Why Stitch Fix Is Doubling Down on Human Stylists as AI Gets Better

The shopping journey has always involved a kind of surrender.

You surrender to a brand’s editorial curation when you browse a site. You surrender to an algorithm when it populates your feed. You surrender to a stylist when you step into a fitting room and admit you need help. Most retail technology has been built to make that surrender faster and more frictionless, and to reduce the human judgment involved.

Stitch Fix has spent the last decade building the opposite model. The premise, from the start, was that human taste and data science were stronger together than either was alone. Now generative AI is sophisticated enough to style, visualize, and recommend outfits at scale, and the obvious question is whether the human stylist still earns their place in the loop. At eTail Palm Springs 2026, Noah Zamansky, VP of Client Experience at Stitch Fix, gave a clear answer: the more Stitch Fix invests in AI, the more it invests in its stylists.

The product that moved the mirror

The conversation opened with Stitch Fix Vision, a new product that represents the company’s most visible AI bet. The mechanics are straightforward: a client uploads a selfie and a full-body image, and the platform generates personalized, shoppable outfit images styled around them specifically. The shift it produces is subtle but commercially significant.

“It moves from ‘hey, that’s a cool top on that person’ to ‘oh my God, that’s a really great outfit on me, and I can visualize it and shop it,'” Zamansky said.

Behind that experience, the engineering complexity is substantial. Zamansky described autonomous quality assurance running before any image surfaces to a client, with dozens of decisions made invisibly in the seconds between upload and result. Facial likeness, body type, outfit layering, personal style preferences, and stylist expertise are all orchestrated before the client sees anything. The outcome, by Stitch Fix’s own measurements, is that Vision clients are coming back more frequently, shopping significantly more, and sharing the outputs with friends and family.

That sharing behavior matters. “We look at styling and shopping as a team sport.” Stitch Fix built a sharing feature directly into Vision, and it is generating a product-led growth loop. Clients share their personalized outfit images with people around them; those people engage with the product; some convert. A flywheel Zamansky described as working exactly as intended.

AI is saving time so stylists can spend it differently

The reflexive concern about a product like Vision is that it makes the human stylist redundant. Zamansky addressed this head-on.

“The more we invest in AI, the bigger the opportunity we find. It’s saving our clients and our stylists time, but the stylists are actually spending more time connecting with clients.”

That reallocation of time is deliberate. Stitch Fix has launched stylist profiles, giving clients visibility into who their stylist is as a person: their interests, their expertise, the sensibility behind the recommendations. It has also launched Connect, a chat-based platform through which clients can message their stylist directly. Questions about specific upcoming events, opinions on a jacket they found elsewhere, full outfit builds for occasions. The connection is real-time and personal.

Zamansky made the commercial case for this investment with a story about a client named Jenna, who had been through a significant body transformation driven by GLP-1 medications. She messaged her stylist after a wardrobe refresh with a note that, in Zamansky’s telling: “You’ve helped give me the confidence to buy nice things again.” The stylist had not simply recommended clothes. She had tracked a client’s evolving relationship with her own body and met her there. “That’s the kind of connection you can’t replace with AI,” Zamansky said.

For senior marketers at brands thinking about the automation of service and support, the lesson here is not about technology restraint. Stitch Fix is disciplined about where human attention delivers something algorithms cannot produce: accumulated relational context applied to a moment of personal significance.

Conviction inside an experimentation culture

On the operational side, Zamansky described a company running thousands of experiments per year, with client satisfaction and behavioral metrics serving as the non-negotiable North Star. The nuance in his framing was worth noting. Experimentation culture is often presented as a mechanism for killing bad ideas quickly. At Stitch Fix, it carries an additional instruction: hold conviction.

“Sometimes it’s not the second or third try that actually works, but the fifth, or the sixth, or the tenth iteration where you actually get the signal you’re looking for,” Zamansky said. The decision to persist or abandon is driven by client metrics, not internal opinion, but the threshold for abandoning is higher than many data-first cultures would set it.

This creates a specific kind of organizational discipline. Teams need to know the difference between a test that is failing and a concept that has not yet been properly expressed. That is a harder call than it sounds, and Zamansky was candid that it requires genuine conviction about the underlying client need. “It always comes back to that,” he said. “What are we trying to solve?”

The subscription is not what it was

On subscription models, Zamansky rejected the fatigue narrative and reframed the question entirely. The model Stitch Fix runs today barely resembles its original architecture. Clients can now add specific items to a fix before it ships. The company is experimenting with larger fix formats, including eight-item boxes, and is building out flexibility across categories, including accessories for women’s clients. The line between curated subscription and inspired shopping is deliberately blurred.

The personalization paradox question produced a similarly layered answer. Some clients want the surprise: the magic moment of opening a box styled entirely by someone who knows them. Others, as they develop confidence in their own style through time with a stylist, shift toward wanting more agency and control. Zamansky’s framing here was grounded in his own origin story with the brand.

“The reason I joined Stitch Fix is because I was a client. I had that magic moment, and I discovered Vuori through my stylist.”

He went on to describe what Stitch Fix is building toward: a style education that moves clients from pure surprise toward informed preference, the way a sommelier might move a wine novice from a single recommendation toward a genuine palate.

The fast fashion question answers itself

When asked whether Stitch Fix and algorithm-driven fast fashion players like Shein are converging, Zamansky was direct. “I think it’s a different game.” His reasoning was not about product or price. It was about client intent. Stitch Fix clients, he argued, are seeking confidence, convenience, and inspiration. The stylist relationship is the mechanism through which those things are delivered. Fast fashion addresses a different need for a different mindset. The products may both involve clothing and algorithms, but the service proposition is not competing for the same customer.

What this means for senior marketers

Zamansky closed with a question about leadership through transformation, and he returned immediately to first principles. He cited Marty Cagan’s books “Inspired” and “Empowered” as guides he returns to regularly, listening in the car on repeat. The reason is specific: Cagan’s framework, he said, is about clarity on the customer problem, empowering product teams to pursue that problem, and resisting the urge to over-specify the solution in an environment where the answer will come through iteration.

 

Stitch Fix is in its third year of a turnaround, by the interviewer’s framing, entering a growth phase while simultaneously running one of the most complex human-AI service operations in retail. The philosophy keeping it coherent is a customer-centric conviction that the right measure of any decision, whether it is a new AI product or a new stylist feature, is whether it produces genuinely happier clients. Measurement and iteration are in service of that signal. The technology follows.