Adobe's Sneaks session at MAX 2026 gave vector designers one demo worth tracking: Project Link Up, a Sneak that turns a plain drag into an editing instruction [1] [2]. In the prototype, an object dragged onto a scene reads the scene's perspective, scale and style, fuses into place and comes out as fully editable vector artwork [1]. The ten-project batch was shown in Miami on November 11 [2], and Adobe now lists Project Link Up on its Labs site under a November 2026 label [3] with a documented argument behind it: chat prompts are inadequate for precise visual work, and the company argues that future tools should support direct manipulation and other ways to express intent [4]. Nothing is available to install today. The actionable step is to watch the project's Labs page, which carries a signup for anyone who wants to know if the Sneak becomes available as an experiment [1].

What the drag changes in a vector file

Project Link Up starts from an observation about a familiar action. The team notes that a drag can carry more intent than the final position of an object, "what the creative moves, where they move it, and how quickly" [1]. In the prototype, the drag gesture becomes what Adobe calls semantic fusion, combining an understanding of both objects with the drag path, its velocity and creative controls, so the two elements influence each other in real time and the result remains editable vector artwork [1]. Adobe names visual continuity as the hardest problem, because small shifts in scale, color or layering can make the interaction feel disconnected from the artwork [1].

The Labs page lists four uses [1]. Illustrators and concept artists can place 2D or vector assets into scenes with automatic perspective matching, "reducing the need to redraw objects at new viewing angles" [1]. Designers can merge illustrated elements with letterforms while keeping the type editable for branding, logotypes and packaging. Two more cover icon design, where elements blend into distinctive icons, and gesture-driven direction, where drag motions guide composition and movement. In a Creative Bloq interview, Hailin Jin, a senior principal scientist and research lead for imaging and design at Adobe, gave the concrete case: placing an illustrated car into a parking space can involve separately adjusting its perspective, scale and style [5]. Jin also confirmed a control surface in the current prototype: "a creativity slider lets the designer guide how subtle or expressive the adaptation should be" [5]. Six team members, led by Ankur Krishna Gautam, built the project [1].

Adobe's own research argues for the gesture

The bet has a paper trail. Adobe Labs published a research article in November describing eight qualitative studies with 82 established and aspiring creative professionals and a review of 27 papers [4]. Its conclusion, in Adobe's words: "Chat is the most common interface for generative AI, but it is insufficient for precise visual work." The same article argues that creatives refine their intent through the process of making rather than describing it up front, and it splits creative work into four phases, from open-ended exploration to detailed edits, with the edit phase requiring precise visual guidance such as annotation, selection or direct manipulation [4].

Adobe also cites a 2024 study applying direct-manipulation principles to chat interfaces, in which participants completed tasks 50 percent faster, using 50 percent fewer prompts that were 72 percent shorter [4]. An art director quoted in the article describes the failure mode: "You would show it a picture of a car in a parking lot…it gives you a new kind of car and the lights are in a different position… it's basically trying to show off, which is not what we need." Adobe lists its own existing tools, including Firefly Boards, AI Markup and Precision Flow, which "incorporates a slider to visually explore variations of an image," and it reads competitors — Anthropic, OpenAI and Google among them — as moving toward canvases and objects [4]. The article states its own limits: the sample skews toward visual disciplines and is not representative [4].

What it is not yet

The boundary is stated on Adobe's own page. Project Link Up is a Sneak, shown as a demo, and availability is conditional: the Labs page offers a signup "if this Sneak becomes available as an experiment" [1]. The reveal sits on the MAX 2026 Sneaks session page as a video [6]. The page frames the capability against "today's vector tools" generally and names no target application [1]. What ships today in the Illustrator ecosystem, per the pages Adobe Labs links as related tools, is narrower: Turntable turns vector assets into 360-degree views that stay editable, and Concept to Vector transforms raster images or rough sketches into editable vector artwork with generative AI prompts [1].

What to watch from here

The rest of the Sneaks batch points the same way: fewer one-shot generations, more controllable edits. Project Scene Sweep understands the characters, locations and scenes across a sequence, so an edit made from a single text prompt or adjusted frame applies across every shot, cut and camera angle; Creative Bloq's rundown includes removing a person or prop, changing a wardrobe without a reshoot and replacing an unlicensed logo or artwork to clear the project [2]. Project Sound Bender lets editors pull out a sound by describing it [2]. Variety's report names Project Live Set, which modifies an existing video with AI-generated effects while the video plays [7]. CNET's coverage frames the batch around "small, precise edits — the very opposite of an AI hallucination" [8].

Creative Bloq's headline makes the stronger claim, that Link Up is making design playful again [5]. The verifiable signals are narrower. The demo becomes something a vector designer can act on only if Adobe moves it from Sneak to public experiment, and the Labs page is where that would show up [1]. If it arrives with the editable-vector promise intact, the drag will have become a working input channel for precise generative work, and the same editability test will decide which of the year's Sneaks demos turn into usable tools.