The goal should be to make everyone a builder.
The Fork in the Road
In my last post, I mapped out the AI adoption continuum and argued that the gap between companies seeing massive gains and the ones seeing nothing comes down to people, not procurement. But there's a question upstream of that, one most companies skip right past: what does "AI transformation" actually mean?
There are two versions of the answer, and they lead to very different places.
Version One: Productivity. Faster. Leaner. More with less. Do the same things, but cheaper. This is the L0–L2 version from the continuum, which is tools-based, procurement-driven, and capped at incremental improvement. It's not wrong. But it's a cost optimization play dressed up as transformation.
Version Two: People. Not "How do we speed up what we already do?" but "what would we build if we started today?" And when you take that question seriously, you land in a place that makes a lot of executives uncomfortable: everyone is a builder. The finance person. The PM. The sales rep. The ops lead. Every single one of them, with the right skills and access, can build things that used to require an engineering ticket.
Most companies pick version one, not because they evaluated both options and chose, but because they never realized there was a choice.
The Default Path: Procurement Theater
If you don't have a clear vision of where AI transformation is supposed to end up, there's a natural, almost gravitational response: you buy the tools, give them to your best and brightest engineers, and let them run. You host some brown bags. You pull together some demos. You shield everyone else from the hard truths...you know, the questions about roles changing, processes dying, or ahem, what happens to a specific role / team. So you alleviate their fears by keeping the change contained. Manageable. Technical. Compliant.
And the technical teams that do have access? They're busy. They're building things, like indexing tools, internal integrations, proof-of-concept demos. A long and impressive laundry list of initiatives without an overarching theme nor vision. But ask anyone to explain in business terms what they've accomplished and why it matters, and the room gets quiet. The work is technically sophisticated, but directionally unclear.
That's Procurement Theater, when you end up engineering the appearance of transformation without creating the conditions for it. And it's not a failure of execution. It's the natural default when nobody has defined what they're transforming into. You gave people a Ferrari and no map, so they're pulling donuts in the parking lot.
Procurement Theater shows up in three paradoxes, where each is a version of the same disease: saying yes at the strategic level and no at the operating level.
Paradox 1: "Be AI-native, but watch the tokens."
Companies set aggressive AI transformation goals and then manage the tools like they're expensing office supplies. Token budgets. Per-seat cost scrutiny. Usage dashboards that flag "heavy users" instead of celebrating them. The person using AI the most, who is actually transforming their workflow gets a tap on the shoulder about cost.
Meanwhile, the person who logged in once and never came back? Access revoked. "They're not using it, so let's save the seat fee." Instead of asking why they're not using it, instead of helping them come up the curve, the company claws back the seat and calls it fiscal responsibility.
So you punish the power users for costing too much and abandon the low users for not adopting fast enough. Who are you actually "transforming"???
Paradox 2: "Experiment freely, but don't touch anything."
The tools get provisioned, but the environment doesn't change. Every system, be it CI/CD, code review, deployment, access control, whatever is designed for professional engineers shipping production software. A PM who writes API specs isn't allowed to write code. A designer can't prototype beyond a static screen. Building is for engineers; everyone else writes documents about what they want built.
So when leadership says "experiment with AI," the only people who can experiment are the ones who already had access. Everyone else gets chat. And chat is fine for summarizing a doc or drafting an email, but it's not transformation. The highest-value use cases, such as the ones that actually rethink workflows, require access that nobody's approved and nobody's willing to approve.
Self-fulfilling prophecy. You restricted the tool to low-value tasks and then used the low-value results as evidence that the tool has low value.
Paradox 3: "Transform everything, but change nothing."
The hardest version. The company says it wants AI-native operations. But the org chart doesn't change. The processes don't change. The way teams are staffed, measured, and incentivized. None of it moves. AI gets layered on top of the existing system. It's an addition, not a transformation.
And meanwhile, the really big procurement decisions keep rolling in non-engineering teams like HR, legal, and finance, where multi-quarter SaaS evaluations, lead to multi-year contracts. Nobody's asking the harder question: does this tool even need to exist in two years? By not even asking, you're locking in pre-AI architecture for a post-AI world, thereby unknowingly enshrining obsolesence.
This is Feedback Theater all over again, which is the performance of the thing replacing the substance of the thing. Feedback Theater is review culture without honesty. Procurement Theater is AI adoption without transformation.
The Builder Path
Choosing the people path is harder. It requires swallowing a pill most organizations aren't ready for: none of the shit you built is going to survive. Not the processes, not the financial model, not the PERF process, the tooling assumptions, nothing. It's ALL getting rethought, whether you're ready or not.
But if you start from the belief that everyone is a builder...and organize around making that true. The to-do list looks nothing like a software purchasing exercise. It looks like a company redesign.
Education changes. Not "here's a 30-minute webinar on prompt engineering." Real investment in developing builder capacity across the entire org, such as prototyping, working with agents, understanding what's possible.
Tooling changes. The next wave isn't 500 engineers shipping 50 services. It's 5,000 people across every function shipping apps, automations, and internal tools. You need repos, sandboxes, lightweight deployment paths, and permission structures designed for a fundamentally different profile of creator.
Access changes. People need to connect AI to the systems where the actual work happens, like the codebase, the data warehouse, the internal APIs. Real tools, real access, real ability to build and ship, with the right guardrails, not the guardrails designed to keep non-engineers out.
Measurement changes. Stop counting active seats, token usage, and lines of code. Start counting what processes got eliminated, what decisions got faster, what shipped that wouldn't have shipped. Otherwise you see tokens as "costs", and neither as signal, nor investment.
And you kill things. The clearest sign of real AI transformation isn't what you bought, it's what you stopped doing. What procurement did you cancel? What manual process did you eliminate? What meeting doesn't happen anymore? If the answer is "nothing," you haven't transformed. You've accessorized.
This is what AI-First Thinking looks like at the org level. If you keep asking "what would we build if we started today?" and the answer keeps being "exactly what we have now, but with an AI layer on top", then you're not asking it hard enough.
Where Are You?
There's a simple way to tell which side of the fork you're on. Ask yourself: are you focused on tools, or are you building builders?
If your AI strategy lives solely in engineering & IT, you're probably seeing it as a tools issue.
If you see your people as those with limitless potential, regardless of function or seniority, then enable them and get out of the way. You're building builders. And if that scares the crap out of you, we've got to get that Ferrari driving to the next frontier.
The tool isn't the transformation. Building builders is.
