Why adopting AI in your SMB does not depend only on technology (and what it does depend on)

What does it take for AI to work in your SMB? Partly technology, and much more the people and the processes. BCG estimates that around 70% of the value of an AI transformation comes from the human and organizational component, and the OECD adds that in an SMB, skills, guidance, data, connectivity, and access to financing also weigh in. Before choosing a tool, it helps to be clear about the problem, who answers for it, and whether there is enough to sustain the change.

Why does the conversation always start with the tool?

Because it is the visible part. In Costa Rica and across the rest of the region, the conversation about AI in an SMB usually starts with a product question: whether ChatGPT Team is worth it, whether a chatbot pays off, what the competition is using. And almost every guide out there answers on the same ground, talking about platforms and integrations.

The problem is that none of those answers touches what is usually holding the matter back. In an SMB, the team's skills, the lack of guidance, the state of the data, the budget, and how the work is organized can all weigh at the same time. That is why an assessment worth doing looks at the whole system before recommending anything.

Is a large company's assessment useful?

It is useful for stealing good questions, not for copying answers. Corporate frameworks are built on samples that look nothing like your company: Cisco's AI Readiness Index 2025, for example, surveyed organizations with more than 500 employees. Taking its general direction is reasonable; using its percentages as a yardstick for a 40-person firm is not.

For an SMB it is better to start from the OECD's specific research and adapt the questions to the case, the resources, and the risk you actually have.

What does the evidence say?

Four recent sources point in the same direction.

BCG, in February 2026, splits the value of an AI project this way: about 10% comes from the algorithms, 20% from the technology and the data, and 70% from the people, the processes, and organizational change. That is the 10/20/70 rule, and it does not say where to spend the budget; it says where to look when something is not working.

McKinsey's State of AI 2025 finds that redesigning workflows and defining who decides what appear among the practices of the organizations that do capture value.

The OECD, in December 2025, lists the conditions an SMB needs: skills, guidance, data, connectivity, computing capacity, and financing.

And Cisco's AI Readiness Index shows that 91% of the organizations it calls Pacesetters had a formal plan to support the change, against 35% of the overall sample, always with the caveat about the size of that sample.

Read together they say something prudent: infrastructure and data matter, and even so they do not replace leadership, skills, process, adoption, and measurement.

The eight dimensions work as a system: weakness in one can limit progress across all the others.

How do we look at it?

At VegasiO we review eight dimensions: strategy and use cases, people and change, leadership and governance, data, process, risk, technology, and measurement. We do not publish a fixed weight for each one because it changes with the case. A regulated workflow demands more on the risk and governance side; an integration that moves transactions demands more on the technical side.

We do use one entry question, and it is practical rather than statistical: whether there is someone in management backing the project and someone carrying it day to day, both of them named. It does not predict the outcome and it does not replace the full assessment. It avoids recommending something no one has the authority or the time to sustain.

Adoption can begin with only a few owners as long as accountability is clear and coordination is consistent.

Which question actually puts the decision in order?

It is not whether you are technologically ready. It is whether you have the conditions for this to work in practice: a problem that hurts, someone who answers for it, information to work with, and team time to adjust.

When the answer is yes, the technology gets resolved in weeks. When it is no, the best tool on the market will be sitting unused in three months, and nobody will be able to explain quite why.

Do you want to know where your company stands?

VegasiO's Discovery web takes 5 minutes, it is free, and it reviews those conditions with what you answer. At the end you get a concrete recommendation: whether the next step is a Diagnóstico de Adopción IA, an initial training session, or starting directly with a first, narrowly scoped project.