The AI stakes couldn’t be higher for your business. Here’s what you can do starting now.

If you are reading this while at the Puerto Rico Convention Center, where the Caribbean AI Summit opens today with some 1,000 attendees and more than 40 speakers, you already sense what is at stake. If you are not at the Summit, know there will be others.

The vital thing is to go to these events and learn, because the window to get this right has begun closing, and the time to act and adapt is right about now.

The generative AI so many of us have come to love and use, namely chatbot assistants that help draft emails, decks and reports, and design pieces and such, is already yesterday’s AI. The technology has moved into a more advanced phase: agents that plan, take actions across a company’s software, write and run code, and keep working for hours or days with little supervision.

It has also settled into a rapid-fire modality, with major new capabilities arriving every few weeks rather than every few months or years. We all had better adapt to that rhythm, or risk falling seriously behind competitors who do.

To adapt means to do three things, which I call the Adopt Protect Create imperative, or APC for short. But before I explain each one, it helps to frame AI in its proper historical context.

We have been here before

I was already at Caribbean Business in the 1980s and early 1990s and experienced, like the rest of the corporate world, the massive transition to personal computers, corporate servers and Office software. The learning curve was steep. Resistance was high. Too many colleagues did not adapt particularly well and fell into the digital divide.

Fast-forward to the dot-com era of apps and more software, cloud computing and social media, all coming at once. From our standpoint, covering how business adapted to the flurry, it has been at once thrilling and painful to watch the companies that did it well and those that did not.

If there is one thing that distinguishes today’s AI transition from previous ones, it’s the raw speed with which it is moving, magnitudes faster than any previous one, and your company cannot afford to watch it from the sidelines.

And now this. And now AI. Many folks make the case that this is the biggest such technology transition of all, perhaps the biggest business transition of all time. And they may be right, but the assumption is not automatic.

Go back further with me here, if you will. Scholars are arguing intensely about whether this PC-to-AI technology revolution of the last 40 to 50 years is, in fact, bigger than the multiple massive transitions of the 40 to 50 years between the 1870s and 1920.

Northwestern University economist Robert Gordon, for one, has argued in “The Rise and Fall of American Growth” that the great inventions that took hold after 1870 changed daily life more deeply than the digital revolution has of late.

It’s easy today to forget how huge the candle and lamp-oil industry, the horse-transportation industry, the manual-agriculture industry and the courier-communication industry all were before the 1870s, and how thoroughly they were replaced, I mean completely wiped out, by electricity, cars/trucks/planes, mechanized and industrial agriculture, and the telephone. Talk about brutal, in many ways ruthless, displacement of human labor, astonishing opportunities, distressing anxieties, and exponential growth curves. Research the simultaneity of those transitions, and we can have that conversation.

The net of the next

In every single one of them, and in others, the new systems ended up creating far more work than the old ones lost. MIT economist David Autor and his colleagues found that about 60% of U.S. jobs in 2018 were in occupations that did not exist in 1940.

For AI, the World Economic Forum’s Future of Jobs Report 2025 projects that 170 million jobs will be created worldwide by 2030, against 92 million displaced, a net gain of 78 million.

Yes, there is, and has always been, pain attached, always winners and transitioners. I’m the first to acknowledge that in the age of AI the contrarian narrative may very well be correct. Autor’s research shows, for example, that since 1980 automation has eroded jobs faster than new work has replaced them. A 2025 Stanford study found early signs that AI is cutting into hiring of young workers in the most exposed occupations. The net, I repeat, is never automatic.

The third front, Create, is the most exciting. This is the innovation and solution-building front, where companies capitalize on the AI boom by launching products and solutions they can sell here and export everywhere. Puerto Rico has a tool built for exactly that.

But the arc has invariably tended toward net-positive progress, and it is the net of the next that we at CB have always focused on. Today, that means doing the math on the complete impact of AI.

We as a newspaper, and each of us as professionals and human beings, must do everything in our power to soften the blow for those who suffer it, and to help companies prepare and capitalize as much as they (you) can. As a small, peripheral market in the grand scheme of things, Puerto Rico is extremely vulnerable to all such transitions. We won the one from an agrarian to an industrial economy in the mid- to late-20th century, and we must win again today.

I’m encouraged by Google’s decision in recent months to name Lila Ibrahim, DeepMind’s first Chief Operating Officer, as its Chief AI Readiness Officer. What’s amazing is the scope of her mandate: helping the world get ready for AI. Google, that is, has named a corporate officer to help the planet adapt to this transition. Think about that for a minute, and then imagine every significant company and government doing the same. Perhaps even yours.

Which brings us to what you as a company can do. In this column, let’s focus on your internal operations. We’ll come back to your contribution to broader readiness later.

Today, let’s focus on a simple three-part framework you can embrace: Adopt Protect Create. Simple to remember, that is, not to execute. But tech transitions have always been challenging to implement, and here we go again.

Adopt: get on the right side of the gap

The first imperative is to close the now well-known and dreaded adoption gap and to land on the right side of it. For you to be, as it were, among the companies that get out ahead rather than fall behind.

The gap is stark. Boston Consulting Group, surveying more than 1,250 companies last year, found only 5% achieving best-practice value from AI at scale, while 60% reported little or none. The study found that leaders involve their workforce far more often when they reshape processes and deploy agents, and nearly 90% of them expect most of their AI value to come from reinventing the way work gets done.

That usually requires hiring help for training, culture change, process redesign and embedding agentic solutions into daily operations.

McKinsey’s more recent 2026 global survey, published in August, found that close to 90% of respondents regularly use AI in at least one business function, and 44% say it is scaling across their enterprise. Yet only 37% report a positive contribution to earnings, and just 6% (a small uptick from BCG’s 5% finding a year earlier) qualify as high performers, attributing at least 5% of EBIT to AI.

Pick one of the many local AI advisory firms in the market, and let them help you join the 5%-10% of companies at the cutting edge of adoption.

The good news is that local players are emerging to provide exactly that help. Among the most innovative I’ve seen is INprende’s INbiz, a finely developed system to help companies implement robust enterprise AI. It is built around what the firm calls Company Brain, a central knowledge layer that gathers a company’s documents, strategies and insights, and connects everyone via a dedicated AI engine.

Another is Routina, an offshoot of Puerto Rico-based Flexio Group. Their approach is to install an AI layer in your company on top of the many softwares you already use and connect them to AI “teammates” that handle multi-step work such as intake, scheduling, routing, documentation and finance tasks. Every agent action is logged, and people approve what gets written back into company systems.

Yet another is BMA Group (formerly Brenda Marrero & Associates), the HR and Talent Solutions firm that takes a more people-based approach to enterprise adoption, including training, culture and process reinvention.

There are others still. Pick one and let them help you join the 5%-10% of companies at the cutting edge of adoption.

Protect: one resilience team, two fronts

In my last column on this subject, El Niño Meets Hugging Face, I matched AI resilience with climate resilience. Both involve emergent, hard-to-predict, potentially catastrophic risk. Both are moving faster than the institutions meant to govern them. Both are following a similar scope and timeline. One team at your company can, and should, manage the risk on both fronts.

On AI, the big and getting-bigger risk is cyber attacks. The Hugging Face hack raised the stakes, because it revealed that autonomous AI agents are already out there invading companies and governments without any human direction or participation.

Because they’re autonomous, they’re also more random and harder to detect and stop than the human hacks you may have already built cybersecurity against.

The news since Hugging Face has only reinforced the point, as we continue to see human AND autonomous AI activity grow.

In September, Google’s Threat Intelligence Group described a financially motivated criminal who used an AI coding assistant and a set of agent instructions to plan, build and run a campaign that compromised thousands of third-party credentials in less than six hours.

In August, CNN reported that hackers had used autonomous AI agents to attack targets in Taiwan. And Anthropic disclosed that a Chinese state-sponsored group had used its Claude Code tool in an espionage campaign against some 30 organizations, with AI performing an estimated 80% to 90% of the work.

In practice, that means testing your defenses against AI-driven, partially or fully autonomous attacks, including how quickly you can detect anomalous coordination inside your systems.

It also means, crucially, governing your own agents: give them only the access they need, log every action, keep people approving anything that writes to core systems, and treat your vendors’ agents as part of your attack surface.

As I wrote last month, your value chain is only as resilient as its most exposed link, and it is an exposure that can place at risk the progress done on the adoption front. One goes with the other.

Create: build here, sell everywhere

The third front is the most exciting one. This is the innovation and solution-building front, where companies capitalize on the AI boom by launching products and solutions they can sell here and export everywhere. Puerto Rico has a tool built for exactly that.

Multiple chapters of Act 60, the Puerto Rico Incentives Code, offer a 4% income tax rate on related business activities, including export services (Chapter 3 of Subtitle B, formerly Act 20). Smaller exporters, with business volume of $3 million or less, pay 2% for their first five years, and a novel pioneer activity can qualify for 1%. Distributions of those exempt profits are fully exempt from Puerto Rico income tax.

The list of eligible export services reads like an AI business plan: research and development, development of computer programs, distribution of software and services through the internet, cloud computing or blockchain, electronic data processing, IT consulting, creative industries such as apps and video games, and telemedicine.

Add Act 60’s research and development tax credits, and you have a framework for building AI products here for markets at scale.

If you’re at the Summit today and tomorrow, you’ll come across companies innovating, but more importantly for you, you’ll likely run into ideas and market opportunities you can innovate yourself.

Now, not later

As you do, Adopt Protect and Create across those ventures, as well. It’s three words, but one imperative.

The transitions of the 1870s and the tech revolution that began in the 1980s have rewarded companies that moved early and punished those that waited.

If there is one thing that distinguishes today’s AI version, it’s the raw speed with which it is moving, magnitudes faster than any previous one, and your company cannot afford to watch it from the sidelines.