Adobe Firefly Adds Project Memory in Private Beta

Adobe has announced an upgraded Firefly creative AI studio in private beta, connecting generation and editing while moving the platform toward project memory instead of one-off prompt sessions. Access through a waitlist limits availability, so the first test is whether creators can return to prior work without rebuilding every setup.

Adobe is pairing that memory pitch with a creator-control caveat. In a survey of more than 16,000 creators, 75% described creative AI as integrated or necessary to their work, while 85% said the final creative decision should remain theirs.

How Firefly’s Project Memory Works

Adobe’s Firefly platform creates and edits media from prompts, but this update changes how repeat work is organized. Elements, a Firefly feature for named visual pieces, lets creators save characters they have already made so they can reuse them across generations.

Adobe’s upgraded Firefly experience enters private beta with persistent context, reusable assets, and organized project workflows. Projects can keep creative context, assets, and generations together so a creator can resume previous sessions instead of rebuilding a character, room, product angle, or visual style from scratch.

For designers and video creators, that difference is workflow continuity rather than another prompt box. A saved room, character, or product angle can become a starting point for revisions, campaign variants, and client feedback instead of a visual setup that has to be recreated by memory.

Earlier in 2026, Firefly Custom Models already gave creators and brands a way to train image generation on their own assets for more consistent output.

Firefly AI Assistant, Adobe’s conversational creative helper, adds adjacent production tools, including brand kit generation for logos and color palettes. Quick Cut assembles clips into a first draft users can refine.

Across Premiere, Photoshop, Illustrator, InDesign, and Frame.io, Adobe’s broader creative agent is also rolling out as public betas. Forest Key, Adobe vice president of agentic AI for creativity and productivity, framed the interface as a flexible layer rather than a mandatory natural-language replacement for hands-on tools.

That control issue becomes practical when teams reuse one Firefly character across a launch campaign. Approvals and handoffs can change an asset, so project memory has to preserve enough revision paths to explain why one visual direction won over another. Without that record, the assistant layer risks hiding decisions that shape a campaign, storyboard, or brand system.

Creative AI Tools Compete for Workflow Memory

Creative AI suites are increasingly converging on workspaces. Canva’s official Magic Studio launch in October 2023 positioned the product early as a unified AI design suite for individuals, teams, and large organizations, and Canva Magic Studio remains a useful comparator for Adobe’s workspace approach.

Figma’s AI tools help craft prompt prototypes, automate repetitive design work, generate or refine images, and add design context to coding tools. Figma Make illustrates the same pressure around prompt-driven prototyping.

Competition also runs through production and video. Leonardo.Ai’s creative platform offers image and video generation, editing, upscaling, and creative production workflows, while Runway’s generative video tools and character products push creative AI toward repeatable production assets.

Adobe’s studio memory gives Firefly a workflow answer: keep the creative state inside the tool instead of asking users to reconstruct it every time.

What Remains Unclear

Adobe’s current pitch ties more assistant setup work to human approval rather than full automation. David Wadhwani, president of Adobe’s Creativity & Productivity business, framed the broader rollout around creators who still set the vision and make final calls. Survey results make that control test central to the private beta.

Private-beta access will test whether saved assets, Projects, and Elements work reliably across real design sessions, not only controlled examples. Teams will also need to know how project memory behaves across collaboration, brand governance, and revisions when several people touch the same creative system.

Adobe has not put a wider availability date in front of users yet, so the next concrete signal is when waitlisted teams can test Projects and Elements against client revisions, handoffs, and brand approvals.