Things nobody will tell
you in an alignment meeting.

01 · The diagnosis

Everyone uses ChatGPT. Nobody uses it for what matters.

For two years we've been asking ChatGPT where to eat, how to write an email, what movie to watch. We use it as a smarter search engine, a spell-checker, a late-night therapist. And that's fine.

The problem is what's not happening on the other side: inside our companies, our teams, our processes that are 3, 5, 8 years old and that nobody has ever questioned because they “work”. There, ChatGPT doesn't exist. There's still a spreadsheet someone updates by hand on Fridays. An onboarding process that depends on someone remembering to send you an email. Meetings to report things a system could measure automatically, in real time, without anyone having to prepare anything.

“We keep using ChatGPT to write emails and look up recipes. But we don't know how to embed it or iterate with it to streamline and automate the processes that cost us the most time and money.”

It's not ignorance. Nobody showed you how. The gap between “using AI personally” and “making AI work for your business” is enormous — and almost nobody is talking about that gap in concrete terms.

MCP servers that connect your tools directly to AI models. Agents that make decisions, query databases and execute actions without anyone calling them. RAG systems that turn your company's knowledge into instant answers. APIs that eliminate the repetitive work your team does every day — work that costs between 3 and 8 hours of salary per week, with nobody measuring it.

And there's a question very few are asking: Does my company show up when someone asks Claude, ChatGPT or Perplexity about a service like mine? Because search has changed. Language models are answering buying questions, comparing vendors and recommending companies — and very few businesses in LATAM are positioned to appear there.

02 · The LATAM moment

The most overlooked region with the greatest untapped potential

I'll be direct: we're the most behind region in the world when it comes to adopting AI tools for business. Not because we lack talent. Not because we lack a market. But because we keep waiting for things to “consolidate” before we move — as if the market is going to pause and wait for us.

Companies in Silicon Valley and the US have had this running for months. In some cases, more than a year. AI agents that filter leads automatically. Pipelines that process customer data and trigger actions without human intervention. Workflows that integrate CRM, support and billing into a single flow. AI search visibility that generates inquiries without paid advertising.

“In Silicon Valley this has been running for a year. In LATAM we're still using ChatGPT as a personal therapist.”

In Argentina, Brazil, Mexico, Colombia, Chile — across the whole region — we're still having that discussion. And that's a gap. Gaps, for those who know how to see them, are opportunities.

I'm not telling you your company is behind to make you feel bad. I'm telling you because the difference between implementing this today and waiting 18 months is enormous. Not in cost — cost is already accessible to any SMB. The difference is in positioning, in processes optimized before your competition, and in the customer data you'll have while others are still debating whether to “run a pilot”.

LATAM has always been the secondary market. The one that arrives late, the one that adopts when everyone else has moved on. My vision is for that to change — company by company, sector by sector. Not because it's a noble cause, but because the opportunity is real and the moment is now.

03 · Short-term vision (2026–2027)

What's happening right now — and most people aren't seeing it

We're at the cheapest and most accessible moment in history to implement AI in a real business. Not in an academic paper. Not in a conference demo. In your business, with your team, with your current processes.

MCP is changing the way models interact with the world. An MCP server lets Claude, GPT or any LLM access your tools, databases and APIs directly — without the user having to copy and paste information between systems. An agent that books meetings, queries your CRM, updates records and answers customers in real time. This isn't the future. This already works. And very few companies in LATAM are using it.

Your company's data is an asset you're not leveraging. Every company has documented (or undocumented) processes, customer conversations, sales metrics, support tickets. All of that is raw material for an AI system that learns from your specific business. The one who starts collecting and structuring that data today will have an advantage impossible to replicate in 2 years. Because data can't be copied. It accumulates.

The operational cost of manual processes is being quantified, for the first time, in real time. It used to be hard to know how much it cost for someone to spend 3 hours a week updating reports. Now there are tools that measure that exactly. And the number always surprises — because it's not just time: it's human error, the delay, the bottleneck that holds up every subsequent process. Automating a process doesn't mean replacing a person. It means giving that person time to do what a machine can't.

“The one who implements today doesn't just save costs. They have data. They have speed. And they have an 18-month head start over the one waiting for the technology to 'mature'.”

05 · Work philosophy

No tech dogmatism. No 40-slide proposals.

I'm not a consultancy. I don't have an 8-week discovery process. There's no team of account managers between you and the person who will actually build the solution.

I work solo. I can assemble a team for projects that require it — but the advantage of working directly with me is that there are no intermediaries, no juniors executing what a senior sold, no status update meetings between people who don't have project context. I talk with you, understand the problem, and solve it.

My process is simple. A 30-minute call to understand the real problem — not the one in the brief, but the one behind the brief. A concrete proposal. A functional MVP in under 10 days.

The MVP isn't a throwaway prototype or something to show in a presentation. It's the first validation step. The question I want you to be able to answer in 10 days isn't “is it well-built?” but “does it solve the problem I had?”. That answer changes everything that comes after — and prevents months of development in the wrong direction.

“I don't choose the tools I know. I choose the ones that solve the root problem. Technology is the means. Your problem is what matters.”

If this week the best solution for your case is n8n with a simple webhook, that's what I build. If it's a multi-step agent with LangGraph and a custom MCP server, same. If in 6 months a better tool appears for what you need, I learn it and use it.

I pursue continuous improvement. If I don't know how to solve something specific to your business, I research it — not because I'm humble, but because the market changes every week and sticking only to what already works is a guarantee of falling behind. Technological complacency is as dangerous as operational complacency.

06 · Who I'm talking to

Those who need to move now — not next quarter.

I'm talking to those who need an increase in speed and efficiency in their company right now. Those who feel they can't keep up with the wave of new tools appearing every week — but find them fascinating as possible solutions to their bottlenecks.

They don't need to know how LLMs work. They don't need to understand what an API is or why Docker exists. What they do need to know is what's holding them back. That's enough to start.

Startups that want to validate an idea before burning their runway. SMBs with manual processes that consume time, generate errors and limit their scale. Companies that understand that adopting AI doesn't mean replacing the team — it means that team can do more with the same resources.

“I'm not here to sell technology. I'm here to solve what's holding you back — with the right tool, not the trendy one.”

What I'm not looking for: clients who want to “do something with AI” without knowing what. Projects where the success criterion is “that it more or less works”. Kickoff meetings with a 40-slide deck before writing the first line of code.

If what you want is an alignment meeting to decide whether to have an alignment meeting — there are plenty of consultancies that'll do that for you.