Adobe’s annual Sneaks session at Adobe Summit in Las Vegas has once again drawn a packed mainstage crowd, and this year it delivered a lot more than just curiosity.
The long-running showcase of experimental ideas has evolved into a proving ground for what’s next in customer experience, with a growing number of concepts making the leap from demo to deployed product.
Designed to spotlight innovation from across the business, Sneaks invites anyone inside Adobe – from engineers to UX designers to pitch their boldest ideas. And in 2026, that pipeline has surged. More than 500 concepts were submitted (up from around 150 last year), fuelled by AI-powered ideation, rapid prototyping and what insiders are calling a new era of “vibe coding.”
The result? A tightly curated set of seven live demos spanning everything from content creation to brand compliance, all built as working prototypes. While still experimental, the gap between Sneaks-stage concepts and real-world rollout is narrowing fast, highlighting Adobe’s future-facing ideas are increasingly becoming near-term realities for marketers.
And with AI connecting data, content and teams more seamlessly than ever, Sneaks 2026 didn’t just hint at what’s coming, it also showed how quickly it could arrive.
From the 500 submissions, seven projects were selected to be demonstrated live, each a working prototype built by engineers and researchers across Adobe. “They’re real prototypes created by real engineers,” Eric Matisoff, principal evangelist, Adobe said. “They haven’t gone through the rigorous QA that we do when we put something into production.”
The session, hosted alongside Iliza Shlesinger, an award winning stand up comedian, actor and writer, focused on how AI is being applied across marketing and customer experience workflows, from creative production through to personalisation and testing.
Following the demonstrations, the audience then. voted in real time, with Project Face Off named the winner.
So what are we waiting for, below are the demos:
Project Face Off (winner)
Project Face Off was voted the crowd’s favourite Sneak. The tool focuses on pre launch testing, simulating A B testing before campaigns go live by generating synthetic user personas and modelling how they interact with different creative options. It predicts which variation is most likely to perform best and explains why. In the demo, it compared multiple webpage designs, selecting a winning option based on readability and clarity, and introduced a “tournament mode” to test multiple variations simultaneously.
Speaking ahead of the session, Matisoff pointed to this as one of the most immediately applicable ideas.
“It’s taking something everyone already does, testing, and compressing it down,” he said. “You’re able to get directional insight before you even go live.” He also identified Face Off as a standout among this year’s cohort, noting it is likely to resonate because it fits into existing workflows.
Project Concurrent
Meanwhile, Project Concurrent introduced “living” design assets.
By linking elements within a creative such as text, pricing or locations directly to live data sources, the design continues to update automatically even after export. In the demo, a tour poster dynamically updated cities, rankings and ticket prices in real time, removing the need for repeated manual edits.
Project Page Turner
Project Page Turner explored real time website personalisation.
Instead of serving pre defined audience segments, the system assembles a webpage dynamically based on an individual user’s intent. In the demo, a single query generated a tailored product experience in seconds, pulling together relevant product information, use cases and supporting content. As the user continued browsing, the system adapted in real time, building a more detailed profile and refining the experience accordingly.
Project Wise Wire
Project Wise Wire focused on email production, converting design files into fully coded, personalised emails.
By analysing the structure and hierarchy of a design, the tool generates production ready HTML, maps content to customer data fields and applies personalisation logic. It also supports dynamic content generation and testing across multiple audience scenarios.
Project Test Kitchen
Project Test Kitchen introduced a collaborative approach to AI assisted design.
Using mood boards, prompts, sketches and brand assets, teams can co create and refine outputs together. The system generates multiple design directions across different styles and compositions, then allows users to break designs into components and recombine them. In the demo, even a hand drawn sketch was incorporated into the final outputs.
Project Asset Amplify
Project Asset Amplify focused on scaling campaigns across channels and audiences. Starting with a single set of assets, the tool generates a full suite of outputs tailored to different platforms and segments. In the demo, a car campaign was adapted into distinct website and social executions for Gen Z and millennial audiences, before extending into a personalised storybook experience.
Project Tailored Takes
Project Tailored Takes, powered by Adobe Firefly, demonstrated how video content can be generated and localised at scale. Using a small number of base assets, the tool creates multiple variations of a video, adapting visuals and details for different regions and audiences. In the demo, a single coffee ad was localised for markets including Japan and South Korea.
