Your AI Problem is Not an AI Problem: Why Every New Technology Fails the Same Way
Remember when Agile was the newest, most need-to-have tech methodology? Then it was Cloud migrations. Now it’s AI adoption.
New technology isn’t ever really new, and yet I’m still watching many organizations make the same integration mistakes that they were making decades ago: Shoot first, aim second.
Before, there was a larger margin for error in implementing anything new. But the implications of AI have a much wider shock radius within companies than any other technology (perhaps except for the emergence of personal computing in the 80s).
With Agile, you could get away with it (it’s called a “practice” for a reason, so you can get better at it over time!). With Cloud, you could mostly get away with it; it was expensive, but you could consolidate the infrastructure quickly. With AI, the bill is coming due on arrival. Almost literally.
You can’t roll out a tool before examining the structures holding everything up, then declare a migration complete and wait to see what happens. When you aren’t seeing the ROI you hoped for, this is usually why.
The guidance that follows is your way out of this pattern.
There Are No More Toasters
Let’s turn back the clock for a moment. Think about how your organization approached Cloud migrations. Did the model change before the technology arrived, or did the technology arrive, and then everyone crossed their fingers, said a little prayer, and hoped the details would work themselves out?
Most orgs started moving servers to the Cloud, saw efficiencies and cost savings, and kept going. Then came the cost overruns, then the corrections, and then the model adjusted. Now, cloud is assumed, and the governance is in place to protect orgs from significant risk.
It has become increasingly clear that AI is too complex and the stakes are too high to simply integrate a new tool ASAP, then deal with the shock waves later.
Enterprise technology used to behave like a toaster. You bought it. You plugged it in. It did its job in a defined corner of your operation and largely stayed there. You didn't have to think about it very hard. The edges were clear, the maintenance was manageable, and the blast radius when something went wrong was contained. (Or, the risk to contain it after the initial blast was low enough that you could worry about it later.)
What you're bringing into your organization today is not a toaster. It's a puppy.
And this sweet, well-meaning puppy runs around and has accidents. It chews things up. It grows. And when it grows, what its needs change completely. It gets its claws into your workflows, your data, your decisions, and your customer interactions. It does not stay in its corner. And it is your responsibility to raise it. You don't plug it in and walk away. You don't get to migrate and call it done.
AI is the most extreme version of this yet. But the dynamic isn't new. Organizations struggling with AI right now are largely the ones that treated the Cloud as a destination, and before that treated Agile as a methodology you implement rather than a way of working you build into your operating model. Same pattern, different puppy.
Why You Can't Kick the Can This Time
There was a time when you could defer the hard work with less fallout.
Mobile created real disruption, but the feedback loop was slow enough that "we'll sort it out" was a viable coping mechanism for a while. With Cloud, too, we saw plenty of organizations spend years in a half-migrated state, running on a patchwork of old infrastructure and new platforms, and the consequences were manageable enough to keep moving. Not ideal, but survivable.
AI is not offering that grace period.
The workforce implications are immediate and they're not subtle. People at every level are already asking what this means for their roles, their teams, their value. That's not a technology question. That's a people question.
With AI, complexity runs deeper because it doesn't live in one system or serve one function. It integrates and compounds. The more capable it becomes, the more of your organization it reaches. You can't manage it from a distance. You can't treat it like infrastructure you deploy and revisit at the next planning cycle. It is, by design, woven into the work, which means the structure of the work has to be ready for it.
There Are No More Finish Lines
The single most expensive habit in enterprise technology is treating modernization as a milestone. Cloud migration: done. AI rollout: complete. ERP: live. Check the box, move the budget, return to normal operations.
There is no returning to normal operations. There are no more lift-and-land projects. By all means, you should still pop the bubbly and celebrate a milestone, but there’s no end-date to tech integration.
The technology you bring in today will keep growing, keep requiring your attention, keep making new demands of your organization. (Puppies, after all, don’t learn how to let themselves outside even when they grow to adulthood.)
And I see this pattern repeat time and time again because, on the surface, it seems easier to buy a new technology than to transform the operating model underneath it. Implementation has a timeline, but modernization doesn't.
You Can't Slap New Tech Onto Old Structures
This is the part technical leaders sometimes bristle at, which is understandable. But the hard truth is that the structure underneath the technology determines whether the investment actually lands.
AI readiness is not a data problem first. It's not a security problem first. It's an operating model problem first.
That’s why, at Tuckpoint, we always say: People and process first, technology second.
Before another dollar goes toward implementation, I invite you to answer three questions honestly:
What problem are you actually trying to solve with AI? (Not "we want to use AI." This needs to be a specific problem. A defined outcome. Something measurable. If the answer is "we're exploring possibilities," that's fine, but be honest that you're in exploration, not implementation.)
Is your organization structured to solve it? (Do the right people have the authority, the clarity, and the capacity to execute? Are your processes set up to support what you're asking the technology to do, or are you asking a new tool to work around an old structure and hoping it figures out the gaps?)
Are you actually ready to talk about “what” and “how”? (The technology conversation about which model, which vendor, which implementation path, etc., is the last conversation, not the first. What conversations around structure, process, and impact should happen before anything else?)
What Comes Next?
AI is revealing how costly it is to continue repeating the pattern. The toaster days are gone (do I need that on a T-shirt?). The puppy has already moved into your house and, probably, chewed up something important.
To figure out what comes next and minimize the blast radius, you need to take a giant step back and examine the underlying operating model to ensure it’s accelerating your technology investments rather than stalling them.
If you’re not sure where to start, the Tuckpoint Operating Model Maturity Assessment is a great jumping-off point. It’s an evidence-based evaluation of your organization that combines quantitative data, qualitative insights, and an expert review of your operating model artifacts to create a complete, unbiased picture of where your organization is today and what it will take to move forward.
This baseline identifies where bottlenecks originate, how priorities get set, when transformation loses momentum, where teams stall, and where ROI questions go unanswered. The outcome is a clear view of where we should focus our efforts, and it’s unique to your organization. Reach out if to learn more about this process.