Why Most Companies Are Getting AI Wrong

I Built, Sold, and Failed at This

Aerospace engineering at ITA, computer science at UT Austin, built and sold a company, led technology at scale. I've been both sides of the table. I didn't study this. I lived it.

What I see now is what I lived then: companies wasting billions on AI because they make the same mistake I made — and only learned to avoid after it cost millions.

This is operator perspective, not consultant theory. The difference matters.

Why Lack of Resources Is Not the Real Problem

Talk to any frustrated executive and he cites symptoms: data not ready, no technical talent, insufficient budget, cultural resistance. All real. None are the cause.

The cause is simple. Most companies don't know what they want AI to do. Not at the strategic level that matters. They know they need it — market says so loudly. Press a little and ask "where exactly does this change your competitive position?" The answer vanishes.

What's left: process automation, cost reduction, perpetual pilots. None of these is the strategic lever.

What Are the Traps That Destroy AI Value?

Trap 1: Why Operational Efficiency Isn't Competitive Advantage

The first is seductive because it feels responsible. Company identifies a cost (customer service, contracts, reports), implements AI, pilot works, ROI is on paper. Approved.

Six months later, the savings exist. The competitive advantage doesn't. Reducing cost of something competitors also reduce isn't advantage. It's maintaining position.

In every company I ran, optimizing what already existed was always tempting. Real value came from identifying where the rules were changing — and acting first. It's not "how do we do this cheaper." It's "what can we now do that was previously impossible."

Things that cost $10 million in human teams now cost $50,000. That's qualitative, not quantitative. Companies that understand this don't cut costs. They build capabilities that didn't exist.

Trap 2: How the Tool Choice Sabotages Strategy

Consultancy, RFP, vendor evaluation. At the end, the chosen tool shapes what seems possible. Strategy is derived from the tool, not the reverse. Like building a company around software you managed to buy.

The company I sold was built differently: real problem first, specific solution for that problem, technology second. The acquirer didn't buy just the product — they bought that clarity. Today most companies do the opposite: buy generic capabilities, force them into real problems, create pilots that work in labs and die in practice.

Right sequence is simple: where your advantage shifts fundamentally → which capability creates that shift → which tool delivers it. Most companies start at step 3.

Trap 3: How Perpetual Pilots Destroy Value Instead of Creating It

I know a well-structured company with 14 simultaneous AI pilots. None in production. Each generates convincing reports. None changed the business. Not transformation. Risk management disguised as innovation.

Pilots exist to learn, not to avoid decision. Difference between transformers and perpetuators is clear: transformers treat pilots as experiments with explicit go/no-go criteria.

Most expensive mistake I made wasn't investing in tech that failed. It was not killing projects fast enough. Indecision costs. In a closing window, that cost is exponential.

What Questions Should Every Leader Answer Before Investing in AI?

After getting these decisions wrong and right, I developed three questions every leader should answer. They didn't come from consulting. They came from operations.

How Do You Know If AI Really Changes the Game in Your Industry?

Fundamental difference between AI that improves what you do and AI that changes what's possible. Improving is necessary, not differentiating. When every competitor improves by same proportion with same tools, net result is zero.

Changing the game is when AI lets you offer something competitors can't — not because you have more money, but because you saw application they didn't. Simple test: competitor buys your same product tomorrow, does your advantage disappear in 12 months? If yes, it's cost of parity, not advantage. Don't abandon it — don't build strategy on it.

Who Should Own the AI Decision in Your Company?

If your AI strategy is led by technical team, you have governance problem — not because technologists are incompetent. Decisions that matter in AI aren't technical. They're business, product, revenue, positioning. "Where does AI change our market position" isn't engineering question. It's CEO question.

Pattern that worked in scaling companies: strong technical leadership for execution, strong business leadership for strategy, CEO who understood AI enough to bridge. Without that, strategy and execution live in parallel universes.

What's the Real Cost of Waiting 6 Months to Decide?

Most underestimate because cost isn't immediate — it compounds. Wait 6 months and you're not just delaying. You're letting competitors accumulate 6 months of data, learning, refinement. When models double in capability yearly, 6 months isn't 6 months. It's exponentially more.

Cost of waiting = missed opportunity + competitor advantage + rising entry cost. All three grow. Third grows nonlinearly. Window closes — not because AI disappears. Because strategic asset isn't tech access. It's accumulated learning, proprietary data, organizational capability. Those can't be purchased when you decide "it's time."

What Actions Should You Take Now Before the Window Closes?

Not saying pivot everything to AI now. Saying strategic conversation needs to happen at right level.

First: map where your advantage amplifies with AI and where it's destroyed with AI in competitor's hands. Two different analyses, both urgent.

Second: move conversation to C-suite — not to present pilot. To make decisions about where to invest, divest, and allocate scarcest resource: leadership attention.

Third: choose one big bet. Not 14 pilots. One bet where AI fundamentally changes how you compete. Serious resources. Clear success criteria. Decision in 90 days — positive or negative.

Market won't wait for you to find the right moment.

Why Waiting for Consensus Is the Worst AI Strategy

Built and sold companies. What I learned isn't in analyst reports: markets don't wait for consensus. In every acquisition I've been part of — on both sides — value wasn't just product. It was conviction the problem would grow, plus execution proving we could solve it. Timing was the thesis.

Same principle with AI. Companies that lead next decade aren't the ones that waited for consensus to form. They're the ones with clarity to act first and capability to execute.

Technology available. Data accessible. What's missing isn't resources. It's clarity about where AI actually transforms business. Clarity doesn't come from consultant or report. Comes from leaders who understand their own business deeply enough to ask right questions.

That's what Zerlotti exists to build.

If you're reading this, you're making decisions that matter about AI. My inbox is open for conversations that actually shift strategy.

Frequently Asked Questions

Why do most corporate AI projects fail?
Because there's no strategic clarity about where AI shifts competitive position. Companies delegate the decision to technical teams, create pilots that work in labs and die in practice. The root problem is leadership, not technology.

What's the difference between AI that improves and AI that changes the game?
AI that improves reduces costs competitors also reduce — zero net advantage. AI that changes the game creates previously impossible capabilities. When every company improves by the same proportion with the same tools, the net result is zero.

How long is the window to build real AI advantage?
The window closes not because AI disappears, but because the strategic asset is accumulated learning. Companies that decide in the next 18 months gain compounding advantage. Those that wait compete with permanent structural disadvantage.

Who should lead AI strategy in a company?
The decision belongs to the CEO, not the CTO. Questions like "where does AI change our competitive position" are business, product, and revenue decisions — not engineering. No significant corporate transformation in history has happened bottom-up.

How do I know if my company is caught in the perpetual pilot trap?
If there are more than three simultaneous AI pilots without explicit go/no-go criteria, that's a signal. Pilots exist to learn, not to avoid decisions. Transformers treat each experiment as a bet with a clear deadline and success or shutdown criteria.