I’ve spent the past 20 years building and scaling operations-intensive businesses, including founding Freshly, which was acquired by Nestlé in a deal valued at approximately $1.5 billion, and leading Petfolk, a fast-growing veterinary hospital platform now backed by more than $150 million in capital. Throughout these experiences, one lesson has become increasingly clear. That means as new technology transforms possibilities, organizations need to rethink how they operate to maximize its value.
Executives and boards of directors across all industries are expressing similar concerns. Despite investing billions of dollars in artificial intelligence efforts, many organizations report little tangible benefit. The frustration is real and well-documented. According to the PwC Global CEO Survey: luck As previously covered, 56% of companies say AI has not yet delivered either cost savings or revenue increases, with only about 12% reporting benefits on both fronts.
talk to luck Mohamed Khande, global chairman of PwC, argued at Davos that what is missing is execution rather than AI capabilities, noting that many companies have “forgotten the basics” such as clean data, disciplined processes and governance.
The conclusion many leaders are reaching is that AI is not living up to expectations.
That conclusion is wrong.
The problem isn’t technology. It depends on how leaders seize opportunities and measure success.
Most companies are implementing AI from an efficiency perspective. They ask where they can reduce labor, automate workflows, or achieve faster return on investment within their existing organizational structure. Then evaluate those efforts using software tools and traditional return-on-investment metrics designed for headcount reduction.
This approach misunderstands what AI actually changes.
AI is not just a better way to do the same task. This is a new economic input that collapses the marginal cost of high-quality analytical and intellectual labor. This change has consequences that most organizations are only beginning to understand.
Total human intelligence time as a new unit of work
All major business transformations of the last century followed the same basic pattern. Basic inputs have become dramatically cheaper and usage has expanded exponentially. During the Industrial Revolution, falling energy costs effectively converted machine power into cheap mechanical labor time, allowing machines to augment manual labor on an unprecedented scale. More recently, cloud computing has collapsed the cost of computing, made storage virtually limitless, and made digital distribution universal overnight.
AI now represents the next rotation of the same economic wheel. It brings the marginal cost of quality thinking closer to zero.
To clearly describe this change, it helps to give it a name. I call it Total Human Intelligence Hour (SHIH).
Synthetic Human Intelligence Hour is high-quality analytical and intellectual work produced by AI at near-zero marginal cost and deployable at scale. They are not androids. They are abilities of synthetic intelligence. A new unit of productive effort.
When we look at AI through this lens, the confusion surrounding its adoption begins to make sense. Organizations are trying to force technology that creates synthetic human intelligence time into systems that were designed to have poor human attention spans.
The discrepancy is clearly visible in the data. A research report based on MIT’s 2025 State of AI in Business study. luck also found that only about 5% of integrated AI pilots deliver measurable value, and about 95% show no measurable financial impact. Researchers describe this gap as the “GenAI divide.”
The report goes on to explain that most failures are not due to the models themselves, but to poor integration into real-world workflows, over-reliance on general-purpose tools, and a tendency for companies to treat AI as a standalone experiment rather than incorporating it into core operations. The findings are based on interviews, employee surveys, and analysis of actual company deployments.
This statistic is often treated as evidence that AI doesn’t work. A more accurate interpretation is that leaders are measuring the wrong thing. They use efficiency-based metrics to evaluate capacity expansion inputs.
It’s not a failure of technology, it’s a failure of leadership.
What does it look like inside a real business?
PetFork currently operates 36 veterinary clinics and is expanding to hundreds of veterinary clinics as part of a more than $150 million-funded initiative to fundamentally transform veterinary care. Our Regional Managers are responsible for virtually every aspect of each region’s store-level performance, including revenue, workforce, scheduling, inventory, procurement, quality of care, compliance, patient outcomes, pricing, customer experience, team development, retention, training and culture.
Each regional manager is effectively responsible for making thousands of detailed decisions per week based on hundreds of reports, dashboards, audits, reviews, and operational signals. Ultimately, all of this is reflected in the performance of individual clinics.
Today, effective regional managers may spend 40 to 50 hours per week reviewing reports, identifying problems, and supporting clinic leaders. Even if you have a talented analyst, your work is time-constrained. Trade-offs are inevitable. Sample the data instead of examining everything. Deeper in some areas and shallower in others.
Our goal for the coming year is to fundamentally break that constraint.
We are working with regional managers to build AI agents that generate synthetic human intelligence time. Its ambition is simple and radical. We want to equate a 40-50 hour human work week to a 500 hour analytical work week without requiring people to work more.
Regional managers still work 40 hours. The remaining 460 hours are SHIH.
These agents review every invoice, every schedule, and every inventory decision. They analyze all NPS scores, eNPS scores, Google reviews, performance metrics, and more. Compare results not only weekly, but across time periods, cohorts, and locations. They work across our learning and development library to create bespoke development plans for each individual team member.
All that intelligence is consolidated and distributed to regional managers. Humans decide what is important. Humans prioritize. Humans communicate and lead.
Functionally, their roles change. Regional managers no longer operate with one person’s analytical bandwidth. They are working with things that previously required an entire team of analysts.
I would never have tried something like this before. Not because it wasn’t worth it, but because it was economically impossible. Human analysis was expensive and not scalable.
AI changes that equation.
Why ROI misses the point early?
One reason many leaders are disappointed with AI is that these changes don’t show up cleanly or quickly in financial results.
Enabling synthetic human intelligence time does not immediately reduce costs. It does not automatically increase your earnings the week it is introduced. In the early stages, the profits are small. Your decision making will be a little better. Patterns will be caught faster. Teams improve incrementally. Waste does not decrease dramatically, but silently.
This is not a defect. That is the nature of the compounding system.
The benefits of expanding intelligence capabilities accumulate over time. Like other combined effects, they appear small at first and are hardly visible on their own. But in the long run, they control the outcome.
Organizations that evaluate AI solely based on short-term efficiency metrics will miss this completely. Organizations that understand that SHIH is a compounding benefit design for durability rather than immediate optics.
This disconnect helps explain why PwC found CEO confidence in sales growth to be at a five-year low. Weak AI returns are increasing strategic uncertainty. This is not because the tools lack power, but because organizations have not been redesigned around them.
Benefits don’t come in one item, but in thousands of iterations of better decisions.
important questions now
As the marginal cost of thinking collapses, the range of analysis an organization can do expands dramatically. There will be no competitive gap between companies that automate faster and those that don’t.
It will be a battle between companies that continue to think in terms of efficiency and those that redesign around capacity and compound interest.
AI will not replace humans. It will redefine what small, focused teams can accomplish.
The question leaders should be asking now is not where they can cut costs.
The question is, if quality thinking was available for almost free, how many hours of synthetic human intelligence would you spend and what problems would you ultimately tackle?
The opinions expressed in Fortune.com commentary articles are solely those of the author and do not necessarily reflect the author’s opinions or beliefs. luck.
