A fascinating stage, but with uncomfortable data
We are living a fascinating stage with artificial intelligence. Every week, new tools, platforms, agents, co-pilots, and productivity promises appear. The enthusiasm is real. The speed of innovation too. According to McKinsey's latest global study, 88% of organizations already use AI in at least one business function, up from 78% the previous year. Adoption ceased to be an advantage: it became an entry-level requirement. But the same data tell a more uncomfortable story. Only 6% of companies manage to capture significant value from AI, and barely a third have moved from pilots to implementation at scale. MIT documented something even more severe in its State of AI in Business 2025 report: 95% of generative AI pilots do not produce any measurable impact on P&L. RAND Corporation comes to a similar conclusion from another angle: more than 80% of AI projects fail, twice as many as traditional technology projects.
The real competition: turning AI into real value
My perception, in this context, is that the real competition of the coming years is not going to be defined by who has the best technology or the most sophisticated model. It's going to be defined by something much harder to build: the ability to turn AI into real value for real customers. Today we see the birth of many AI platforms. Some are very good. Others are very striking. Many solve one-off problems with impressive speed. But many are also born with a major weakness: they don't have a solid customer base, they don't have deep relationships with the market, they don't have a thorough understanding of the processes where they want to intervene, and they're still looking for the real problem that someone would be willing to pay for on a recurring basis.
What fails is not the model
This is not theoretical. Deloitte reports that 42% of companies abandoned at least one AI initiative in 2025, with an average sunk cost of $7.2 million per canceled project (S&P Global). What fails, almost always, is not the model: it is the lack of a well-defined business problem, reliable data and real integration into the workflow. And that's where the conversation gets interesting. Because, on the other hand, there are companies that already have something that is not built overnight: customers, trust, historical data, operational knowledge, understood processes, commercial channels, support, reputation and long-term relationships.
The advantage of those who already have customers
These companies, if they manage to incorporate AI intelligently, have a huge opportunity. Not because they're going to create the most advanced model in the world, but because they can apply that technology directly to real problems, in real industries, with customers who already trust them. The probability of selling something new to an existing customer is 60–70%; to a cold prospect, only 5–20%. That difference is a competitive advantage that no platform buys with a good demo. AI, on its own, is not a strategy. It needs context. It needs data. It needs distribution. It needs knowledge of the business. It needs to understand exceptions, rules, operational frictions, regulations, user habits, and industry-specific pains. And that, many times, is already in the hands of companies that have been serving their customers for years.
Know-how as a real differentiator
There's one piece of data from the MIT study worth highlighting: AI solutions purchased from specialized vendors with deep workflow integration are successful 67% of the time; those built in-house, only 33%. The pattern is clear: what matters is not building AI, but integrating it where there is already a business that works. I recently had a conversation with the CFO of a fast-growing Mexican company. His question was straightforward and very reasonable: "why implement this with you and not with using AI to create what I need?" My answer was also straightforward: our know-how is what makes us different. Our product incorporates AI capabilities, yes, but on a solid foundation — a proven product, used by the market, with customers who already know it, processes that already work and learnings accumulated over years of real operation. AI is a layer that powers something that already has traction, not a promise that is just beginning to look for where to land. That conversation, for me, sums up well the logic I see in the coming years. The right question is not "established company vs. AI startups." The right question is: on what basis is that AI mounted? When there is proven product, real customers, and operational knowledge, AI multiplies value. When that foundation doesn't exist, AI hardly builds it in time — and numbers from MIT, RAND, and Deloitte confirm this every quarter.
The three forces that will decide who wins
That's why I believe that the future will not simply be of new AI platforms, nor will it automatically be of traditional companies. It will be for those who manage to combine three forces:
- Well-applied AI technology. Not to use AI for fashion, but to integrate it into products, services and processes where it really improves the experience, reduces friction, increases efficiency or generates better decisions. McKinsey found that workflow redesign is the variable that most correlates with actual financial impact, and yet only 21% of companies using generative AI have redesigned at least some of their processes. The rest are adding AI on top of old processes, and that's why the ROI doesn't come.
- In-depth knowledge of the business. Understand the real problem, not just the superficial use case. Knowing where it hurts, where time is wasted, where there is risk, where there is opportunity and what it means to generate value in that industry. This knowledge is not downloaded from a model: it is built with years of operation, conversations with customers and corrected errors.
- A customer base that already trusts you. In business, distribution matters. Trust matters. Closeness matters. The cost of acquiring a new customer is five to twenty-five times greater than that of retaining an existing one, and a mere 5% increase in retention can translate into 25–95% more profit. When a company already has a built relationship with its customers, it has a powerful advantage in bringing them new AI-based capabilities.
AI doesn't eliminate the importance of having customers. It multiplies it.
Many new AI platforms are going to face a strong challenge: moving from attractive demonstration to sustainable value. Many established companies will face another equally important challenge: moving from viewing AI as a threat or a fad, to integrating it as a new strategic capability. The former will have to build trust. The latter will have to move fast. The advantage will not be for those who simply "have AI". That will soon be common. The advantage will be for those who know how to turn AI into better products, better decisions, better processes and better results for a community of customers that already recognizes their authority and value. In other words: AI doesn't eliminate the importance of having customers. It multiplies it. And maybe the real question for companies isn't: "Which AI platform are we going to use?" But rather: "How are we going to use AI to become much more relevant to the customers who already trust us?" That, I believe, will be one of the great battles of the coming years.