Adobe Sneaks AI Tools Focus On Gestures And Conversation Instead of Prompts
Adobe's Max Sneaks lean on dragging, talking and copy-pasting, but only one of the nine projects named across two outlets has a stated way to try it.
Adobe used its Max conference this Wednesday to show off Sneaks, the experimental tools it builds to test ideas before deciding what deserves a place in Creative Cloud.
The Verge, which covered the showing and says almost all rest on generative AI, describes them as tools that turn images into explorable 3D scenes, let you design characters by talking to them, and make photo editing a copy-and-paste job.
Read together, they share something less obvious than the AI label.
Three tools try to stop you from re-typing the prompt
Say an AI edit lands almost right. Today you reword the prompt and try again. Project Hands On analyses an image and adds control points, so you reshape details by dragging instead, per Creative Bloq.
Project Link Up reads a drag gesture, using what the two objects are plus the path and speed of your drag, and keeps the result editable as vector art.
That editable vector output also connects with the wider range of graphic design software, including Illustrator and Photoshop.
Rinse Repeat, from The Verge’s list, copies an edit you made on one subject and adapts it to others in the same photo. Retouch one face, and it works out the rest of the group. Shadows get the same treatment across products of different shapes.
Rinse Repeat is built around Photoshop, and The Verge says the copied edits are recreated with native Photoshop tools, so they stay editable.
Talking to a character and stepping into a photo
Project Screen Test is the odd one. You give a character a few traits, say “anxious, shy, awkward,” pick a voice, then talk with it, FaceTime-style, to fill in its backstory, fears and quirks, The Verge details.
Unlike consumer chat apps like Character.ai, which focus on open-ended social roleplay, Screen Test treats the dialogue as a production audition to lock down consistency.
A Personality tab lets you edit what comes out of those chats by hand. The report calls it a standalone exercise, with no clear route into Adobe’s video tools, and says its characters felt more consistent than most AI ones.
Project Sight Span turns a flat image into a 3D space you can walk around in. The preview reportedly looks pixelated until you settle on an angle. It also says the result is hard to still call a photograph.
Only Deep Brush has a stated way in
Project Deep Brush splits an image into what Creative Bloq calls “physically decomposed layers,” so a logo wrapped on a bottle follows its curves and lighting. It’s the only Sneak with a stated way to try it: Adobe’s new Labs Playground site.
Then Project Sound Bender. The Verge has you describe the sound you want, such as “bubbling sci-fi cauldron noises,” and the tool generates it, synced to your clip, on a square canvas you drag around.
The Verge notes no official guarantees that these experiments will launch. Balancing this heavy AI push with legacy software updates, like its hesitation around retiring Adobe Animator, shows the company carefully weighing its next steps.
But as Adobe research scientist Gautham Mysore describes the aim as “more precise control”, the takeaway is clear: the future of creative software is about making technology feel as responsive as a physical tool.
Source: Adobe’s AI experiments let you walk around in photos and interview your characters
