Adobe has spent years building agentic AI, and this year’s Adobe MAX conference in Miami suggests that work is reaching its next stage. The company is substantially widening its Creative Agent, a tool that takes instructions written in plain text and carries out editing work on the user’s behalf. Those instructions can be issued inside Adobe’s Firefly AI platform as well as within Creative Cloud applications such as Photoshop and Premiere.
From Assistant to Agent
Most AI features in creative software so far have been single-purpose helpers, such as a button that removes a background or a slider that generates a fill. An agent works differently. Rather than handling one isolated step, it interprets a broader goal, breaks it into smaller actions, and works through them in sequence. For photographers and videographers, that could mean describing the result you want and letting the software assemble the steps needed to get there.
The practical appeal is clear. Much of post-production is repetitive: batch color corrections, consistent cropping across a shoot, selecting subjects, cleaning up distractions, or organizing clips before the creative work begins. These tasks are necessary but rarely inspiring, and they can consume hours that would otherwise go toward shaping a final image or cut.
Why It Matters Across Photo and Video
Bringing the agent into both Photoshop and Premiere matters because modern content creators rarely work in one medium. A still photographer may produce a short video for social channels, and a video editor often needs to pull frames for thumbnails or marketing stills. A shared, text-directed approach across both environments could reduce the friction of moving between them.
Firefly sits at the center of this strategy. By allowing direction from the Firefly platform as well as from inside individual apps, Adobe is positioning its generative and assistive models as a connective layer across its ecosystem rather than a feature confined to one program.
Creative Control and Professional Caution
For working professionals, the important question is not only what an agent can do but how much control remains in human hands. Photography has long depended on intent, and retouching decisions carry ethical and commercial weight, especially in editorial, documentary, and journalistic work. An agent that makes changes at scale needs to be easy to review, easy to reverse, and transparent about what it altered.
Adobe has previously emphasized content provenance through its Content Authenticity Initiative and Content Credentials, and industry observers will likely watch closely to see how those systems interact with more autonomous editing. Clear records of edits will matter to clients, publishers, and anyone who relies on the integrity of a image or recording.
What to Watch Next
Expanded agentic tools promise to change daily workflows, but their value will depend on reliability, precision, and how well they respect a photographer’s or editor’s established style. Creators should evaluate new features through real projects rather than demonstrations alone, paying attention to output quality, editing history, and any limits on file types or subscription tiers. Readers interested in exact availability and capabilities should consult Adobe’s official MAX announcements for current details.