I was the guinea pig: what I think of Xentris and automating a country starting with yourself

The guinea pig was me

The Xentris team published an article with a title that sounds big: «How to automate a country (and why we start with ourselves)». I read it with a mix of pride and healthy embarrassment, because I show up in one paragraph: the CTO, in Colombia, who was the first one to climb onto the operating table. In other words, the guinea pig was me.

And I think it's fair to leave my version here, on my own site, because the company telling it is one thing and what I actually lived is another. Our rule is simple and a little uncomfortable: we don't sell anyone an automation we haven't suffered through ourselves first. It's not marketing. We've burned ourselves enough times that we never want to burn a client.

Why I believe a country is automated one person at a time

When you hear «automate a country» you picture a national plan, a government contract, a photo op with suits. I saw none of that. What I saw was much smaller and much more real: my own week. Invoices I built by hand, reports I copied from one place to another, follow-ups I kept forgetting. I started there, with the boring stuff, not with the epic stuff.

The trick wasn't to replace me. It was to give my work a layer of agents that draft the work, a dashboard where I watch in real time what they're doing, and a memory that doesn't wipe between one day and the next. I still decide. They just stop wasting my time. That border —who decides and who executes— is the whole thing.

A country isn't automated with a decree. It's automated when one person gets their Monday back.

The compounding effect, seen from my desk

The article calls it the «compounding effect» and draws it as a staircase. I watched it happen, step by step, which is why I believe it:

• Automate a task → you get hours back. (For me it gave back my nights.)
• Automate a role → you get a person back to think, not to copy and paste.
• Automate a company → you get a whole team back to do what a team should actually do.
• Automate a sector → now you start moving an industry. And several industries make a country.

The order is what matters. Nobody reaches the last step without stepping on the first. And the first is so unglamorous that almost everyone skips it: sitting down to look at what you do with your hands that a machine could do while you think.

Why from Latin America it makes more sense than it seems

There's an idea I repeat a lot, and it sometimes stings: the biggest return is wherever the most is still done by hand. And in Latin America a huge amount is still done by hand. What sounds like a disadvantage is actually the most fertile ground there is for this. Where everything is already automated, automating gives you 3% more. Here, starting from zero gives you leaps that would be impossible elsewhere.

I write this from Colombia, not from Silicon Valley, and on purpose. I don't think the next wave of productivity will come from a giant company abroad. I think it comes from the accountant, the small-business owner, the engineer in a mid-sized city who one day decides that their repetitive work should be done by a supervised agent. Xentris is betting on that, and that's why I'm inside.

Automating isn't firing (and I mean it)

This is the part where I cool the hype down. Automating well means erasing tasks, not people. The machine keeps the repetitive part; the human keeps what the machine shouldn't touch on its own: judgment, decisions, responsibility. In everything we build, the person stays at the wheel and the agent asks permission before it acts for real.

What I stand behind about Xentris… and what I hold back

Since I'm giving my opinion about the company I help build, I'll be honest in both directions. What I stand behind without hesitation: the discipline of testing everything at home before selling it, the obsession with keeping the human deciding, and the idea that this gets democratized from the bottom up and not from a tower. I'll sign that with my eyes closed.

What I hold back —or rather, where I'm hardest on myself: making sure the talk never runs faster than the results. It's easy to write «automate a country»; it's hard to hold that up client by client, without over-promising. My job, more than writing the headline, is to make sure it's true. That's why this article isn't applause: it's a commitment.

Start with one task

If you keep one idea, let it be this: don't wait for the grand plan. Pick one repetitive task you'll do this week, put a supervised agent on it, measure whether it gave you time back and only then scale. That's how I started, and that's how —one person at a time— a country actually gets automated.

If you want the other side of the story, the company's, read it at the source: «How to automate a country (and why we start with ourselves)». Here I put down what I saw from the inside.