Stuck in AI pilot purgatory? The real win is focusing on business problems, not hype, and scaling smartly with the right mix of tools.
Many of the leaders I talk to are struggling to move past the pilot stage of AI. They are in pilot purgatory. It’s not a lack of ambition that holds them back — it’s a lack of clarity. Questions about where to build, where to buy and how to structure teams for scale create a kind of organizational gridlock. I’ve faced the same challenges in my own experience leading digital transformation efforts, and I’ve found that what breaks the cycle is not hype or vision, but a practical, focused approach.
AI doesn’t need to be complicated to be effective. But it does require structure, prioritization and a willingness to rethink how work gets done.
Here’s how I think about it.
Buying versus building: Start with a philosophy, not a spreadsheet
Buying versus building: Start with a philosophy, not a spreadsheet
Early on, I heard IT leaders in my network trying to justify AI investments by comparing cost and performance between vendor platforms and in-house models. But I quickly realized that no spreadsheet could capture the broader trade-offs at play. Speed matters. So does the ability to scale, maintain and continuously improve what you build.
Over time, I settled on a general rule of thumb: aim to buy about 80% of the capabilities you need and build the remaining 20%. Some organizations may shift closer to 70/30, but the underlying idea holds: don’t reinvent what’s already working elsewhere.
Evidence backs this up. In a July 2025 Morgan Stanley AI Adopter Survey, analysts reported that AI is already delivering tangible returns across industries such as financial services, consumer goods and real estate — many of which are benefiting from off-the-shelf tools with embedded intelligence. That market reality is mirrored in the vendor landscape: enterprise software platforms are evolving at a remarkable pace. Vendors are embedding significant AI capabilities directly into their products. Gartner predicts that by 2026, more than 80% of software vendors will have embedded generative AI into enterprise applications, up from just 1% in February 2024.
That’s why I’ve found it far more effective to treat enterprise software platforms as the foundation and then identify where custom development is truly needed. This frees up internal teams to focus on high-value use cases rather than duplicating what’s already available.
Of course, out-of-the-box solutions don’t cover every scenario. In some cases — like stitching together data and workflows across platforms — agentic AI or internal development has been essential to close integration gaps. It’s not about choosing one approach over the other; it’s about knowing when each one makes sense.
Prioritizing problems, not projects
AI is not the strategy. The business is the strategy. That’s a lesson I learned after watching too many proofs of concept fail to scale because they didn’t tie directly to outcomes that mattered.
To make progress, it’s best to shift the conversation away from AI as a technology and toward AI as a tool for solving real business problems. That starts with listening — really listening — to what customers, frontline employees and partners are experiencing.
I attended the CIO 100 event in August, where Avery Dennison received an award for Program Loop, our AI program. While there, I had the chance to compare notes with other IT leaders about how they were approaching AI opportunity selection. One common theme was the use of workshops with functional leaders to surface pain points from the customer’s perspective. From there, several leaders described bringing cross-functional teams together to weigh potential solutions.
What struck me most was the simplicity of the prioritization methods. One approach rated opportunities across three dimensions: business impact, level of effort and urgency — each on a simple high, medium or low scale. When you map those ratings, the opportunities that score high in impact and urgency but low in effort naturally rise to the top. Those become the logical first focus areas, and the clarity it provides helps build early momentum.
What made this approach effective was how it aligned teams around what truly mattered. Several leaders described reframing challenges as “How might we…” questions, which naturally broadened the discussion and sparked more creative solutions. AI was then introduced not as the starting point, but as one of many possible enablers.
From my perspective, this mindset shift changes the tone of the entire conversation. It’s no longer a debate about whether to adopt AI — it becomes a dialogue about driving business value. That’s where real momentum begins to build.
From central authority to shared ownership
In conversations with other IT leaders, I’ve noticed a common pattern in how AI programs evolve. Most began with a centralized team — a logical first step to establish standards, consistency and a safe space for early experiments. But over time, it became clear that no central group could keep pace with every business request or understand each domain deeply enough to deliver the best solutions.
Many organizations have since shifted toward a hub-and-spoke model. The hub — often an AI center of excellence — takes responsibility for governance, education, best practices and the technically complex use cases. The spokes, led by product or functional teams, experiment with AI features embedded in the tools they use every day. Because they’re closer to the business, these teams can test, iterate and deliver solutions at speed.
When I look across industries, the majority of AI innovation is now happening at the edge, not the center. That’s largely because so much intelligence is already embedded into enterprise software. A CRM platform, for instance, might now offer AI-based lead scoring or predictive churn models — capabilities a team can enable and deploy with little to no involvement from the center of excellence.
According to McKinsey, “Over the next three years, 92 percent of companies plan to increase their AI investments. But while nearly all companies are investing in AI, only 1 percent of leaders call their companies ‘mature’ on the deployment spectrum, meaning that AI is fully integrated into workflows and drives substantial business outcomes. The big question is how business leaders can deploy capital and steer their organizations closer to AI maturity.” But the real insight is not the use of AI itself — it’s the decentralization of how it gets done.
This doesn’t mean the center becomes irrelevant. It means the center becomes an enabler. Its job is to create the conditions for responsible scale, not to own every initiative. That’s a subtle but important shift in how leadership needs to think about structure.
When intelligence becomes invisible
Some of the most powerful AI features are also the quietest. That’s something I’ve come to appreciate as adoption has matured. Early on, AI was treated as a distinct effort — something big and special. Today, the most valuable capabilities are often the ones that feel invisible. They just work.
Consider a finance process like accounts payable. In the past, automation meant scheduling payments strictly on invoice due dates. Now, embedded intelligence can recommend optimal timing based on supplier behavior, working capital goals and historical cash flow patterns. There’s no project name attached to it — just business value being delivered.
This shift — from AI as an initiative to AI as part of the foundation — is one of the most exciting developments I’ve seen. It doesn’t always require data scientists or specialized teams. It calls for product managers and business owners who understand the process and the outcomes they want to improve.
As more enterprise tools embed intelligent features, the role of IT leaders is increasingly about helping teams spot what’s already available, test it safely and scale what works.
Measuring the value of AI
One of the biggest hurdles I hear from leaders is how to measure the value of AI. The truth is, you don’t need a brand-new framework. Most organizations already track ROI, EBIT or productivity gains from the projects their IT teams deliver. The key is simply connecting those outcomes back to the fact that many of those projects are now powered by AI.
Take enterprise software as an example. Modern SaaS platforms — whether CRM, ERP, HR or supply chain — are increasingly embedded with AI features. In fact, most new systems already include intelligent capabilities as part of their core functionality. That means when IT delivers a new SaaS-based pricing system with an estimated ROI of 25% and $6 million in EBIT, the leader doesn’t need to invent a separate “AI metric.” All they need to recognize is that the $6M of EBIT is profitability driven in part by AI. There is no need to overthink it because in a few years, we will not be able to distinguish AI from modern software.
There’s another benefit: the stories about AI become much easier to tell. You no longer have to lean on anecdotes about experimental pilots running in some distant site. AI is already present in many of the enterprise solutions you’re deploying. That means you can add to the pure economic value of the new CRM system and also share stories about how it’s helping the commercial team work more efficiently — or even improving customer NPS.
Of course, not all value will come from enterprise software. There are also opportunities to monetize AI-first solutions — such as digital avatars, copilots and industry-specific applications — as well as custom-built models that solve unique challenges. These, too, should be captured and reported as part of the overall AI value story.
The main point is this: you don’t have to overengineer the measurement process. Simply ride the coattails of existing projects that are already leveraging AI in some way. By doing so, you give stakeholders both the numbers and the narratives they need to see AI as a driver of progress, not just an experiment.
Final thoughts
AI is not optional. But it’s also not magic. Turning AI into an advantage requires a clear sense of purpose, a willingness to let go of central control and the discipline to focus on what really matters.
I’ve learned that success has less to do with how advanced the technology is and more to do with how well it’s aligned with business needs. That means starting with the problem, picking the right approach and building a structure that allows teams to learn and adapt as they go.
It’s not about pilots. It’s about progress. And that starts with being honest about where you are, and bold and responsible enough to move forward.
This article is published as part of the Foundry Expert Contributor Network.
Want to join?
