AI-enabling the past is why your transformation is underperforming: Cannes 2026

The promise was that this would fuel faster, better work. But for a lot of employees, the reality feels more complicated. When tools don’t sync with each other and work has to move between disconnected systems and teams, people are left still stitching the process together manually.

In short: AI has done a great job at speeding up individual tasks, but it hasn’t necessarily improved overall work flows.
 
That tension was the underlying refrain we kept hearing in the DEPT® Secret Garden at Cannes, leaders from some of today’s most prolific brands talked candidly about what it takes for AI to change how organizations actually work.

Most businesses have had no shortage of AI activity and experimentation over the past two years. But because AI tools are typically dropped into the same disconnected processes, the “transformational” benefits stalled, making the impact simply another layer of technology on an already complicated way of working.

eBay CMO Adrian Fung captured the problem with a line that feels uncomfortably familiar to anyone who has been navigating a rushed AI rollout: “It’s like you went into an office building, gave everyone a chainsaw and said, ‘This is the best thing to cut stuff down. Go.’”

From the employee side, that’s how a lot of AI adoption has felt. New tools, little training, and even less coordination. People are told to experiment and find efficiencies with AI, bringing it into their day-to-day work. Then they return to processes that were built for a different pace of production, with the same approval paths, review habits, team boundaries, and inherited assumptions about how work should move. The result is often somewhere between chaos and frustration.

This is where the gap between tool adoption and transformation starts to show up. A designer can use AI inside Figma and move through concepts more quickly. A developer can use AI to support front-end coding and move faster, too. But if the broader process still treats design and development as separate stages connected by the same old clunky handoff, the organization has only made two existing steps faster. The shape of the work hasn’t really changed, nor has the outcome.

The larger point was that tool-first adoption puts too much emphasis on usage-for-usage’s sake and too little on improving the system as a whole. 

Once AI enters the work, the harder questions become more important: how it changes the interactions between teams, how it changes the relationship with the customer, and what it makes possible that the business couldn’t do before.

Fung shared a phrase from eBay’s internal approach that nails the direction organizations should be moving in: “Don’t AI-enable the past; AI-invent the future.” 

AI-enabling the past starts with how the company already works and by looking for places to layer AI on top to shave time. The bigger shift starts when AI is allowed to challenge the structure of the work itself.

Most organizations are still evaluating AI through the lens of existing tasks. Can it write the brief, summarize research, or generate asset variations faster? Those are all helpful areas to gain efficiency, but they keep the focus on isolated moments inside a workflow. Real AI transformation,
productivity gains, and business growth are shaped by the connections between those moments, where work is handed from one team to another, translated across functions, slowed down by approvals, or reshaped to fit systems that were never designed for this level of speed.

AI-inventing the future doesn’t begin with a tool. It has to start with the outcome the business is trying to create: faster campaign production, more relevant customer experiences, shorter paths from insight to activation. Once that direction is clear, AI becomes part of a larger redesign rather than another layer added to the existing process.

For many companies, the answer to AI transformation isn’t to rip out every platform they’ve already invested in and start again, nor is it to add on even more standalone AI tools. Typically, what’s missing is an orchestration layer between what already exists.

An AI-enabled orchestration layer connects the tools, data, teams, and decisions involved in getting work from idea to execution. It’s less of a destination employees have to visit and more of an intelligent connective tissue across the work itself. An orchestration layer can, for example, carry context between steps, trigger the next action, and translate information between the systems. 

Altogether, this changes the role of AI from making individual tasks faster to helping the entire process move more intelligently—while allowing organizations to get more from the tools and platforms they’ve already invested in.

AI-inventing the future isn’t about designing people out of the process. It actually means being more intentional about where they enter it and placing greater value on the unique emotional and decision-making skills that only a human can leverage. 

Today, people are often brought into every stage of a process because the systems around them can’t make work move on their own. They transfer information between tools, chase approvals, reformat outputs, and review work simply because the workflow requires someone to keep it moving.
An empathy-first approach to AI creates an opportunity to change that by identifying the moments that genuinely require human context, judgment, taste, or an understanding of another person, and allowing AI to take on more of the coordination, translation, production, and repetition around those moments.

Instead of saving human judgment for a final review at the end of a long process, organizations can move it earlier, where it can shape the direction of the work. People can spend more time interpreting context, making decisions, and responding to customers, audiences, or colleagues, while AI handles more of the work required to carry those decisions through the system. That’s augmentation in a much more meaningful sense. 

Rather than placing emphasis on acquiring every new tool that enters the market and using AI to help someone complete the same task 20% faster, this approach to AI focuses on designing the workflow so the human contribution is focused where it creates the most value.

AI-enabling the past will keep producing a familiar pattern: lots of visible experimentation, some meaningful efficiencies, and results that don’t quite match the ambition. Getting more from AI needs a more deliberate redesign—connecting the tools, teams, and data already involved in the work and redesigning the path from idea to output.

Start with the outcome. Identify where human empathy, context, and judgment matter most. Then orchestrate AI around those moments so the rest of the work can move faster and more intelligently.

The measure of transformation is whether the business works differently because of AI, not whether more people are using it.