Almost every customer conversation we have these days starts with the same story. The C-suite told the organization to adopt AI. So they bought AI tools and told their teams to use them. Most of those tools are priced on usage, measured in tokens. The more people use them, the more tokens they consume, and the bigger the bill gets.
Now, a year later, the bills are coming due. Finance is seeing AI spend show up in budgets across the organization, and that spend is getting the attention of the same executives who mandated adoption in the first place. Only this time, they’re asking a different question: What are we actually getting for this investment?
That question changes the goal. It’s no longer about using more AI. It’s about integrating AI into the business in a way that is safe, sustainable, and delivers measurable value. Then operationalizing that process so adoption accelerates instead of stalling.
In my previous blog, I laid out five factors that separate AI projects that ship from those that stall:
- Start with a use case, not a science experiment
- Put decision-makers in the room from day one
- Get to a proof of concept fast, and be willing to kill it (fail fast)
- Fix the use case data first, or don’t start
- Coordinate everything. The model isn’t the hard part
That last one is what I want to expand on here, because it’s where most organizations are struggling right now.
Long-term Success Rests on Coordination, Not Technology
Handing everyone an AI license and turning them loose is not a strategy. But neither is locking AI down, mandating a single tool, or writing policies so restrictive that nobody bothers to try.
Success comes down to how your organization approaches use case development and governance. And that is best handled by a cross-functional group that owns it.
Call it an AI Committee. Or an AI Adoption Committee. Or an AI Center of Excellence (CoE). The name doesn’t matter. The function does.
This group should include line of business leaders and key stakeholders who will shape and oversee five critical functions as adoption grows: business alignment, policies and governance, use case intake, tooling selection, and AI lifecycle management.
Business Alignment
AI has the power to change how work gets done. The biggest wins don’t come from doing existing tasks a little faster. They come from doing things that were previously impossible. But what that looks like inside your organization will be unique to your business.
As employees bring use case ideas forward, every proposal should map to your overall business strategy. Above all else, AI has to move your business closer to its goals and objectives. If it doesn’t, there’s no reason to use it. Each use case also needs a clear ROI, so the investment required to move it into production makes sense.
The AI CoE owns that alignment.
Policies and Governance
AI is changing work whether organizations are ready or not.
Many companies learned early on that once employees got the green light to use AI, they started building it into their daily work faster than the company could redesign its governance, data, and operating models.
What happened next? Out of fear, many companies rushed to put policies in place to restrict AI. The problem? In their haste, they skipped the first and most important step: establishing guiding principles.
Policies without principles are all about prevention. They tell people what they can’t do without explaining why. That doesn’t work very well. All it does is create friction and push people toward shadow AI.
Principles must come before policies. Principles define what you're trying to achieve. Policies define how you achieve it safely.
In other words, principles tell people how to say yes. Policies make sure they say yes safely.
Use Case Intake
Good ideas for using AI can come from anywhere in your organization. Line workers. Field technicians. Warehouse managers. Salespeople. Back-office administrators. The list goes on.
Many of these people already use AI in their personal lives. They see the benefits firsthand. And they want the same capabilities at work.
So how do you capture those ideas and evaluate them?
Through a formal use case intake process. It could be a simple form, a ticketing app, or something else. The format matters less than having one. Make it available to everyone, and ask for specific information: a detailed description, the scope, the data involved, and the expected ROI.
Keep the ROI criteria narrow but meaningful. Require every proposal to explain how the use case will increase revenue, lower cost, mitigate risk, or improve customer satisfaction.
An intake process like this does two things. First, it gives the AI CoE the information it needs to evaluate every proposal consistently. Second, it drives employee engagement, which is a key ingredient for adoption.
When employees see that they have a voice and that their ideas get a fair hearing, they’re more likely to get on board when a new use case is deployed. It also gets them thinking creatively about what else AI could do for their work.
Tooling Selection
Getting the outcomes you want often comes down to having the right tools. But with AI, there is no one-size-fits-all solution.
The right tool depends on the use case. Once you have a well-defined use case with its data sources identified and outcomes specified, you can select the tooling to deliver it.
Picking a tool first and then trying to back into an outcome rarely works. It encourages action without direction. It’s also how a lot of organizations ended up with the AI bills I described at the start.
What can be worse is to blanket restrict all but some tools. Instead, adopt guiding principles like “We want to avoid vendor lock-in” and then employ policies that support safely adopting those principles. In this way you avoid forcing users into shadow AI. You now position yourself to saying “yes” instead of saying “no” to your community with AI adoption.
AI Lifecycle Management
Infrastructure and software both have limited lifespans. That’s why we manage them across their lifecycle, refreshing, replacing, or retiring them when they become obsolete.
AI should be treated the same way.
That requires a defined process for AI lifecycle management. Once a use case moves past ideation and into production, it needs to be reviewed on a regular schedule.
Two reviews are key:
- Value and business alignment review. Is the use case still aligned with your business objectives, and is it still delivering the value it promised? If not, it should be changed or retired.
- Relevance review. Is the use case still relevant to how your company operates? Things change. Priorities shift, processes evolve, and the technology improves. What makes sense today may not be necessary 18 months from now.
Final Thoughts
We often talk about AI as a journey, getting from point A to point B and on from there. That framing makes sense, up to a point.
But AI adoption is less a journey with a destination and more an ongoing, iterative process. Yes, you need to build a solid foundation first. After that, it becomes a repeatable cycle: decide what your company wants to accomplish, determine how AI can help, execute within a proven framework, and measure the results. Then do it again. That cycle is exactly what an AI CoE is built to run, and it’s how you have a real answer ready the next time leadership asks what your AI investment is delivering.
In my next blog, I’ll dig deeper into why AI principles need to come before policies, and how to establish them. Keep an eye out for it.
If you want to talk through your specific business goals and how AI can help you reach them, get in touch with us. We’d love to chat.