendor.

The Accuracy Question Nobody’s Answered Yet

What WGSN hasn’t published is how accurate Fashion Vision’s tagging actually is once you drill down to the brand level. Aggregate trend calls — “cargo pants are up,” “brown is replacing black” — can tolerate some noise in the underlying image tagging and still land on a directionally correct answer. A dashboard built for teams to run their own comparisons is a different product: it invites exactly the kind of granular, brand-by-brand query where a model’s blind spots show up fastest, and there’s no published error rate to tell a user when to trust a filtered result versus double-check it.

It’s a version of a problem this site has covered before in AI fashion tools generally: a model’s real-world reliability hinges on what its training data actually represented, and vendors rarely publish that breakdown up front. WGSN framing this as a tool built at clients’ request doesn’t resolve that question — it just means the industry is about to find out the answer through use rather than through a disclosed benchmark.

FAQ

Is the WGSN Catwalks AI Dashboard replacing WGSN’s analyst-written trend reports?

No. WGSN’s existing Catwalks service, which combines live show coverage with analyst commentary and seasonal wrap-ups, continues alongside the dashboard. The dashboard is a separate, self-serve layer for teams that want to query the raw runway data directly.

What technology powers the dashboard’s filtering?

Fashion Vision, WGSN’s AI image recognition model, tags runway looks by category and color to enable the filtering and city/season/brand comparisons, with data updating daily during fashion month.

How does this compare to other runway-data tools on the market?

It puts WGSN more directly in competition with companies like Heuritech, Tagwalk, and Stylumia, which already sell queryable runway and social-media trend data rather than only narrative trend reports.

Source: WWD — “WGSN Launches Platform to Analyze Fashion on the Runway,” and FTW — “Primark Pays £90m to Deliver, BoF Questions Fashion’s AI Image Habit, WGSN Sells Runway Data Direct,” https://ftw.pi.tv/primark-pays-90m-to-deliver-bof-questions-fashions-ai-image-habit-wgsn-sells-runway-data-direct/

By Michelle Jones, Fashion News GF

A new industry analysis is complicating the usual “AI is coming for fashion jobs” narrative: by 2030, up to 40% of workers in developed economies may need to reskill or shift into new roles as fashion companies adopt AI — but the argument from inside the industry is that the technology is meant to amplify scarce expertise, not simply eliminate headcount. The figure comes from the McKinsey & Company and Business of Fashion State of Fashion 2026 report, and it’s the anchor for a September 3, 2026 industry analysis from Lectra — a major fashion design and production software company — that’s worth unpacking for anyone trying to separate real workforce change from AI hype.

What the Reskilling Number Actually Means

The 40%-by-2030 figure isn’t a layoff forecast. It describes the share of the fashion workforce in developed economies whose day-to-day role is expected to look meaningfully different because of AI and automation — new tools to run, new decisions to make with AI-generated input, new skills to learn on top of what the job already required. That’s a distinction that gets lost in a lot of AI-and-jobs coverage, which tends to collapse “role changes” and “role disappears” into the same story.

Writing in Textile World, John Brearley, President of Americas at Lectra, frames the shift this way: “Technology alone will not solve the talent gap, but when applied intentionally, it can significantly amplify scarce expertise.” His argument is that fashion has a structural talent problem already — specialized pattern-making, fit, and production-planning skills that take years to build and aren’t being replaced fast enough as experienced staff retire — and that AI’s more realistic job in the near term is closing that gap rather than shrinking the workforce that’s left.

Where AI Is Actually Slotting In

The specific mechanism Brearley points to is knowledge capture: “AI-driven solutions can capture institutional knowledge, standardize best practices and provide real-time insights that support better decisions.” In practice, that’s less about a chatbot replacing a designer and more about software that watches how a company’s most experienced pattern-makers or planners actually solve problems, then makes that judgment available to someone five years into their career instead of twenty.

That framing matters for how brands should be evaluating the current wave of AI fashion tools. A lot of what gets marketed as “AI for fashion” is aimed at shoppers — virtual try-on, style recommendations, AI-generated campaign imagery. The reskilling argument points at a quieter, less visible category: tools built for the people inside a fashion company, doing production planning, quality control, and design iteration, where the payoff isn’t a flashier storefront but fewer bottlenecks caused by knowledge that used to live in one person’s head.

Why This Is Worth Watching, Not Just Reading

Reports like State of Fashion get produced every year, and it’s easy to treat the reskilling stat as one more abstract projection. What makes this one worth tracking is that it’s already shaping how a company that actually sells software into fashion production — not a consultancy with no skin in the outcome — is pitching its own tools to the industry. When the vendor’s public argument shifts from “automate this task” to “capture this person’s expertise before it walks out the door,” that’s a signal about where fashion companies are actually spending budget, not just where the marketing copy points.

The open question is whether “amplify scarce expertise” holds up as adoption scales, or whether it turns out to be the more palatable way of describing the same headcount pressure other industries have seen from AI. Worth watching over the next year: whether fashion companies talk publicly about reskilling programs tied to specific AI tools, or whether the tools simply get adopted and the roles they were meant to “amplify” quietly shrink anyway.

FAQ

Does the 40% reskilling figure mean 40% of fashion jobs will be cut by 2030?

No. The McKinsey/Business of Fashion State of Fashion 2026 report projects that up to 40% of workers in developed economies may need to reskill or transition to new roles as AI and automation change how fashion companies operate — that’s a description of role change, not a job-loss forecast.

What kind of AI tools is this actually about?

The analysis focuses on AI used inside fashion companies for things like production planning, pattern-making support, and capturing institutional knowledge from experienced staff — a different category from the shopper-facing AI tools (virtual try-on, style recommendations) that get more headlines.

Who is making this argument?

John Brearley, President of Americas at Lectra, a fashion design and production software company, writing in Textile World and citing the McKinsey & Company / Business of Fashion State of Fashion 2026 report.

Source: Textile World — John Brearley (Lectra), “AI Can Strengthen Fashion’s Skilled Workforce,” published September 3, 2026, https://www.textileworld.com/textile-world/knitting-apparel/2026/09/ai-can-strengthen-fashions-skilled-workforce/

By Michelle Jones, Fashion News GF

Upload a selfie, paste in a prompt, and let a chatbot tell you your “color season,” your undertone, and the exact palette that supposedly flatters you. That workflow — call it AI style analysis — went mainstream in 2026, powering everything from viral ChatGPT threads to paid styling apps and color-matching widgets built into retailer websites. A recent benchmark study is a useful reality check on how well the technology underneath actually reads a human face.

What the Research Found

The study, released as a preprint by Haoming Lu of Topaz Labs, introduces a dataset called TrueSkin: 7,299 images — 1,790 photographed and 5,509 synthetically generated — sorted into six skin-tone categories the paper labels Dark, Brown, Tan, Medium, Light, and Pale. The structure borrows the six-point layout of the Fitzpatrick scale but grades on plain visual perception rather than dermatological criteria, which is closer to how a styling tool would use it.

When the researcher tested a range of general-purpose multimodal models on that dataset, none classified skin tone correctly more than roughly half the time. Reported accuracy came in at 44.31% for LLaMA 3.2, 40.45% for LLaVA-NeXT, 48.83% for Janus-Pro-7B, 43.12% for Qwen 2.5, and 41.40% for Phi-3.5. A model trained specifically on TrueSkin reached 74.18%, an improvement of more than 20 percentage points over the off-the-shelf systems.

The mistakes were not random. The paper describes a consistent bias toward lighter skin tones, alongside a countervailing pattern in which many mid-range brown samples were pushed the other way and misread as dark. Image-generation models showed a related weakness: when the researcher asked systems such as SDXL, SD3, and FLUX.1-dev to render a specified skin tone, unrelated details in the prompt shifted the result — adding “braided hair,” for instance, tended to make the generated skin deeper than requested.

Why This Matters for AI Style Analysis

Seasonal color analysis is, at its core, a classification task performed on a person’s coloring — skin depth and undertone first, then hair and eyes. If a model’s read of skin tone is skewed before it ever gets to picking colors, every recommendation downstream inherits that skew. And because the study found the error is not evenly distributed, the people most likely to be handed an inaccurate palette are those with medium-to-deep skin — the same groups that face-analysis systems have historically served worst.

This is not limited to novelty chatbot prompts. Fashion and beauty retailers are actively wiring color-matching and AI stylist features into their apps and product pages, and many of those features call the same class of general-purpose vision models the study evaluated. A styling widget that quietly nudges warm-toned or deeper-skinned shoppers toward the wrong palette is a fairness problem and a conversion problem at the same time.

The Practical Takeaway

The research also points to a fix that already exists: a purpose-built model beat the strongest general chatbot in the same test by a wide margin. That is a strong argument that “ask a big AI model” is the wrong tool for this specific job, and that anyone shipping — or paying for — AI style analysis should know whether it runs on a specialized system or a generic one.

If you are using one of these tools yourself, treat the output as a starting hypothesis rather than a verdict, especially for deeper skin tones. Shoot your reference photo in indirect natural light, skip makeup, and wear something neutral or bare your shoulders so fabric color does not bounce onto your face. Run it more than once and compare. And if the result genuinely matters to a wardrobe decision, a human color analyst is still the more reliable check — the automated version has caught up on convenience, not yet on accuracy.

FAQ

What counts as “AI style analysis”?

It is an umbrella term for tools that analyze a photo of you — usually a selfie — to estimate your undertone, contrast level, and “color season,” then recommend a color palette or specific outfits. Some run inside general chatbots through a prompt; others are standalone apps or features built into retailer websites.

Does this study mean AI color analysis is useless?

No. It can be a helpful, low-cost first pass. But the findings suggest you should be skeptical of any single result, particularly if you have medium-to-deep skin, and aware that a general-purpose chatbot performed worse than a model built specifically for skin-tone classification.

How do I get the most accurate read from one of these tools?

Use indirect natural light — not a ring light, overhead bulbs, or direct sun — remove makeup, and keep strong colors away from your face by wearing white, a skin-tone top, or baring your shoulders. Take several photos, compare the results, and cross-check against a human analyst if the stakes are high.

By Michelle Jones, Fashion News GF

Source: Haoming Lu, Topaz Labs — “TrueSkin: Towards Fair and Accurate Skin Tone Recognition and Generation,” arXiv preprint arXiv:2509.10980 (v2, revised February 2026), https://arxiv.org/abs/2509.10980

PointAI, a Simulation AI company built by a former Microsoft Research scientist, says it can render a photorealistic virtual try-on in under a second — without touching a generative AI model at all. That claim is now being tested inside one of India’s largest fashion retailers, and it points at a fork in the road for how “AI try-on” gets built going forward.

Two Very Different Ways to Fake a Fitting Room

Most of the virtual try-on tools that have made headlines this year — including the DressX-powered luxury pilots we covered a few weeks back — work the same way under the hood: a generative model looks at a photo of a shopper and a photo of a garment, then predicts, pixel by pixel, what the combination should look like. It’s often convincing, but it’s still a guess. The model has never actually draped that fabric on that body; it’s pattern-matching against everything similar it’s seen before.

PointAI is pitching something structurally different. Rather than asking a generative model to imagine how a garment would fall, its Simulation AI pairs proprietary AI models with physics-based simulation — the same category of engine used to animate cloth in film and games — to calculate how a specific fabric actually behaves on a specific body shape. According to the company, that combination renders a result in under a second, trained across more than 200,000 body-type variations, at roughly 1/100th the per-image cost of generative try-on APIs, which the company says typically run around $0.15 per image and take 30 to 60 seconds to generate.

Why a Retailer the Size of ABFRL Is Paying Attention

The technology isn’t just a lab demo. PointAI is already working with Aditya Birla Fashion and Retail Limited (ABFRL) — one of India’s largest fashion retail groups, with a footprint that runs into the thousands of stores — where the two companies showcased an in-store virtual trial setup and an AI styling advisor at ABFRL’s Fashion Excellence Day in Mumbai this week. Nothing is formally signed for a nationwide rollout yet, but the demo reportedly covered a mix-and-match trial room experience alongside a broader evaluation across ABFRL’s roughly 3,000 stores. PointAI has also shown the same underlying tech at Bharat Tex 2026 alongside Gokaldas Exports, in front of global manufacturing clients including Adidas, Aerie, Banana Republic, M&S, and Puma, and separately lists Amazon, Flipkart, Myntra, and Rakuten among its retail relationships.

That’s a meaningfully different sales pitch than most AI fashion tools make. The company isn’t leading with “look what generative AI can do” — it’s leading with independence from generative AI. Its argument is that not relying on a commercial foundation model API from OpenAI or Google gives enterprise retailers something generative try-on can’t: predictable unit economics, no exposure to a third party’s pricing changes, and tighter control over how shopper photos are processed. For a retailer running try-on at the scale of a few thousand stores, a $0.15-per-render generative pipeline and a 30-to-60-second wait aren’t rounding errors — they’re the difference between a feature that scales and one that gets quietly shelved.

What This Signals for the Rest of the Industry

PointAI founder Nitin Vats framed the bet in blunt terms: “The last decade of commerce was built for people who browsed. The next is built for people who delegate to agents.” That’s consistent with what we’ve been tracking across AI fashion coverage all year — brands are racing to make discovery and fit-checking instant enough that a shopper (or eventually a shopping agent acting for them) doesn’t bounce before converting. Where PointAI’s approach stands out is in treating “instant and cheap” as a technical requirement to be engineered around, rather than a trade-off to accept as the price of using generative AI.

None of that guarantees physics-based simulation wins out over generative approaches long-term — plenty of retailers will still choose GenAI try-on for its flexibility in rendering totally novel styling combinations a physics engine hasn’t modeled. But it’s a useful reminder that “AI try-on” isn’t one technology; it’s a category with genuinely different engineering bets underneath it, and the bet that wins at enterprise scale may not be the one that gets the most demo-day attention.

FAQ

Is PointAI’s virtual try-on already live in stores?
Not at full scale yet. PointAI has demoed its technology with Aditya Birla Fashion and Retail (ABFRL) at a Mumbai event this week, with a broader rollout across ABFRL’s roughly 3,000 stores reportedly under evaluation — no formal nationwide agreement has been announced.

How is “Simulation AI” different from the generative AI try-on tools other brands use?
Generative try-on tools predict what a garment-on-body image should look like based on patterns learned from training data. PointAI instead pairs AI models with physics-based simulation that calculates how a specific fabric actually drapes on a specific body shape, which the company says is faster and cheaper to run at scale.

Which retailers is PointAI already connected to?
Beyond the ABFRL pilot, PointAI lists relationships with Amazon, Flipkart, Myntra, and Rakuten, and it showcased the technology alongside manufacturer Gokaldas Exports at Bharat Tex 2026 in front of brands including Adidas, Aerie, Banana Republic, M&S, and Puma.

By Michelle Jones, Fashion News GF

Source: The Print / ANI Press Releases, “PointAI, AI Partner to Aditya Birla Fashion, Eyes Rollout of One-Second Virtual Try-On Technology Across ABFRL Stores” (published August 22, 2026), https://theprint.in/ani-press-releases/pointai-ai-partner-to-aditya-birla-fashion-eyes-rollout-of-one-second-virtual-try-on-technology-across-abfrl-stores/3022239/