The Smallest Seat on the Plane, the Biggest Mistake in AI Transformation

What does an airline seat have to do with enterprise AI?

What if the most important question about generative AI is not how fast employees are experimenting, but who gets to occupy the scarce, high-leverage space in an organization? That sounds like a travel question, until you notice the strange logic that connects it to corporate transformation. In aviation, a business class seat uses far more space and produces far more emissions than economy. In organizations, a few premium behaviors also consume disproportionate resources: executive attention, management time, capital, and political bandwidth. The result is the same in both cases. A system can look efficient from the outside while quietly becoming much more expensive than it needs to be.

The real lesson is not about flying. It is about allocation under constraint. Climate-friendly travel asks us to stop treating comfort as invisible and start seeing the footprint of our choices. Gen AI asks companies to do something similar: stop treating adoption as a collection of individual enthusiasms and start seeing the organizational footprint of how work is actually redesigned. The danger in both cases is the same too: we mistake personal behavior for systemic change.

---

The hidden footprint of privilege

Business class is not just a more comfortable seat. It is a bigger claim on shared space. That is why the emissions are so much higher. The same structural pattern appears in companies when AI is introduced as a perk for a few power users rather than a redesign of the work itself. A handful of employees become impressive experimenters, but the organization as a whole remains unchanged, and therefore leaves most of its value on the table.

This is the central tension in gen AI adoption. Employees are already using the tools, often enthusiastically, often ahead of leadership. Yet organizational maturity lags badly. That gap is not a minor implementation issue. It is the equivalent of installing a more fuel efficient engine while leaving the aircraft design untouched. You may improve a few parts of the journey, but you do not change the economics of flight.

Here is the deeper pattern: the benefits of a transformative technology are captured only when the surrounding system is redesigned to match its logic. Gen AI is broadly accessible, which tempts leaders to think access alone will drive value. But access is not transformation. A widely available tool can either become a productivity layer spread across the organization or a novelty that burns time, creates scattered use cases, and never compounds into real advantage.

The largest gains do not come from giving everyone the same tool. They come from changing the system so the tool changes how the system works.

That is why the most useful question is not, “Who is using gen AI?” It is, “What part of the organization has been redesigned because gen AI exists?”

---

Why experimentation does not equal transformation

Most companies begin the same way: people try prompts, draft emails faster, summarize documents, and prototype ideas. This stage matters. It lowers fear, builds fluency, and creates demand from below. But experimentation can become a trap if leaders mistake excitement for impact. It is like letting travelers choose their own offset scheme and calling the trip sustainable. The behavior feels responsible, but the underlying system remains the same.

The deeper issue is that individual productivity gains do not automatically become organizational value. If one manager writes status updates 20 percent faster, the company has not yet transformed. If ten people save time but nobody reinvests that time into better customer service, sharper strategy, or faster product cycles, the firm merely creates pockets of spare capacity that evaporate into busyness.

This is why the idea of a business-led center of excellence matters. Not because every transformation needs another committee, but because organizations need a mechanism that can answer three hard questions:

1. Which use cases matter most?
2. Which should be scaled, and which should be stopped?
3. How do gains in one area change the way adjacent functions work?

Without that discipline, AI adoption remains fragmented. With it, the company can move from isolated use to domain-level redesign. That distinction is crucial. Domains like customer service, product development, and marketing are where work actually happens across boundaries. They are the practical unit of transformation, because they reveal the interdependencies that point solutions ignore.

Think of it this way: a company does not become healthier because one department buys a gym membership. It becomes healthier when the daily routines of the whole organism change. AI works the same way.

---

The real constraint is not technology, it is operating model gravity

The most revealing idea in all of this is that technology adoption is usually described as a software problem when it is actually an operating model problem. Tools are easy to demo. Workflows are hard to change. A model can draft a memo in seconds, but if the approval chain still requires four meetings, the memo is not the bottleneck. The bottleneck is the organization’s gravity: its habits, incentives, reporting structures, and assumptions about who does what.

This is where the climate analogy becomes unexpectedly useful. Flying economy reduces emissions not because the traveler becomes morally superior, but because the same flight carries more people in the same aircraft footprint. The system becomes more efficient through better density. In companies, gen AI should be used to increase the density of value creation: fewer handoffs, faster iteration, sharper decisions, more time spent on judgment and relationships, less on repetitive administration.

That is why one of the most promising applications is in management itself. If gen AI can surface coaching prompts, prepare performance notes, or draft routine communications, managers can spend more time on the work that actually cannot be automated: developing people, resolving conflict, and making tradeoffs under uncertainty. In other words, AI should not make management more mechanical. It should make management more human by stripping away the mechanical parts.

The goal is not to automate people out of the picture. The goal is to reclaim human attention for the work that only humans can do well.

This also explains why talent strategy matters so much. A transformation is not a hiring spree. It is a recalibration of skills. Employees need prompt writing, contextualization, and data-driven decision making, but also judgment, strategic thinking, and social intelligence. Tech talent must learn to translate business needs into technical solutions. Leaders must learn to use AI safely and visibly. HR must become an engine of capability building, not merely a support function.

The organizational challenge is therefore double. First, companies must redesign processes. Second, they must redesign the people system that makes those processes durable.

---

The 5 to 1 rule: why people matter more than software

There is a memorable ratio that captures the economics of transformation: for every $1 spent on technology, $5 should be spent on people. Whether or not every company can literally apply that ratio, the principle is powerful. Most technology investments fail not because the tool is bad, but because the organization underinvests in the behavioral and skill changes required to realize the tool’s value.

This is the part leaders resist most. It is tempting to believe that once a model is available, adoption should follow naturally. But human systems are not plug and play. They need role modeling, communication, training, incentives, and reinforcement. If leaders do not visibly use the tools themselves, employees get the message that AI is optional theater. If performance metrics do not reflect new ways of working, old habits remain rational. If training is generic, adoption stays shallow.

The deeper insight is that AI transformation is really a trust transformation. People must trust that the tools are useful, that the risks are managed, and that the company is serious about changing how work gets done. That requires governance, yes, but also emotional credibility. Employees need to believe the organization is not merely extracting efficiency from them, but creating a better way to work.

A practical way to see this is to compare two companies:

• Company A gives everyone access to a model and says, “Experiment.”
• Company B identifies the top three domains where AI can change cycle time, retrains the relevant teams, updates performance metrics, and publishes measured outcomes.

Company A gets anecdotes. Company B gets compounding advantage.

The same is true in climate-conscious travel. Telling individuals to care more about emissions is useful, but changing defaults, norms, and system design is what creates durable impact. Organizations should think this way about AI too: not as a collection of heroic individual choices, but as a redesign of defaults.

---

A useful mental model: density, not just speed

Most AI discussions obsess over speed. Faster writing. Faster coding. Faster analysis. Speed matters, but it is not the right north star. The better metric is density of high-value work.

Density means that more of an employee’s day is spent on work that requires judgment, creativity, empathy, or cross-functional coordination. It means managers spend less time on administration and more time on people development. It means customers encounter faster, smarter service. It means product teams spend less time wrangling information and more time making decisions.

This is the same logic as efficient seating on a plane. The point is not that economy is virtuous in the abstract. The point is that the same physical system can carry more value with less waste when resources are allocated differently. In business, AI should help organizations carry more strategic value per unit of attention, not simply produce more output per hour.

That framing also clarifies what success looks like. A company does not need every employee to become an AI power user. It needs the right domains to be redesigned so that AI meaningfully changes throughput, quality, and adaptability. Some use cases should be scaled. Others should be stopped. The deciding factor is not novelty. It is whether the change alters the system’s density of value.

When leaders adopt this lens, they ask better questions:

• Where are humans still doing low-value coordination work?
• Where are approvals, drafts, and reviews creating unnecessary friction?
• Which roles need augmentation rather than replacement?
• Where will AI free people to do more of the work that creates trust and differentiation?

These are not software questions. They are design questions.

---

Key Takeaways

1. Treat AI adoption as a redesign problem, not a usage problem. If work flows stay the same, the tool will only create isolated productivity pockets.
2. Focus on domains, not scattered use cases. Customer service, product development, marketing, and performance management are where AI can reshape the operating model.
3. Invest in people at least as much as technology. Training, governance, role modeling, and updated incentives determine whether adoption compounds.
4. Measure density of value, not just speed. Ask whether AI is increasing the share of time spent on judgment, creativity, and human work.
5. Use governance to scale what works and stop what does not. A center of excellence should not only encourage experimentation, but also decide what deserves scale.

---

The future belongs to organizations that learn to fit more value into less waste

The common mistake in both travel and AI is to focus on the surface choice and ignore the system behind it. It is easy to tell people to choose economy or to experiment with prompts. It is much harder to redesign the machinery that makes those choices matter. But that is where durable change lives.

The most effective organizations will not be the ones that merely let employees play with gen AI. They will be the ones that use it to rethink how work is organized, how talent is developed, how managers spend time, and how value moves through the company. In that sense, AI is less like a tool and more like a stress test. It reveals whether the organization is built for habits or for adaptation.

And perhaps that is the final connection. Climate-friendly travel asks us to recognize that every seat choice participates in a larger system. Gen AI asks the same of every workflow choice. The real question is not whether individuals can act wisely. It is whether institutions can learn to make wisdom scalable.

The companies that answer that question well will not simply use AI better. They will become more like the kind of system the future rewards: denser, faster, more human, and less wasteful all at once.