When was the moment you realized AI could change everything?
For me, it was a simple conversation. I started asking an LLM questions about an internal data set I knew inside and out. In seconds, I had meaningful and articulate analysis, key trends and patterns. I walked away with new ideas and perspectives.
Whenever I talk to CEOs just starting with generative AI, I ask them to do the same thing: plug a trusted data set into an enterprise LLM. Then, start asking questions.
Often, the experience is eye-opening.
If AI can do all of this with one data set, they ask, what can it do in the hands of every employee?
That’s what the entire world is wondering, along with other questions like: What is the ceiling for AI’s capabilities? How far and how quickly will it advance? And how do we start using it safely and effectively today?
I recently had the pleasure of speaking with AI expert and Wharton professor Ethan Mollick. Ethan’s perspective on our current AI moment was sharp and pertinent. If you haven’t seen our “Executive Exchange,” I highly recommend watching the full video. It’s worth your time.
Since then, I’ve been thinking more about our conversation. I’ll share some of my thoughts below. If you’re a business leader in any capacity, this article will help you understand how to start successfully deploying AI in your business. I’ll also give you a 30-day road map to fast-track your success.
Nerves are leading nowhere
“Three sleepless nights” is what Ethan said it costs executives to get to know AI. And I agree. As I mentioned, I had my own moment of disbelief where I was amazed at what AI could do. Then, I couldn’t stop tinkering. What about this? Can I connect these two things? What happens if I do this?
Most leaders who start experimenting with AI have a similar experience. And most agree that three days of little to no sleep is a small price for gaining a better handle on such a disruptive technology.
And yet, many CEOs are still waiting, wondering how regulation and the competitive landscape will play out. This is a mistake.
I told one of our customers recently that if their AI project is waiting just behind their master data management project that’s been in the works for decades, their AI initiative will never launch.
That doesn’t mean clean, high-quality data isn’t important. In fact, it’s foundational. But it does mean you can’t wait until everything is perfect to get started. The stakes are too high to sit on the sidelines.
You have to start building processes around governance and implementation now. Generative AI isn’t like other technologies. As Ethan pointed out, you can’t delegate it to IT and ask them to roll it out because its implications are too far-reaching. It changes the entire scope of work. Realizing that is the first step toward successful implementation.
Productivity needs a target
The second thing you have to understand about AI going into 2026 is that productivity isn’t a magic potion. It will not magically generate more revenue for your business.
I was speaking to a colleague recently who joked about how one employee used AI to create a dense PowerPoint, and another asked AI to distill it into bullets. Yes, creating the PowerPoint was more “productive,” but it became a pointless circle of productivity.
Another example that Ethan shared: You can bring 400 PowerPoints to a sales call, but having more PowerPoints isn’t the outcome you want. I agree with Ethan here: productivity is only valuable when it’s aimed at a specific business outcome. It’s a lever, but it has to lead to better results. Start with the end and optimize from there.
How you view your talent will determine everything
For years, we’ve heard “people, process, and technology” as the driving force behind digital transformation. The strange thing with AI is that many leaders are inverting this.
They’re looking at the technology first, then building processes to support it, and finally, they’re looking at people as a variable cost. Unfortunately, many organizations may be tempted to substitute talent with tech.
This is a losing formula. It creates short-term savings at an untold cost. Ask yourself who wins: the companies using AI to do the same amount of work they do today, or the ones expanding their capacity by using AI in smart and imaginative ways?
You need a new operating model
To make AI work for your organization, you need a new organizational framework. Ethan talked about one that I believe is pragmatic and practical: Leadership, Lab, and Crowd.
The leader is you, it’s me, it’s anyone making decisions for the business. The Crowd is your employees, the ones using AI tools every day. And the Lab, perhaps the most pivotal part, is a dedicated team of employees constantly experimenting with AI, the ones ideating new use cases, benchmarking them against the status quo, and figuring out safe and smart ways to implement AI into core workflows.
If your organization is anything like mine, you have strict rules and regulations around how you build and ship and deploy things. The established processes around these things are crucial. But they’re slow. They weren’t designed for AI.
The idea behind the Lab is to isolate a group of specialists and let them experiment. At Alteryx, we implemented something similar and made one rule non-negotiable: nothing leaves the Lab until it’s reviewed and approved. Then, we encouraged them to be as ambitious as they wanted with AI.
I love Ethan’s framework because it feels like a missing piece. At Alteryx, we gave all of our employees access to the enterprise version of ChatGPT. The question we kept asking ourselves, though, was how to properly incentivize employees to share their generative AI workflows and use cases across the org.
In Ethan’s framework, everything filters through the Lab. The Lab understands what leadership wants, and the Crowd can freely bring their ideas to the Lab. Then, the Lab can refine, ship, or reject those ideas, and leadership can reward employees accordingly. Suddenly, you’ve activated your entire organization as a sort of AI lab. You’re innovating and experimenting as a business.
This is just one framework, of course. The takeaway is that you must change how you incentivize employees using AI. Otherwise, as Ethan pointed out, they’ll keep their breakthroughs to themselves, especially if sharing them leads to coworkers losing their jobs.
The CEO’s playbook for the next 30 days
If you’re a leader trying to move from ambition to impact, I would focus on these four steps.
1. Go hands-on with a frontier model using real data
Start with a trusted internal data set, something you already use to run the business, whether that’s pipeline, churn, budget-to-actuals, support tickets, or customer feedback. Load it into an enterprise LLM and start asking questions you might ask in a leadership meeting.
Once you’ve done this a handful of times, you’ll start to understand what these systems are good at and what they can and can’t do. You’ll gain context to understand where and how you can start applying it in your business.
2. Stand up a Lab
Give a small, dedicated team permission to move quickly and freely, with one clear rule: nothing leaves the Lab without review and approval. The Lab’s job is to take the best ideas from the business, turn them into repeatable workflows, and then put those workflows into the hands of employees.
3. Create a risk-tiered policy that enables usage
Trying to mitigate all risk will stall your progress forever, but that doesn’t mean you have to blindly accept all risk. Give people access to enterprise-grade LLMs; otherwise, they’ll use their own versions and potentially leak your data to unapproved tools. In addition, prioritize data platforms that have transparent, auditable, and governable data.
Then, once you’ve set your guardrails, encourage employees to freely experiment. You may be surprised by what they come up with. I know I’ve been.
4. Pick two high-value workflows and infuse AI into them
Choose two high-value workflows that happen every week but have a lot of manual work. Then, integrate AI directly into those workflows so that it actually changes how work gets done. Deploying two workflows successfully will give you a framework and ideas for bringing other pilots into production.
Success starts with you
AI is moving fast. At times, it’s hard not to feel like we have tech whiplash. But as business leaders, it’s our job to set the pace. AI change management has to come from leadership. It has to come from you.
Successful AI implementation requires rethinking your entire organizational structure and incentives. And to do that successfully, you have to understand what AI can and can’t do. You have to get your hands dirty. Experimentation leads to realization, which leads to successful implementation.
In some ways, our job as leaders has never been harder. But it’s also never been more exciting. We have a chance to rethink how we work and how we can make our employees’ jobs more impactful and more meaningful. Doing that will change the trajectory of your business.
Watch the Executive Exchange with Ethan Mollick to see how leading teams are governing and scaling AI workflows across the enterprise.