If We Can Anticipate Buyers, We Can Also Anticipate Borrowers

I’ve worked inside the financial system, as Head of Fintech at one of Latin America’s largest banks. I’ve also built outside of it, launching products designed for people the banks never understood. From both sides, one thing is clear: the way we design financial infrastructure today is outdated, reactive, and blind to the lives it claims to serve.

We live in a world where tech companies can anticipate a shopper’s behavior with eerie precision. Amazon knows what you’ll buy before you do. Google serves ads at the exact moment you’re likely to convert. And yet, when it comes to lending, we still rely on backward-looking credit scores, outdated paperwork, and rigid repayment plans. This isn’t just inefficient, it’s inhumane.

Billions of people live outside the financial system not because they want to, but because it was never built with them in mind. In Mexico, for example, low-income families face ultra-expensive credit with motorcycle debt collectors chasing down missed payments. Rural areas still operate almost entirely on cash. Government transfers often disappear without accountability. Even the most respected microfinance institutions remain years behind on technology, and AI isn’t even part of the conversation.

Meanwhile, one of the most deeply human financial behaviors, expense sharing within families and small communities, remains completely unsolved. I’ve spent two years building Compago to address this. And still, there’s no global product that enables real-time shared responsibility for household expenses. It should exist, but it doesn’t.

So here’s the question: if we can anticipate buyers, why can’t we anticipate borrowers?

What AI Can Do That Finance Still Can’t

Traditional finance operates on delay. It’s slow to listen, and quick to punish. It lends based on static assumptions and reacts to defaults with consequences rather than with support. But AI, and particularly large language models, can change that dynamic entirely.

What if, based on subtle behavioral shifts, like a delayed income deposit, a dip in discretionary purchases, or a canceled subscription, AI could anticipate a borrower’s stress before they miss a payment? What if, instead of penalizing them after the fact, the system automatically offered a restructured payment plan? A lower monthly obligation, a deferred installment, or a timeline adjustment tailored to the borrower’s reality.

This isn’t speculative, it’s the same class of machine learning that already powers the most profitable recommendation engines in the world. But instead of maximizing sales, it could prevent defaults. It could preserve loan portfolios for lenders and dignity for borrowers, protecting the relationship between them. It could even prevent a borrower from being shut out of the system for years, which is often what happens today after a single late payment.

I’ve experienced this personally. When I couldn’t pay the total statement balance of my credit cards, my balance grew, my FICO score fell, and, despite never defaulting, I wasn’t eligible for refinancing. So I was stuck paying ultra-high interest rates on credit cards. The system is too slow to understand what’s actually happening because the entire process reflects a system that sees numbers, not context.

AI can see context. And that’s where everything changes.

The Infrastructure We Could Actually Build

This isn’t about replacing banks, it’s about enabling a smarter, more adaptive layer beneath the financial institutions we already have. A layer that listens, understands, and intervenes early. A layer that makes the system more human.

If I had access to large resources and compute, here’s what I’d do. First, I would create a global network for intelligent financial interoperability. A shared data layer, accessible with full user permission, that connects banks, payment processors, and wallets across geographies. This layer wouldn’t compete with financial institutions, it would empower them to coordinate proactive interventions for clients before problems escalate.

Second, I would deploy embedded AI financial copilots. These could live inside everyday tools, like WhatsApp, point-of-sale systems, or government portals, helping people understand and manage their obligations in real time, in their own language, in a way that makes sense to them.

Third, I would redefine creditworthiness based on behavior rather than bureaucracy. Let’s build models that capture real-world trust: informal income, cash-based activity, reputation within a community. Let’s give people a way in, not a way out.

The Real Work Is Infrastructure

This isn’t a call for better apps or a friendlier interface. It’s a call to reimagine the underlying systems, the architecture beneath everything, that shape how people access credit, and build financial lives worth trusting. The challenge isn’t about innovation at the edges, it’s about changing what happens at the core.

Today, when a borrower’s life shifts, when income dries up, when a child gets sick, when a family member loses a job, the financial system offers only rigidity, and no fast adjustment. We’re talking fees, damaged scores, and rejection that stays on record for years. It’s a system built to protect institutions, not to protect relationships. And the result is exclusion at scale, not because people are unworthy, but because the system is blind.

But it doesn’t have to be this way. If we can anticipate purchases before someone clicks, we can anticipate repayment strain before it turns into a default. If we can model a customer’s behavior to optimize conversions, we can model a borrower’s behavior to optimize their chances of staying afloat. The tools already exist in the open platforms being built by companies like OpenAI and xAI. The question isn’t whether this can be done, it’s whether we’ll build the infrastructure to do it for everyone, not just the top ten percent.

Because the longer we wait, the more people fall through the cracks. Not from lack of effort, but from lack of adaptation. And adaptation is exactly what intelligence, real, dynamic, machine-enabled intelligence, can finally offer.

If we can anticipate buyers, we can also anticipate borrowers. And if we can do that, we don’t just prevent defaults or reduce risk. We change the role of the financial system itself. From one that reacts, to one that responds. From one that punishes missteps, to one that protects momentum. From one that is close to most, to one that is open to all.

That’s the infrastructure worth building. And it’s within reach with AI.