What has caught my attention at Adobe MAX in Miami is how AI is finding a place within the creative process. Artists and designers can use it to bring their ideas to life faster, and explore possibilities that would have been much harder to pursue before. Their ingenuity is still very much at the centre of the work.
I find that encouraging. A model can produce an impressive image in seconds. Developing an idea you would be proud to put in front of a customer takes longer. It has to answer a brief, reflect what the customer needs to understand, and live up to the promises the brand has already made. The choices along the way matter: what to try, what to discard, and when an apparently impressive result has missed the point entirely.
The announcements at MAX make me think about how those choices could change. For CMOs, there is more to consider than whether the next campaign can be produced more cheaply. I am interested in what our creative colleagues could achieve if more of their time and energy were available for the work they most want to do.
Focus On: More room for craft
Across Adobe’s Creative Cloud and Firefly announcements, a common direction is becoming visible. Generating, editing, reviewing, and adapting work are being brought closer together, with more scope for people to move between AI assistance and the precise controls they already know.
Firefly Design, introduced in beta, brings several creative disciplines onto one canvas, with collaboration and review included. Firefly Graph, also in beta, lets people build reusable workflows connecting AI models and Adobe tools. Its expanded AI assistants, still in beta, are designed to carry out multistep tasks within creative applications, leaving people able to refine the work themselves.
Adobe also previewed integrations through which an external agent could work across its applications while preserving editable project structures, including layers, vectors, and video sequences. A private beta is planned for early 2027. That is a preview of a possible working method, not something a team can deploy today. And a convincing demonstration is not yet a proven production process.
Creative agents themselves are not new to MAX: Adobe had announced an expansion in June. What interests me is how the tools are developing around professional creative work. Adobe says Photoshop’s Generative Restyle turns prompt-based edits into adjustable sliders, while Firefly’s Plan Mode lets someone revise an agent’s proposed steps before it runs. An editable file at the end is useful; being able to express and revise an intention during the work matters more.
Imagine a campaign for an outdoor clothing brand’s repair service. The product, awkwardly for a conventional launch, is a jacket the customer already owns. A designer could explore documentary photographs of worn kit, an illustrated repair manual, or patches treated almost like medals. This is a hypothetical brief, but it shows why the process matters. Making each route presentable used to limit how many could get a fair hearing. AI may make it easier to explore the less obvious ones.
Suppose the repair-manual route wins. The jacket looks suspiciously new, and the stitching should feel hand-drawn. Can the designer change those details without losing everything useful in the first attempt? Can they keep working directly on the composition? More possibilities are of little help if refining any one of them becomes a struggle with a prompt box.
When everything can be made, choosing what should be made becomes much more important. A wider range of routes asks more of the brief and the review. What will this idea help a customer understand? Which treatment deserves another hour, and which one is merely plausible? The experienced creative who can explain those choices gives colleagues something more valuable than a folder of generated options.
That is the kind of augmentation I am excited about. Taking repetitive adaptation off someone’s hands could make room for a more ambitious idea or for details a deadline would previously have forced them to leave alone. We should ask creative people where they see those opportunities. They know which parts of the process help them think and which simply consume the afternoon.
Bringing our own creative knowledge into the process
The wider alliance announced by Adobe and Disney offers another perspective on that opportunity. It includes custom Firefly Foundry models tailored to Disney properties, building on work already underway with Imagineering. Adobe describes models supporting sketch-to-image concept development and the generation of franchise-specific creative assets.
We do not all have Disney’s characters, resources, or production requirements. Still, the principle is worth examining: creative tools become more useful when they can work with the particular material and knowledge that make a business distinctive.
Many marketing organisations have years of photography, illustration, design, campaign concepts, and experience of what customers respond to. Some of that knowledge is documented; some remains with the people who made the work. Making it accessible could help a new idea develop from a much better starting point than a generic prompt.
I wouldn’t begin by deciding that everything in the archive should train a model. Some material may be useful as a reference, some may justify a customised tool, and some will require rights we do not have. Experienced colleagues also understand why a creative choice worked. That knowledge is not always in the final files.
Their involvement is essential if we want useful creative context rather than an automated way of reproducing yesterday’s campaign. The archive should help us attempt something new, with enough understanding of the brand to make that attempt worthwhile.
Learning to let go of parts of the work
For all this technical progress, I think the harder gap remains change management and education. We can make a tool available quite quickly. Helping people become comfortable with it takes more care.
Someone who has spent years developing a craft may be understandably reluctant to hand part of the process to AI. There is a perception that using AI is somehow cheating, as though the assistance diminishes the creative contribution. But reluctance can also come from knowing precisely how a piece of work can fail.
I don’t think that is a helpful way to judge the work. We would hardly suggest that an illustrator’s ingenuity counts for less because they use Illustrator. Computer-assisted illustration introduced efficiencies and changed how work could be made; the person still had to have something worth making. The comparison does not settle every question about generative AI, but it should make us careful about equating manual effort with creative merit.
In a MAX interview, Newell Brands CMO Nick Hammitt described his team adapting approved content for different retailers, formats, and languages. He was much more cautious about autonomous visual creation. In pilots, he said, logos and other details have been wrong, particularly consequential for products where trust matters. This is one customer’s account, in a programme that identifies Adobe as a partner, rather than an independent benchmark. It is still a useful distinction: delegation that works for an approved source asset may not work equally well when the tool has to invent the asset itself.
There are genuine questions about generative AI that deserve their own answers, including quality, the rights to source material, and how creative contributions are acknowledged. A colleague asking about those things is not necessarily resisting change. The organisation should clarify which tools and materials people can use, so they can concentrate on learning instead of resolving every uncertainty themselves.
None of this means people should be embarrassed about preferring to do a particular task themselves. Sometimes working through a detail is how an idea develops. We need enough familiarity with these tools to decide where assistance helps, and enough freedom to decline it when it gets in the way.
A demonstration followed by an invitation to experiment over the weekend would not be much of an education programme. I’d want people to have time to work on real creative problems with colleagues who can help them. Let a designer compare an assisted approach with their usual method, examine what improved and what became awkward, and discuss the result without feeling obliged to declare the technology wonderful.
It would also help to keep examples of how a piece developed, including the human interventions that made the difference. A final asset can conceal the process that produced it. Seeing the decisions, rejected attempts, and refinements would give colleagues a better understanding of where AI was useful and where the craft remained demanding.
Leaders have a part to play in that confidence. If we say AI gives people more room for craft, the work we commission and the time we allow should reflect it. Filling every released hour with more production would make that promise rather difficult to believe.
What we could do with the possibilities
My ambition would be to explore more creative territories, push the quality of the work, and attempt things that previously fell outside the available time or budget. Personalisation belongs in that ambition too. A smaller audience might receive an execution developed around its actual needs, rather than a hurried variation of an asset made for somebody else.
We would still have to establish that the additional work was worth doing. Producing more versions is easy to count; learning whether they improve the customer experience or commercial result takes a better comparison.
I would begin with one campaign where the team sees a genuine opportunity. Agree what we want to improve before choosing the tools. It might be the quality of the concept development, the range of relevant audiences we can serve, or the time available for refinement. Compare the assisted workflow with current practice, including the effort spent briefing, reviewing, and correcting the work, rather than measuring the generation step in isolation.
Then assess the work against a shared creative standard, and test its effect in market. Randomised audiences or suitable market tests can compare the new approach with the existing one at comparable budgets, measuring incremental sales or contribution margin over an appropriate buying cycle. Production and review costs belong in that assessment. Brand recognition and customer response can help explain the results without becoming substitutes for commercial impact.
That gives us a basis for deciding where to expand, and for explaining the investment to colleagues in finance. It also leaves room to discover that a more ambitious approach produces better work without reducing the total cost. That may be a very good investment, if the improvement earns its place.
For me, the interesting question after MAX is what we choose to commission. More versions of the familiar might be useful. So might an idea we previously dismissed because we could not afford to explore it properly. Our job is to tell the difference, give the better possibilities a fair test, and make sure the people with the craft have room to do their best work.
Follow me
That’s all for this week. To keep up with the latest in generative AI and its relevance to your digital transformation programs, follow me on LinkedIn or subscribe to this newsletter.
Disclaimer: The views and opinions expressed in Chronicles of Change and on my social media accounts are my own and do not necessarily reflect the official policy or position of S&P Global.
