By: Federico Pascale

A few weeks ago, news about humanity’s return to space dominated the headlines. Taking advantage of a long weekend, I rewatched some films on the subject – among them, Apollo 13.

That mission suffered an explosion in an oxygen tank on the way to the Moon. Suddenly, three astronauts were trapped in a damaged spacecraft; the lunar landing was canceled, and on the way back, another problem emerged: the oxygen filters in the command module were square; those in the lunar module were round. They faced the challenge of making a square fit into a circle.

While the crew tried to keep the spacecraft stable more than 300,000 kilometers from Earth, in Houston, United States, the team of engineers monitoring every piece of mission data from Mission Control did not stand idly by. One of them walked into a room, dumped everything the astronauts had available on board onto a table, and said something like: ‘Gentlemen, we need to solve this with what’s on the table.’ No more, no less. They couldn’t order new parts. They couldn’t change the design. They could only use what they had.

And they solved it.

Now let’s think about our world. The world of enterprise technology in 2026. Our problem is exactly the opposite.

The Apollo 13 engineers had constraints: limited materials, limited time, no possibility of adding anything new. We have the opposite problem. Every week – without exaggerating – a new tool appears, a new model, a new provider, a new framework. Generative AI, agentic AI, copilots, code assistants, no-code platforms, orchestrators, autonomous agents. The menu is enormous.

And yet, when we sit down in front of a concrete client problem, we often find ourselves paralyzed. Not for lack of options. By excess.

Apollo 13’s lesson for the agentic AI era: solve real problems with what’s on the table.

Paralysis by abundance and back to the table

There is a concept in psychology that describes exactly this: the paradox of choice. When we have too many options, it becomes harder to decide, we are less satisfied with what we choose, and in many cases, we end up choosing nothing. The same thing happens when we try to pick something to watch on a streaming platform on a Saturday night. And the same thing happens when a client brings us a real problem and we arrive at the meeting thinking about tools instead of solutions.

I believe the Apollo 13 lesson remains deeply relevant. Not because we need to work under extreme constraints, but because the discipline of looking at what we have on the table and solving with that is a skill we are losing.

Before going out to search for the perfect tool, it is worth asking the usual questions: What is the real problem? What do I already have today that works? What do I genuinely need to add, and what is just noise?

The NASA engineers solved the problem because they understood exactly what they needed to achieve and worked with what they had. Without getting distracted. Without asking for more. Without waiting for something better to appear.

Constraint as an advantage

It sounds contradictory, but clear constraints produce better solutions. When the problem is well defined and resources are limited, creativity focuses. There is no room to wander.

Those of us who work in the GeneXus ecosystem know that logic well. We have spent decades solving complex business problems with an approach that prioritizes understanding the what before choosing the how. Modeling reality before jumping to implementation. That has not changed with AI. If anything, it has become more important.

Because in a world where tools are abundant, what is scarce is judgment. The ability to look at a table full of options and say: this is what I need, this is what I don’t, and with this I solve it.

What’s on the table

And here is where the metaphor closes in a way I did not expect when I started writing this article.

Because from GeneXus by Globant, that table already exists. And it has concrete, non-experimental elements: tools that are already running in production.

Globant Enterprise AI is the operational AI Platform that connects data, models, and agents at an enterprise level, with governance, observability, and control. It is not just another tool on the menu: it is the infrastructure that allows everything to be orchestrated without depending on a single provider and without losing traceability of what agents do.

GeneXus for Agents, launched in April of this year, opens our Knowledge Bases (KBs) so that AI agents can read, propose changes, generate code, and explore the existing system through standard protocols such as MCP. But here lies the key: agents propose, GeneXus validates and generates deterministically. The speed of AI with the consistency and business rules that critical systems require.

And GeneXus, with more than 35 years of history, remains what it has always been: the Platform that models business reality before jumping to implementation. Except that now, in addition to the traditional IDE, it allows interaction with your Knowledge Base from a CLI, from agents such as Globant CODA, Claude Code, or Codex, without losing any of what makes GeneXus reliable.

And for those who choose to keep working from the IDE, the table also has something that deserves special attention.

Patterns technology has been solving for years what many are now trying to reinvent with generative AI: automating application construction, reducing complexity, and guaranteeing consistency – allowing a development team to automate up to 60% of building a GeneXus application, with production-ready interfaces, integrated security, and scalable design. It is not new, nor experimental – it works. Sometimes the most powerful thing on the table is what was already there before the noise started.

It is not about choosing between generative or deterministic AI, between the new and the proven, between agents and the IDE. It is about having everything on the same table, working together, with clear rules about what each thing does.

The Apollo 13 engineers solved the problem because they looked at what they had available and got to work. Today, those of us who work in the GeneXus ecosystem have on the table a set of tools that did not exist a year ago. I invite you to sit down, make some coffee, and start with the right question: not ‘what tool do I use?’ but ‘what problem am I solving?’.