Adobe MAX 2026 previews nine AI experiments; two are public

By Tom Whitfield

The Nov. 11, 2026 Adobe MAX Sneaks rundown named nine experimental AI projects, even though the accompanying headline counted 10; only two—Project Deep Brush and Project Hands On—accept public trials now.

With MAX 2026 in full swing, Adobe presented the projects as experimental generative-AI and workflow technology spanning imaging, video and audio. Preview author Dunja Đuđić Kalinin viewed the program as Adobe’s response to growing interest in agentic AI, not a broad product launch. She also warned that some demonstrations may never become shipping features.

What can the public test now?

Adobe has opened Deep Brush and Hands On to anyone through the new Adobe Labs Playground, marking the first time two projects from this year’s Sneaks lineup accept public trials.

Deep Brush uses AI to interpret an image’s underlying shapes, colors and lighting, then makes brushstrokes follow surface contours while preserving natural shadows.

Hands On adds on-canvas controls that use AI understanding so a creator can manipulate scene elements directly with a hand.

Kalinin, a Novi Sad, Serbia-based photographer with 15 years of experience, wrote: “I am difficult to impress, but Project Deep Brush has left me speechless.” The preview supplied no benchmarks, pricing, operating-system details or export limits for either experiment. A hands-on trial may show the interface, but it does not establish production reliability.

Takeaway: Public testing currently means two experiments, not a general Adobe release.

Which tools could change production work?

Seven other projects remain demonstrations. Their proposed uses matter because each targets labor, consistency or control in creative work:

• Project Scene Sweep tracks characters, locations and scenes across a video sequence, then applies one edit or adjustment consistently across cuts and multiple camera angles.
• Project Screen Test lets creators define a fictional character’s look, voice and personality, then hold natural-language voice chats and virtual auditions to test responses in different scenes.
• Project Live Set alters subjects, transforms environments or redirects camera angles while a video scene continues playing.
• Project On Cue turns a hand-drawn signal, or one extracted from video and audio tracks, into an editable timing guide for generated visual actions and effects.
• Project Sound Bender lets users describe effects with text, extract or generate audio inside a mix, and shape the result in real time on a spatial control surface.
• Project Sight Span converts flat 2D images, photos or illustrations into navigable 3D environments, opens new camera angles and generates content beyond the original frame.
• Project Rinse Repeat records specific edits or retouches on one image element, adapts them across other people or objects, and leaves the result open to refinement with native Photoshop tools.

Kalinin singled out Rinse Repeat for product, fashion and portrait photographers: “Product, fashion and portrait photographers would love this one, as it makes tedious edits so much simpler.” Her examples included cleaning a beauty shot with several people and removing pimples from teenagers in a group shot after cleaning one face.

The practical promise is less retouching time and more consistent delivery, which could affect studio labor costs and turnaround. The report offers no measured speed, consistency or failure rate, however, and Adobe has not said whether Rinse Repeat will ship.

Takeaway: Rinse Repeat offers the clearest potential labor saving, but a conference demonstration is not yet a production commitment.

What does the MCP announcement change?

For commercial and enterprise users, Adobe announced a Model Context Protocol (MCP) integration designed to let selected high-volume users connect Creative Cloud with business software including Slack, Drive and Jira.

Unlike Blackmagic Design’s implementation of an Open MCP protocol, Adobe focuses on handoffs across separate apps. Its proposed agent can round-trip graphics, motion design, video and enterprise brand assets throughout Creative Cloud.

The target users are marketing teams, commercial studios, production houses and technical engineers. Adobe says they could enforce house standards for branding, replace logos across multiple campaign documents and complete repeatable processes through natural-language prompts instead of hours of human effort.

The announcement gives no general-availability date, pricing, eligibility rules, permission model or measured time savings. It also does not say whether MCP access will come with existing Creative Cloud subscriptions or require a separate fee. Studios therefore cannot yet price the workflow or determine which staff and assets it will cover.

For a working studio, the useful test goes beyond moving a file between apps. It requires proof that an agent can preserve brand rules, file versions and approval ownership across the stack; the announcement documents none of those safeguards.

Takeaway: MCP could centralize repetitive asset work, but its commercial terms and operating controls remain unspecified.

What should photographers watch next?

Kalinin, describing herself as an old-school Photoshop user, called the demonstrations science fiction and said the experience was fun but somewhat frightening. Her closing question was which projects she would want implemented; commercial buyers need firmer answers on release dates, supported files and repeatable results.

Only Deep Brush and Hands On have public access, and the preview documents no commercial terms for either. Takeaway: As Adobe decides which demos become products, photographers should wait for release dates, pricing and repeatable performance rather than treating the 2026 showcase as production-ready.