Virtual try-on has evolved from AI demo to business infrastructure. DMEXCO 2026 explores how AI is making the same shift across fintech, healthtech, deeptech and SaaS through scalable, measurable impact. 1 hour ago
For a decade, virtual try-on technology has been the industry’s perennial “almost.” The idea—letting shoppers see how a jacket or a shade of lipstick would actually look on them—was always compelling. The execution was not. Brands had to feed in expensive 3D product data, tools choked on badly lit selfies, and the results looked more uncanny than useful.
That changed fast. Generative AI can now turn a standard product photo into a 3D model that simulates cut, drape and fabric in real time, without the manual 3D pipeline that made earlier versions unworkable. ASOS now lets shoppers upload a photo or build a digital twin from their proportions and preferences. Breuninger became the first German fashion retailer to integrate Google’s virtual try-on technology into its app. Maybelline lets users try shades via upload, digital model, or live camera. None of this is a lab demo. It is live, in production, driving measurable results.
That distinction—between AI as a pilot and AI as infrastructure—is exactly why virtual try-on has become a reference case for an entire industry. Fashion e-commerce has quietly carried two expensive, unsolved problems for years: purchase hesitation from not knowing how a product will look, and return rates driven by size uncertainty. In the US, the National Retail Federation estimated that 19.3% of online fashion purchases were returned in 2025, with Gen Z shoppers returning close to eight garments on average. Arnold Pötsch, lead author of the BVDW working group paper on 3D in e-commerce, put it plainly: advances in computer vision, AI and real-time rendering have turned virtual try-on into “a clear competitive advantage for forward-looking retailers,” blurring the line between physical product and digital twin, and giving retail the key to a personalized shopping experience, fewer returns, and greater sustainability.
The reason this matters beyond fashion is the pattern underneath it: AI that finally scales past the demo stage, tied directly to a P&L. That pattern is showing up across every vertical Tech.eu covers. In fintech, AI-driven underwriting and fraud detection are moving from add-on features to default infrastructure. In healthtech, diagnostic and triage models are clearing the trust threshold that kept them in pilot purgatory for years. Deeptech ventures are proving that foundational research—in robotics, materials, or computer vision—can be productized on realistic timelines. SaaS vendors are being judged less on their AI features list and more on measurable retention and margin. And sustainability has stopped being a compliance checkbox, becoming a genuine efficiency lever as AI helps reduce waste, from fewer fashion returns to leaner data center loads.
This is precisely the shift DMEXCO has chosen as its 2026 theme: “Scaling Intelligence”—the pivot from AI experimentation to real value creation. It’s a fitting frame for where the market actually is. The easy wins from bolting a chatbot onto an existing product are gone. The harder, more interesting work now is building AI that holds up at scale, earns trust from real users, and shows up in the numbers investors and operators actually track.
DMEXCO 2026 takes place September 23–24 in Cologne, and brings together decision-makers across agencies, commerce, tech and media to work through exactly these questions. For founders, operators, and investors tracking where AI is actually converting into revenue and retention rather than headlines, the agenda spans World of Commerce, World of Tech, and dedicated summits addressing the same scaling challenge across sectors—from fintech and healthtech to deeptech and sustainable innovation. It’s a chance to see, case by case, which parts of the AI hype cycle have turned into working infrastructure, and which are still stuck in pilot mode.
That mix of formats is deliberate. The Expo floor puts scaled products from established players next to early-stage tech in the Start-up Area, so visitors can compare a live enterprise deployment against the pitch that might become one in eighteen months. The Conference stages pair that with the strategic view: operators explaining what actually broke when they moved a model from pilot to production, and where the ROI showed up first. For a European tech audience in particular, that combination matters. Much of the AI scaling conversation is still dominated by US platforms and Chinese manufacturing; DMEXCO’s floor and agenda are unusually dense with European builders—in fintech, healthtech, deeptech and SaaS—solving the same problem under different regulatory and capital constraints.
Virtual try-on took ten years to go from novelty to necessity. The vertical-specific AI applications now underway in fintech, healthtech, deeptech, and SaaS are unlikely to take that long—but they will face the same test: does it work at scale, for real users, with numbers that hold up. That is the conversation DMEXCO 2026 is built around.