The company is positioning the rollout as part of a broader rebuild of its digital foundation. Using unified data on Google Cloud alongside AI-ready architecture and governance controls, Gap said it is embedding AI across customer journeys and internal workflows. The objective, it said, is to make intelligence a core part of its operating model rather than a bolt-on feature.

Gap chief technology officer Sven Gerjets said the focus is practical outcomes rather than novelty. “We are not pursuing AI for novelty. These partnerships are about solving real customer problems – helping shoppers feel confident about fit and making it easier to complete a purchase. They also reflect the holistic AI strategy we’ve built to scale intelligence across the enterprise in a disciplined way that drives measurable value over time.”

Sizing is a persistent friction point in online apparel, and Gap is targeting that directly by integrating Bold Metrics’ predictive fit technology into its AI-driven shopping flows. Instead of directing customers to static size charts, the system is designed to deliver personalised size recommendations within live, conversational interactions—placing fit guidance inside the purchase path rather than outside it. Gap said the intent is to make sizing advice an always-on component of the transaction experience and a key enabler for conversational shopping.

Alongside fit personalisation, Gap is preparing for discovery and checkout to happen increasingly inside AI-powered environments. By supporting Google’s Universal Commerce Protocol, the retailer aims to keep product listings accurate and transaction-ready within services such as AI Mode in Google Search and the Gemini app. In practice, this could allow customers to complete purchases directly from these interfaces, shortening the distance between finding an item and paying for it.

Gap said the pairing of predictive sizing and AI-native commerce should “reduce friction” both in size selection and at checkout, improving confidence and conversion while meeting customers in the digital spaces where they are increasingly browsing.

Gap sells clothing, accessories and lifestyle products across Old Navy, Gap, Banana Republic and Athleta. The company reported total net sales of $15.4 billion for fiscal 2025, reaching the upper end of its outlook

The report suggests retailers expect tangible operational benefits to arrive quickly. Many anticipate “meaningful” impact within the next 12 to 18 months, while clearer and more scalable returns on investment are expected in roughly two years. That timeline, Retail Economics argues, creates a narrowing window: delaying capability building risks leaving retailers exposed as AI shifts from optional pilots to a baseline requirement for competing.

Today, the report estimates around 32% of marketing and ecommerce tasks are already exposed to being supported, enhanced or partly automated by AI. The impact spans core activities including personalisation, analytics, campaign execution and product discovery—areas where AI can compress time-to-decision and increase the precision of targeting. Data and analytics functions are among the most immediately exposed, the report notes, because AI systems excel at pattern recognition, forecasting and optimisation. Customer personalisation and experience execution are also highly exposed, reflecting retailers’ push for more relevant, real-time engagement across channels.

Brand and creative functions appear less automatable in direct terms, but still represent a meaningful share of exposed budgets because creative and brand-building remain large line items in marketing spend. In other words, even modest automation or augmentation in these areas could move substantial amounts of AI retail marketing spend.

The research covers four European regions—Benelux, DACH, the Nordics and the UK—and paints a market in transition. Almost all retailers (95%) have tested AI tools in marketing, with many beginning in 2023 through early generative AI and large language model experimentation. But the report finds a split between those still exploring what AI can do and those embedding it into daily operations.

Around one in four retailers remain in exploration or pilot-scaling phases, often constrained by data quality, limited in-house skills or uncertainty over governance. By contrast, 45.3% are described as “operational,” meaning AI is integrated across multiple workflows and is influencing day-to-day execution. Another quarter say AI is embedded at a strategic level, shaping planning, prioritisation and decision-making across functions rather than being limited to isolated use cases.

Despite growing adoption, commercial proof remains uneven. Only 5% of retailers say AI is currently delivering clear, scalable ROI—highlighting a gap between activity and outcomes. Still, most expect that to change quickly as tooling improves and organisations mature their data and operating models.

Skills and governance are cited as major barriers in the UK. The report notes that cultural hesitation and governance concerns constrain three quarters of UK retailers, while two thirds point to a lack of internal expertise as a core obstacle to deploying AI effectively. Data compliance concerns also slow progress, particularly as retailers balance personalisation ambitions with privacy and regulatory expectations.

Voyado’s chief product officer Felix Kruth argued that the next phase of value will come from autonomous systems operating behind the scenes. “What’s most exciting is that we’re still very early in the journey, and the AI we’re using today is likely the least impressive version we’ll ever see. This is still young technology. Generative AI has already delivered significant efficiency gains, but it’s agentic AI, built on the right data foundations, that will prove real commercial value.

“In the race to demonstrate AI progress, it is easy to focus on what is visible: new interfaces, chatbots, and features. But without the right data foundation and context, the impact simply does not materialise. The real value in retail is created by AI working in the background – prioritising the right customers, optimising decisions and ensuring the right thing happens at the right moment.”

Retail Economics CEO Richard Lim described the next two years as decisive. “The next two years represent an inflexion point as AI shifts from experimentation to competitive necessity. Retailers are on a journey, and while most have begun testing and deploying AI, few have reached a stage where it is delivering consistent commercial returns.

“As AI transforms retail tasks, it is reshaping how marketing and ecommerce spend is executed. The retailers that succeed will be those building the right data foundations, skills and operating models now, as AI becomes a core requirement for competing effectively in retail.”

As AI adoption accelerates, the report’s central argument is that AI retail marketing spend will increasingly be determined by data readiness and organisational capability, not just tool availability—separating retailers that industrialise AI from those that remain stuck in pilots.

The retailer is positioning the move as more than a set of disconnected experiments. While much of the industry is still piloting narrow AI features, Gap Inc. says it has reworked its digital backbone so AI can run through the entire customer journey. The foundation is built on unified Google Cloud data, an AI-ready architecture and governance standards designed to scale machine-driven decisioning across the business, rather than treating AI as a standalone initiative.

“We are not pursuing AI for novelty,” said Sven Gerjets, Chief Technology Officer, Gap Inc. “These partnerships are about solving real customer problems; helping shoppers feel confident about fit and making it easier to complete a purchase. They also reflect the holistic AI strategy we have built to scale intelligence across the enterprise in a disciplined way that drives measurable value over time.”

Sizing uncertainty remains one of e-commerce apparel’s most persistent conversion blockers, and Gap Inc. is targeting that pain point directly. Through its partnership with Bold Metrics, the company is embedding predictive fit guidance into AI-driven shopping interactions. Instead of sending customers to static size charts, the system is intended to deliver personalised size suggestions within conversational flows—right when shoppers are close to purchasing. By making fit intelligence part of the purchase path, Gap Inc. is effectively treating sizing as a core capability for agentic commerce, where the “shopping assistant” experience guides decision-making in real time.

At the same time, the company is preparing for a shift in how consumers discover and buy products as search behaviour migrates toward AI answer engines. Gap Inc. said its support for Google’s Universal Commerce Protocol is designed to ensure its product data appears correctly and remains ready for transaction in conversational AI contexts, allowing shoppers to move from product discovery to checkout with fewer handoffs.

Under UCP, Gap Inc. expects to make its catalogue purchasable across AI-native environments, including Google Search’s AI Mode and the Gemini app. The aim is to let customers complete purchases while interacting with AI tools, rather than forcing them to revert to traditional browsing and checkout patterns. In practical terms, it is an attempt to keep Gap’s brands present and “buyable” wherever the next generation of shopping journeys occurs—an approach that aligns with the company’s broader push to operationalise agentic commerce rather than treat it as a future concept.

Warehouses are becoming machines, not just buildings

The modern fulfilment centre looks less like a sea of shelves and more like a choreographed production line. Where traditional warehouses depended on staff walking long aisles to find items, today’s facilities increasingly deploy robotics, automated storage and retrieval systems and AI-based orchestration to compress time and distance. Robots can bring inventory to the picker rather than the other way around, raising throughput and improving accuracy. That reduction in mis-picks matters when operations are processing tens of thousands of parcels a day and a small error rate becomes a costly flood of customer service cases.

Automation also stabilises performance during demand spikes. When a promotion hits or seasonal volumes surge, automated workflows can scale more predictably than purely manual operations, while also lowering the risk of mistakes that erode margin through reships and refunds.

From one mega hub to many smaller nodes

The second structural shift is geography. To meet the promise of next-day and same-day shipping, retailers are moving away from reliance on a single central warehouse and toward distributed networks of fulfilment sites placed closer to where customers live. With inventory positioned in the right places, shipping distances shrink, delivery times tighten and carrier costs can fall.

For many retailers, this distributed model is too complex to run without specialist support. Inventory has to be balanced across locations, demand has to be forecast at a regional level and replenishment decisions have to happen continuously. As a result, many businesses lean on fulfilment partners to manage the complexity and keep the network responsive.

Micro fulfilment rises as speed expectations harden

The push for faster delivery has also created space for micro fulfilment small, highly automated facilities located in dense urban areas or even integrated into existing stores. These sites are built for rapid processing of local orders, reducing the last-mile distance and helping retailers meet tight delivery windows.

Grocery has been a major catalyst here, where speed and freshness are non-negotiable. But micro fulfilment is increasingly relevant across categories where consumers expect near-immediate availability, particularly in major cities.

Software now dictates flow, not just labour

Hardware gets attention, but the real nervous system of fulfilment infrastructure is software. Advanced warehouse management systems provide real-time visibility into inventory, optimise where products are stored and coordinate the choreography of picking, packing and dispatch. With machine learning layered on top, these tools can anticipate demand shifts, flag stock risks and suggest rebalancing moves before service levels drop.

That intelligence reduces two expensive failures at once: stockouts, which cost sales and damage trust, and excess inventory, which ties up capital and increases markdown pressure. When demand changes quickly, visibility and responsiveness are the difference between a smooth operation and a costly scramble.

AI and predictive analytics reshape planning

Predictive models are increasingly being used to decide what stock should sit where, and when. By analysing historic sales, seasonality and behavioural patterns, retailers can pre-position inventory across regions so orders travel fewer miles and process faster. The same forecasting logic can be applied to workforce planning and transport routing, cutting overtime costs and improving utilisation.

This is also where many retailers gain the confidence to promise faster delivery: not by working harder, but by planning earlier and moving stock before customers click “buy”.

Last mile delivery is still the hardest mile

Even the best warehouse cannot fix a broken last-mile network. The final journey from a local depot to the customer’s door remains the most expensive and operationally messy step in e-commerce logistics. To reduce friction, retailers are combining fulfilment operations with alternative delivery models, including local courier partnerships, parcel lockers and crowdsourced networks. Longer term, autonomous vehicles and drones remain on the horizon, but even without them, the goal is the same: increase density, cut failed deliveries and reduce per-drop cost.

Sustainability becomes operational, not optional

As e-commerce grows, so does the environmental burden of packaging, transport and energy-intensive facilities. Many retailers are now investing in lower-impact operations: energy-efficient warehouses, route optimisation, recyclable packaging and renewable power sources such as solar. Some are electrifying delivery fleets where feasible, reducing emissions in urban zones.

The shift is partly reputational, partly regulatory and partly economic. More efficient routes and better packaging choices often reduce cost as well as carbon.

Returns management is a profit lever

Returns are a defining reality of online apparel and footwear retail, and they can quickly erode profitability. Modern systems are therefore being designed to process returns faster, sort items accurately and route them to the most value-preserving outcome restock, refurbish, secondary channels or liquidation. Automated sorting and tracking tools help compress cycle time, which is critical because resale value often drops as a season moves on.

Data is also becoming a prevention tool. By analysing return reasons and patterns, retailers can improve sizing guidance, imagery and product descriptions, reducing avoidable returns before they happen.

Stores rejoin the network through ship from store and BOPIS

Physical stores are being repurposed as fulfilment assets. Ship-from-store turns local inventory into a delivery advantage, enabling faster fulfilment and better stock utilisation. BOPIS buy online, pick up in store offers convenience for customers and lowers last-mile delivery cost for retailers, while also creating opportunities for incremental in-store purchases.

The pandemic locked in the new playbook

COVID-19 accelerated many of these moves. As online demand surged, retailers invested rapidly in automation, logistics software and new fulfilment locations. Those capabilities did not disappear when stores reopened; they became the baseline. Consumers learned to expect speed and reliability, and retailers learned that logistics excellence is a core competitive advantage.

Where fulfilment goes next

Fulfilment will keep evolving as robotics improves, AI becomes more predictive and data becomes more real time. The retailers that win will be those that treat fulfilment not as a cost centre, but as a strategic platform that shapes customer experience, margins and brand trust.

In the end, the front-end of online retail may be what shoppers see, but fulfilment infrastructure is what they feel every time an order arrives on time, intact and exactly as expected.

Building on the first OdorCrunch system, Polygiene says the updated version delivers stronger absorption performance and works across a broader set of materials. That includes lightweight 100% polyester—often a challenge for odor management—alongside blended constructions and natural fibres. The mechanism is intended to capture environmental odor molecules and prevent them from being released during wear, extending the “fresh feel” window and reducing the need for immediate laundering.

A key differentiator, according to the company, is the chemistry profile. Polygiene says Polygiene OdorCrunch2.0 is heavy-metal free and uses a capture approach that retains odor molecules while maintaining fabric hand feel and technical properties. The finish is designed to avoid trade-offs commonly associated with odor solutions, preserving comfort, breathability, appearance and performance.

From a manufacturing perspective, the technology is built to slot into standard textile finishing lines. Polygiene says it can be applied through widely used methods such as padding and exhaust processes, making integration straightforward for mills without requiring major equipment changes. The company adds that performance is durable and has demonstrated a high level of odor reduction under ISO 17299 testing, including results on lightweight 100% polyester.

While activewear and performance categories are obvious targets, Polygiene is also pitching the technology for products that are difficult to wash regularly—such as garments requiring dry cleaning or specialised care. By improving freshness between cleaning cycles, the company argues the finish can support longer garment life by reducing washing frequency, while also improving wearer confidence across additional apparel segments.

Polygiene also frames the launch in the context of sustainability expectations. The company says the technology contains no heavy metals and no PFAS, and notes that biodegradability characteristics are currently being evaluated. By enabling more wears between washes, Polygiene suggests the finish has the potential to reduce water and energy use over a product’s lifetime.

With broader fabric compatibility, mill-friendly application methods and a sustainability profile designed to meet tighter chemical expectations, Polygiene positions OdorCrunch2.0 as the newest step in expanding its odor control portfolio.

That’s the central message of Linnworks’ 2026 State of Commerce Ops Report, based on input from 500 companies across the two markets. All respondents sit in the mid-market band, with annual revenues between $7.5m and $100m, and are navigating the complexity of running across multiple digital channels and fulfilment networks.

The study indicates that growth is broad, with 89.6% of UK respondents and 88.8% of US respondents reporting moderate to significant year-over-year improvement. Yet Linnworks says scaling remains uneven: “scalability is selective,” and tends to correlate with automation levels and channel maturity.

Linnworks CEO Jon Bahl said: “As retailers navigate the ever-evolving ecommerce industry, one thing remains clear: growth and operational maturity are inseparable. Our new survey data sheds light on the ongoing growth of the market while providing clients with strategies for scalability and progress.”

Inventory visibility emerged as a persistent weakness for mid-sized online sellers. Only around one-third of respondents said they have excellent visibility across all channels and warehouse locations, a shortfall that can magnify errors as mid-market ecommerce retailers add marketplaces, geographies and fulfilment nodes. Linnworks noted that inaccurate stock tracking, inconsistent system reliability and fulfilment execution issues remain common as companies expand.

The report also suggests AI is moving rapidly from optional to foundational. More than half of those surveyed said they already use AI chatbots for customer support, including 60.8% in the US and 54.8% in the UK. Overall, more than 95% said they use AI somewhere in their operations, and most expect to increase adoption in 2026 once they identify “effective” use cases. Linnworks said AI is also being deployed in demand forecasting, inventory management, marketing optimisation and fraud prevention.

Automation is increasingly viewed as standard operating practice rather than a differentiator. In the UK, 64.8% of surveyed businesses reported that their operations are majority or highly automated; in the US, the figure was 60%. Linnworks says the depth of automation still varies widely, and those differences show up in resilience, growth outcomes and leadership confidence—separating companies that can scale smoothly from mid-market ecommerce retailers that grow into operational friction.

On fulfilment, the report argues that logistics choices remain central to customer experience and operational stability. Retailers working with multiple carriers were described as more resilient during peak periods, encountering fewer disruptions and having greater flexibility to manage delivery speed and cost.

Most wearable robotics still depend on stiff components that can restrict movement, reduce comfort and limit everyday usability. Soft robotic elements promise more discreet support, but achieving meaningful force and motion inside a flexible garment has been a persistent engineering challenge.

The breakthrough comes from EPFL’s Soft Transducers Lab (LMTS) in the School of Engineering, where PhD student Huapeng Zhang and LMTS head Herbert Shea developed an actuator textile using shape memory alloy (SMA) fibres arranged in a periodic X-crossing architecture. SMA fibres, made from nickel-titanium, shorten and stiffen when heated by an electrical current. While the material itself is powerful, its output has often been diluted in textiles because conventional knits and knotted structures make fibres pull in different directions, causing forces to counteract one another.

EPFL’s X-crossing design addresses that problem by aligning each fibre intersection with the intended direction of movement, so the forces add up constructively rather than cancelling out. The team also says the geometry improves wearability: the fabric can stretch to 160% of its original length, helping garments remain easy to put on and comfortable in motion.

“We realised that the orientation of fibre crossings plays a critical role in how forces add up inside a textile actuator,” explains PhD student Huapeng Zhang. “By aligning the crossings, we ensure that the forces generated at each intersection contribute constructively, rather than working against each other, resulting in a textile actuator that significantly outperform previous knitted or knotted designs.”

To show application potential, the researchers integrated the textile actuators into two prototypes. In one demonstration, the actuator was mounted on a mannequin’s arm as a sleeve to assist elbow flexion, lifting a 1 kg bag through a smooth 30-degree range of motion. In a second test, the actuators generated on-body compression, a key requirement for medical sleeves and performance athletic gear.

Beyond the textile architecture, the team developed a mechanics model that better predicts performance by capturing how SMA stiffness changes with temperature and stress along each fibre as phase transitions occur. The model is designed to help engineers forecast force and contraction under different loads, thermal conditions and geometric layouts, supporting more reliable design and scaling.

Shea also highlighted efficiency as a practical advantage, noting that the X-crossing configuration can maintain compression pressure with zero ongoing energy input.

“While textiles traditionally serve solely as passive apparel, the transition to fabrics that function as powerful actuators will enable a new class of comfortable, unobtrusive, practical wearable robotics that provide support for daily activities to be designed,” he says.

The project targets a recognised vulnerability: ageing can reduce the body’s ability to regulate temperature, while chronic conditions, medications and frailty can further increase risk—especially for people living alone. The vest is intended to provide sustained monitoring and, if necessary, trigger an alert when temperature readings suggest a potentially dangerous trend.

The research is being led by Dr Theo Hughes-Riley from Nottingham Trent University’s Advanced Textiles Research Group (ATRG). The garment integrates four miniature thermistors directly into the yarn structure to detect longer-lasting shifts in temperature. Each thermistor measures roughly 1mm by 0.5mm and connects to a Bluetooth-enabled microcontroller, enabling real-time transmission to a mobile phone or other device. The system could raise an alarm if unusual readings persist.

To ensure practicality, the thermistors are sealed in resin so they can withstand washing, and are protected within a polyester braid. They are positioned at key sites for temperature monitoring—two on the chest and two over the scapula—and are designed to be imperceptible to the wearer.

The prototype has been tested through everyday movements such as sitting, walking, jumping and reaching to confirm that measurements remain reliable during real-world use. Trial data showed distinct temperature-change patterns that could support earlier identification of abnormal physiological responses before they escalate.

“Hypothermia is a very dangerous condition, particularly for those who are elderly and live alone without anyone to raise the alarm for them should they become ill,” says Dr Hughes-Riley, of the Nottingham School of Art and Design. “By combining electronic textiles with an everyday garment such as a vest, carers and medical professionals would be able to respond immediately to any detected risk and help save the lives of older people who may need urgent support.”

Technically, the sensors are placed on flat contact surfaces to maintain steady skin contact and use a semiconductor property in which resistance drops as temperature rises. The thermistors are calibrated to the yarns used in the garment, and the readings are used to infer an estimate of core body temperature.

ATRG senior research fellow Dr Arash Moghaddassian Shahidi said the aim is consistent, round-the-clock oversight for those most at risk: “By utilising smart textiles technology in this way, we can help ensure that vulnerable older people are monitored around the clock for serious health conditions like hypothermia, where it is paramount that they receive urgent treatment,” he observes. The team believes the smart textiles vest could ultimately support faster intervention and better outcomes for older people who need urgent help.

The programme’s full title “Thermal Insulation and Acoustic Optimisation through the Use of Residual Materials as Fillers in Near-Net-Shape 3D Woven Structures for Textile Panels” signals its technical direction: fully textile, near-net-shape 3D structures that can be filled with residual-material-based fillers. The work began at the start of the year and is supported through Germany’s Central Innovation Programme for SMEs (ZIM).

“With WAVE, we are addressing a highly relevant topic that meaningfully combines ecological requirements, material innovation, and regional value creation,” explained professor Dr.-Ing. Frank Ficker, project lead and Head of the Institute of Materials Science (ifm) at Hof University of Applied Sciences.

The project is responding to growing demand for building products that are environmentally compatible without sacrificing performance. Many conventional insulation and acoustic materials depend on fossil-derived inputs, use multi-material composites that are difficult to separate, and often fall short on recyclability or biodegradability. The WAVE consortium aims to demonstrate an alternative: a textile-based panel system engineered to deliver insulation and sound optimisation while supporting circular end-of-life pathways.

“We see a clear market demand, but also enormous untapped potential in regionally available raw materials such as wool,” emphasised Isabell Korn, research associate at the ifm and project manager. “Our approach is to make this resource functionally usable in a targeted way and to transfer it into a market-ready textile system.”

A key technical premise is that engineered textile structures can integrate multiple requirements function, stability and design flexibility within one material system, rather than relying on glued layers or difficult-to-recycle composites.

“Textile structures enable us to combine functionality, stability, and design within a single material system,” explained Corinna Anzer, deputy project lead and head of weaving, braiding, and yarn development at the Institute of Materials Science (ifm). “This opens up new pathways for sustainable insulation and acoustic solutions in interior construction.”

Development will begin with lab-based design and testing at the ifm, where researchers will evaluate textile structures and filler concepts. From there, the work will move toward industrial-scale implementation at Möbelstoffweberei Reich e.K., with continuous knowledge transfer built into the process. The target at the end of the WAVE research project is a market-ready prototype demonstrating verified thermal and acoustic performance and meeting current expectations for modern interior construction.

“For us, the close integration of research and practice is a key success factor,” said Claudia Schödel-Reich of Möbelstoffweberei Reich e.K. “Early involvement in the development process ensures that a good idea can ultimately become an industrially viable product.”

The partners say the resulting textile insulation panels could offer a lower-impact alternative for offices, educational facilities and cultural buildings, supporting improved acoustics and thermal comfort while aligning with environmentally responsible construction choices.

“Particularly in interior construction, there is growing demand for flexible, sustainable, and aesthetically appealing solutions,” added Korn. “Textile panel systems can offer genuine added value here.”

The initiative also reflects a regional innovation pipeline. The concept emerged from discussions at TextilTreff Oberfranken, a networking event organised by Bayern Innovativ GmbH and the Association of the Bavarian Textile and Apparel Industry (VTB), where representatives from Möbelstoffweberei Reich, the ifm and Felix Meier of neowistra developed the joint proposal that was later approved under ZIM.

London, 17 February 2026 — ASOS has rolled out a new hybrid virtual try-on experience designed to help customers judge how selected items may look before committing to a purchase. The feature gives shoppers two ways to preview products: either by uploading their own image or by selecting an AI-generated virtual model created to resemble them.

ASOS says the dual-option approach reflects how differently people feel about virtual try-on tools. Some prefer the familiarity of using a personal photo, while others are more comfortable exploring outfits through a digital likeness that matches their size, proportions or style—without sharing an image of themselves. By offering both routes, the retailer aims to make online fashion browsing feel more personal and reduce uncertainty at checkout.

The launch is being delivered in partnership with AI fashion platform AIUTA and begins with around 10,000 products available through the ASOS iOS app. The feature will initially be offered to selected customers in the UK and US, with wider rollout planned over time. ASOS also highlighted performance, saying the try-on experience typically loads in four to seven seconds, faster than many industry alternatives, so customers can test looks quickly and stay engaged.

Melissa Lim, Head of Digital Product at ASOS, said: “We know customers want the confidence of seeing how something will really look but don’t want to be pushed into doing it one way. Our hybrid approach meets them where they are, giving everyone a try‑on option that feels right for them.”

ASOS, founded in 2000, serves 17 million active customers across more than 200 markets and sells its own brands—including ASOS DESIGN, ARRANGE, COLLUSION, Topshop and Topman—alongside a large selection of partner labels. The company says this hybrid virtual try-on release fits its broader strategy of removing friction from online fashion shopping while giving customers more control over how they discover and buy clothes.