88% of companies use AI. Only 6% see a business impact.

AI makes your people faster. But company productivity doesn't move. Every individual responds better, summarizes better, analyzes better — and yet the business outcome stays exactly the same. Here I break down why this happens, where the value gets lost, and what has to change for AI to stop being an individual improvement and start becoming an organizational capability.

88% of companies already use artificial intelligence. Studies show that at the individual level we're achieving gains of between 14% and 56% in speed and quality. Faster answers, better summaries, code written in half the time.

And yet.

McKinsey found in 2025 that only 6% of companies see a real impact on their operating results. MIT studied 300 implementations and concluded that 95% didn't move their P&L by an inch. The gap between what AI does for a person and what it does for an organization is enormous. And it has a name. We've lived through it before.

The paradox we already know

In 1987, Robert Solow said something that became one of the most quoted lines in modern economics: "You can see the computer age everywhere but in the productivity statistics." That line stood as a warning about what happens when a technology gets adopted without changing the way work is done.

Computer productivity didn't show up in income statements until the 1990s. It took nearly a decade to have a real impact. And it didn't appear when companies bought more computers. It appeared when the way we work was redesigned — when the way of managing and running the "business" was redesigned around the emergence of a new technology.

Today we're repeating the exact same story. AI is on every earnings call, in every budget, in every boardroom, in every conversation. We spend the day asking "how can we do this with AI?" But the value doesn't show up. It's not in productivity. It's not in the results. Today we're in a pure state of FOMO (Fear of Missing Out)!

This isn't a technology problem. It's a problem of how it's used. It's a cultural problem! (HR, where are you?)

How AI is used today in most companies

An analyst asks AI to summarize a report. It's done in seconds. It used to take an hour. Individual productivity, no question about it.

Now let's think about what happens to that summary. The analyst sends it to their boss. The boss reads it, discusses it in a meeting. Someone asks for a presentation. The presentation gets built. Ten people debate the arguments, question the data, and ask for a new summary with more detail on the same thing. The new summary comes out. Another meeting gets scheduled.

And throughout that entire journey, nobody caught that the metric had deviated before it caused impact. Nobody broke the problem down to find out where it was concentrated. Nobody assigned a plan with an owner and a deadline. And nobody verified whether the decision actually moved the number.

The analyst was faster. The organization spun in the same old circle.

HBR Analytic Services published a data point in April 2026 that defines the current state of affairs: only 18% of organizations have AI integrated into their work processes. 34% use it as a tool sitting alongside the process. When AI operates outside the workflow, it lacks the context to produce business results. People were handed AI without changing the processes they work within. Four out of five companies are still layering AI on top of processes that haven't changed.

AI has become an extraordinary individual tool bolted onto an organization that still operates the way it did ten years ago.

What's happening in Latin America

For companies in the region, the gap between individual and organizational productivity sits on top of another gap that already existed. Latin America's labor productivity went from 69% of that of the United States in 2000 down to 56% in 2019. It didn't rise. It fell.

AI is a huge opportunity to reverse this brutal (negative) trend. And what's worse, a WEF and McKinsey study from January 2026 found that only 23% of organizations in the region generate any economic value from AI. And the most revealing finding is that organizations that don't generate impact are three times more likely to have low data maturity. In my article "Your AI Doesn't Understand Your Business" (link) I go into a bit more detail on the "why" and the opportunities facing a world that's still unfamiliar to many parts of an organization. The phrase "I won't start with Data and Analytics until I have all my data nice and clean" is perhaps the most destructive phrase at an organizational level today.

In Mexico, 92% of companies plan to increase their AI investment, but only 1% reached what IMD 2025 (source needed) defines as full maturity — the complete integration of AI into workflows as a driver of business results. One out of a hundred. And to complete the picture, data management at Mexican companies didn't improve, it got worse, falling from 33% to 31% between 2024 and 2025. Data is moving backward while AI investment moves forward. That disconnect explains why we won't see the expected value, at least not in the way we're expecting it.

AI was adopted as individual behavior before it became reliable infrastructure. And that is exactly the difference.

Why individual productivity doesn't add up

It sounds intuitive to think that if every person works faster, the company should move faster too. But that's not how it works.

Between the person who responds faster and the company that improves its results, there's an invisible layer. It's the layer made up of processes, integrated data, business rules translated into a language the technology can understand, and a cycle that connects the signal to the action and the action to the result.

Individual productivity is measured in task speed. Organizational productivity is measured in outcome speed. And between the two there's a gap that no chatbot solves.

The cycle that closes the gap

If the problem isn't AI, but rather where and how we operate, then the question changes. It's no longer "which model should I use" but "what cycle do I have to turn a data point into a verified result."

There's an opportunity that few are seeing, because most still believe that having a chatbot connected to the data is enough. It isn't. What's missing is a closed-loop management cycle, and it has four moments. Each one feeds the next, and none of them work on their own.

Detect. Continuously monitor critical indicators (KPIs) against defined thresholds, and alert the owner before the deviation becomes irreversible. Don't wait for Monday's report. Don't depend on someone opening a dashboard. The organization finds out in real time, with the data, the context, and the trend. What today reaches the Director as a surprise in Monday's meeting should reach them as an alert the previous Thursday, with time to act.

Diagnose. When something deviates, break the indicator down through the organization's hierarchy to show exactly where the problem is concentrated, in whom, and how far each owner is from the threshold. Today, that work takes an analyst days: opening a dashboard, filtering level by level, exporting to Excel, and building a PowerPoint. By the time the answer arrives, the data has already gone stale. The alternative is an instant scan that shows everything at once, in seconds, and connects directly to action.

Act. Assign initiatives with an owner and a deadline. Not a note in a notebook, not a verbal agreement that dissolves between one meeting and the next. A concrete, visible plan that someone pursues until it's done. The difference between organizations that execute and those that just have good intentions lies right here — in the ability to turn a decision into a commitment with an owner, a date, and systematic follow-up.

Verify. Confirm whether what was done actually moved the indicator. Executing tasks and talking to an AI agent is not the same as creating value. If the result didn't improve, alert again and recommend a different course of action — having clarity on what the organization is doing to impact that objective becomes vital. This is the moment almost nobody has. Companies close tasks, file away plans, and move on to the next topic without checking whether anything actually changed. Verifying is what turns a management cycle into a results cycle.

80% of organizations get stuck somewhere between the first and second moments. They detect late, diagnose by hand, act with good intentions, and never verify. Value is lost at every transition. And it's lost silently, because nobody measures what was never closed.

What separates a productive organization from one that merely has productive people is the ability to close this cycle end to end, systematically, on the indicators that sustain the business.

What changes when the cycle closes

Each of the pieces I just described exists today separately, in some tool or another. Dashboards show data. Copilots answer questions. Project managers chase tasks. The problem is that none of them closes the full cycle.

An alert without a diagnosis is noise. A diagnosis without a plan is just another meeting. A plan without follow-up is a presentation. And a closed task that never checks whether the indicator moved is well-intentioned bureaucracy.

BCG found that the 5% of companies generating real value with AI get 1.7 times more revenue growth and 3.6 times more shareholder return than the 60% still seeing no results. The difference isn't which model they use or how much they spend. It's that they redesigned core processes, measure impact with discipline, and execute end to end.

They did the organizational work. The work nobody wants to do, because it doesn't show.

This is what we've built at PeopleOPTI with Amaia. A closed-loop management cycle that detects, diagnoses, acts, and verifies. Not one more dashboard. Not a copilot that only answers when asked. A system that closes the gap between knowing something is happening and making sure it gets resolved.

The productivity that matters

Solow was right in 1987. And he's right now. Computers didn't improve productivity until companies stopped using them to do the same paper-based work, just faster. AI won't be any different.

The difference between the 88% that use AI and the 6% that see results isn't technological. It's organizational, it's cultural. It's the real, deep redesign of how we work.

We've even learned to be hyper-efficient at Excel and dashboards. The ones seeing results don't have better models. They have redesigned processes, integrated data, and a cycle that connects every signal to an action and every action to a verified result.

For Latin America, this isn't a theoretical conversation. The region's productivity has been falling relative to the United States for twenty-five years. AI is the biggest opportunity we've had in decades to reverse that. But only if we stop confusing individual speed with organizational capability.

The productivity that matters isn't measured by how long it takes an analyst to put together a summary. It's measured by how long it takes an organization to turn a deviation into a result.

That's the only metric that moves the business. Everything else is noise disguised as efficiency.

Sources

McKinsey, "The State of AI in 2025. Agents, innovation, and transformation," November 2025. https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

MIT Media Lab / Project NANDA, "The GenAI Divide. State of AI in Business 2025," July 2025. https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf

Humlum and Vestergaard, "Still Waters, Rapid Currents. Large Language Models, Small Labor Market Effects," NBER Working Paper w33777, 2025. https://www.nber.org/papers/w33777

Daron Acemoglu, "The Simple Macroeconomics of AI," Economic Policy, vol. 40, January 2025. https://academic.oup.com/economicpolicy/article-abstract/40/121/13/7728473

WEF and McKinsey, "Latin America in the Intelligent Age. A New Path for Growth," January 2026. https://www.weforum.org/publications/latin-america-in-the-intelligent-age-a-new-path-for-growth/

Needed, EY, AMCHAM Mexico and KIO, "Digital Maturity Index 2025." https://amcham.org.mx/imd-2026-mexico-vive-su-mayor-salto-digital-en-seis-anos-de-medicion/

HBR Analytic Services, "What Drives AI Value. Why Modernization and Workflow Integration Matter," April 2026. https://appian.com/resources/analyst-reports/hbr-report-what-drives-ai-value

BCG, "The Widening AI Value Gap. Build for the Future 2025," September 2025. https://media-publications.bcg.com/The-Widening-AI-Value-Gap-Sept-2025.pdf

Federal Reserve Bank of San Francisco, "The Productivity Puzzle. AI, Technology Adoption and the Workforce," April 2025. https://www.frbsf.org/research-and-insights/publications/system-research-richmond-fed/2025/04/the-productivity-puzzle-ai-technology-adoption-and-the-workforce/