In the 1990s, Business Process Reengineering (BPR) was the Big Bet. Companies launched tightly controlled pilot programs with hand-picked teams, custom software, and executive backing. The results dazzled on paper.
But when it came time to scale? Reality hit. People weren’t ready. Systems didn’t connect. Budgets dried up. The pilot became a cautionary tale, not a blueprint.
We’ve seen this before with Lean, Agile, even digital transformations. Now it’s happening again with AI—only this time, the stakes are different. Because we’re not just implementing a new solution, we’re building into a future that’s unfolding. Technology is evolving faster than most organizations can learn, govern, or adapt right now. That uncertainty doesn’t make transformation impossible—but it does make it easier to get wrong.
And the dysfunction is already showing up—just in two very different forms.
Two roads to the same cliff
Today, we see organizations falling into two extremes. Most companies are either overdoing the control or letting AI run wild.
Road 1: The free-for-all
Everyone’s experimenting. Product teams are building bots, prompting, using copilots. Finance is trying automated reporting. HR has a feedback chatbot in the works. Some experiments are exciting. Most are disconnected. There’s no shared vision, no scaling pathway, and no learning across the enterprise. It’s innovation by coincidence.
Road 2: The forced march
Leadership declares an AI strategy. Use cases are approved centrally. Governance is tight. Risk is managed. But the result? An impressive PowerPoint, a sanctioned use case, and very little broad adoption. Innovation is constrained before it ever reaches the front lines.
Two very different environments. Same outcome: localized wins, system-wide inertia.
The real problem: Building for optics, not for scale
Whether you’re over-governing or under-coordinating, the root issue is the same: Designing efforts that look good but aren’t built to scale.
Here’s the familiar pattern:
- A team builds something clever.
- It works in their context.
- Others try to adopt it.
- It doesn’t stick.
- Momentum dies. Energy scatters. Or worse, compliance says no.
Sound familiar?
It’s not that the ideas are flawed. It’s that they’re built in isolation with no plan for others to adopt, adapt, or scale them. There’s no mechanism for transfer, no feedback loops for iteration, and no connection to how people actually work across the organization.
So, what starts as a promising AI breakthrough—a smart bot, a helpful copilot, a detailed series of prompts, a slick automation—quietly runs out of road. It works for one team or solves one problem, but without a handoff or playbook, there’s no way for others to plug in. The system stays the same, and the promise of momentum fades, lost in the gap between what’s possible and what’s repeatable.
We’ve seen this before
These aren’t new problems. From BPR to Agile, we’ve learned (and re-learned) that:
- Experiments are not strategies. Experiments show potential, not readiness for adoption. Without a plan to scale, they become isolated wins—interesting, but not transformative.
- Culture is the operating system. If the beliefs, behaviors, and incentives underneath aren’t aligned, the system breaks—no matter how advanced the tools.
- Managers matter. Without their ownership and support, change stalls.
- Behavior beats code. Tools don’t transform companies. People do.
Design thinking promised to bridge this gap with user-driven iteration and empathy. But in practice? Most efforts skip the hard parts. We tinker, test, and move on—without ever building the conditions for adoption.
AI and the new architecture of work
Many organizations treat AI like an add-on as if its something to bolt onto existing systems to boost efficiency. But AI isn’t just a project or just a tool—it changes the rules of how decisions are made, how value is created, and what roles even exist. It’s an inflection point that forces companies to rethink how work gets done.
Companies making real progress aren’t just chasing use cases. They’re rethinking how their organizations operate, end to end. They’re asking:
- Have we prepared people to reimagine how they work with AI—not just how to use it?
- Are we redesigning workflows, decision rights, and interactions—not just layering new tech onto old routines?
- Do we know what success looks like when it’s scaled and sustained—not just when it dazzles?
If the answer is no—whether you’re too loose or too locked down—you’re not ready.
The mindset shift AI demands
AI isn’t just a tech rollout. It’s a mindset shift that asks leaders to reimagine how value gets created, how teams operate, and how people grow. But that reimagination isn’t about the tools. The tools will change…rapidly. It starts with new assumptions, new stances and a new internal leader compass.
Here are three essential mindset shifts every leader must make, not just to keep up with AI but to stay relevant in a world being reshaped by it:
1. From automation to amplification
Old mindset: AI automates tasks and cuts costs.
New mindset: AI expands and amplifies human potential enhancing our ability to think strategically, learn rapidly and act boldly. The question isn’t what AI can do instead of us but what it can do through us – helping people make better decisions, move faster, and focus on higher-value work.
2. From efficiency to reimagination
Old mindset: How can we use AI to make current processes more efficient?
New mindset: What would this process look like if we started from zero with AI as our co-creator, not a bolt on?
3. From implementation to opportunity building
Old mindset: Roll out the tool. Train everybody. Check the box.
New mindset: AI fluency is a new core human capability creating new realms of curiosity, sophistication in judgement and opportunity thinking. Soon, AI won’t be a one-time training. It will be a part of how we define leadership, collaboration and value creation.
From sparkles to scale
In most organizations, the spark isn’t the problem—good ideas are everywhere. What’s missing is the ability to translate those isolated wins into something durable, repeatable, and enterprise wide.
Too many pilots are built to impress, not to endure. They dazzle in one corner of the business but aren’t designed for others to adopt, adapt, or sustain. The result? Innovation that stays stuck in the lab—or dies.
Designing for scale means thinking beyond the “what” to the “how”:
- How will this spread?
- What behaviors and systems need to change?
- Can this live in our whole world, not just my ‘sandbox’?
It’s not about chasing the next use case. It’s about setting up the conditions that allow innovation to take root, grow, and multiply—without starting from scratch every time.
Here’s how to make that shift:
1. Test in the wild, not just the lab
Skip the polished demo. Put your solution in the hands of real users, in real conditions, with all the friction that comes with it. Use messy data. Invite resistance. That’s where the insights live—and where scale begins. If it only works in ideal settings, it doesn’t work.
2. Mobilize managers
Executives sponsor. Front lines experiment. But it’s team leaders who connect and spread. Equip them as translators and expediters, not blockers. Every leader is a change leader.
3. Hardwire behaviors, not just tools
The biggest unlock in AI is not the model—it’s the muscle. Invest in shared language, habits, and peer learning that support new ways of working. Focus on developing behaviors that scale, such as:
- Change readiness: the ability to constantly spot opportunity, turn obstacles into new possibilities and help teams pivot
- Coaching: getting the best out of your AI “co-workers” just like human ones
- Critical thinking: applying human judgment where it matters most—context, nuance, and ethics
4. Align to a future state vision
To scale beyond one-off wins, people need a shared sense of where they’re headed. A clear future-state vision acts as an enduring focus—allowing everyone to innovate in concert. That alignment doesn’t stifle innovation. It multiplies it—turning a thousand disconnected pilots into a coherent transformation.
5. Track adoption, not just ’wins’
Don’t mistake a shiny, cleaver prompt for progress. A great experiment means nothing if it can’t be repeated by many, many people. From day one, design with scale in mind: Can this be adopted elsewhere? What would need to change for it to work across teams, roles, or regions? Build for transfer, not just applause.
The real opportunity
AI will not fail because the tech wasn’t good enough. It will fail because we mistook experiments for solutions—or because we governed innovation into paralysis.
You don’t need more control. You don’t need more chaos. You need design for scale—not just scale in hindsight.
Let’s stop chasing sparkles. Let’s build systems that spread.
