Fashion buyers have long acted as the industry’s quiet tastemakers, the people who can sense desire before it’s formed. But now, facing tighter margins and the pressure of precision, they’re meeting these demands with the help of AI.

With the ability to process vast amounts of previously siloed data — search behavior, click patterns, regional preferences, and product performance across markets — AI is rapidly moving beyond simple sales forecasting. Buyers and merchandisers say it’s now reshaping how they build, refine, and scale assortments, as decisions become more data-led than ever.

Instead of relying solely on past sell-through or personal intuition, buyers can access real-time signals about what shoppers are searching for, clicking on and saving globally. “AI is more of a tool that extends their reach,” says Rich Shepherd, VP of product at Lyst. “The best buyers still lead with instinct — AI just gives them a clearer view of where that instinct might resonate most strongly.”

From luxury groups to global e-commerce platforms, a new model is emerging: AI-powered recommendation systems and pattern-surfacing tools that analyze data, while human buyers interpret those insights and make strategic decisions. The balance between the two is becoming a competitive advantage.

Real-time demand insights

Tapestry, parent company of Coach, Kate Spade, and Stuart Weitzman, uses AI behind the scenes, helping buyers to make smarter decisions about what to order, how much to stock, and where to allocate inventory.

“We always understood that to digitalize this process and scale fast, we had to build a capability to host and share data easily across the business,” says Fabio Luzzi, chief data and analytics officer at Tapestry. The company invested in building a centralized data repository — what Luzzi calls its “proprietary data fabric” — which makes it easy to model data around customers, locations, and supply chains. “It makes the digitization of processes very easy, as well as the ability to use AI across multiple steps in the value chain.”

Coach’s buying teams are already using shared data sets to compare regional buying patterns in real time, adjusting depth and allocation before products hit stores. These insights reveal demand earlier, with more precision than historical sell-through alone.

In practical terms, a member of the team might open a live, shared dashboard, which will show a particular silhouette over-indexing in the southwest US while underperforming in the northeast — information that previously arrived weeks later via sell-through reports. That signal allows them to adjust the allocation before stock is committed, rather than having it sit in the wrong warehouse. Luzzi positions AI as an embedded decision-support system across design, inventory, and pricing, accelerating analysis and interaction while leaving final product and merchandising judgments with human teams. He says this is freeing up buying and merchandising teams’ time so they can focus on more strategic work.

At Coach, core leather goods and seasonal staples — categories with years of sell-through data behind them — are increasingly handled through automated replenishment models, with the system flagging reorder points, adjusting depth by door, and reallocating stock between regions without manual input. The time it frees up is significant: merchants who previously spent the majority of the buying cycle managing the known quantities of their assortments can redirect that capacity toward the categories where data offers less certainty and human judgment carries more weight.

More time for merchandising and trends

For trend-driven or new pieces, it’s a different story. Their sales performance depends on cultural timing, editorial context, and early signals that historical data alone can’t yet model. What this means in practice is that buyers are spending less time making buying decisions around known quantities and more time on the harder question of what the customer doesn’t know they want yet — the part of the job that requires taste, not just analysis, according to Farfetch chief technology officer Luis Carvalho.

“We believe in empowering our customers’ individual style, not dictating it,” Carvalho says. Farfetch’s personalization engine refines what shoppers see based on style signals rather than pure popularity. “Advances in AI — from data processing to predictive modeling — help us navigate vast amounts of information and connect each customer to the most personalized products across our network.”

These advances include AI’s ability to process billions of signals — search behavior, click patterns, product metadata, and regional buying differences — at a speed beyond the reach of human teams. As AI capabilities have expanded, fashion aggregator marketplace Lyst has moved from broad catalog ranking to style-level recommendations, matching products to individual shoppers based on taste, price sensitivity, and occasion.

“Before, merchandising was just about what the first six products you’d see in a feed were,” Miyon Im, VP of product design and editorial at Lyst, says. “But with AI, we can get more sophisticated — around styling, outfitting, or even event-based suggestions. If we can use AI to create something like an office party recommendation, where every piece feels right because we understand your taste, preferences and price sensitivity, that’s the future.”

In practice, this means that Lyst’s merchandisers receive regular data briefs, including surfaced items gaining unusual traction in searches or saves, which they then interrogate for fashion context before any recommendation is made. When the data flags spikes in colors or textures, it’s not published as a trend immediately. A human has to ask why: was it a runway moment, a celebrity citing, or a cultural reference? Only then is it incorporated into the merchandising.

Balancing data with intuition

Executives say that, as of yet, AI’s promise arrives with structural limits. Machine learning models are only as reliable as the data they are trained on, and fashion’s historical biases — in sizing, representation, geography, and taste hierarchies — can easily be reinforced rather than corrected. If past sales skewed toward narrow size ranges or specific demographics, those exclusions don’t disappear in machine learning models — they scale. Experts say that AI can’t yet replace the cultural intelligence, intuition, and narrative instincts that shape fashion.

“AI is here, and an incredible tool to enhance one’s work,” says Julie Gilhart, a fashion consultant who spent 18 years overseeing buying decisions at Barneys New York. “But the real magic comes from human intuition; the instinctive sense that data alone cannot replicate. The brands that get it right will let creativity lead, with AI enhancing the vision rather than replacing human touch.”

As brands integrate more data-driven tools, Gilhart says a new role is emerging: merchandisers who can translate analytical signals into creative strategies. “You have to be curious,” Shepherd says. “You don’t need a computer science background, but you need to understand how the technology works to solve problems for users and partners.”