We founded Maladeta Studio in February 2021. This post is going up in August, which means I'm about six months late to my own anniversary. I kept putting it off because I didn't have a clean way to close a chapter that, if I'm honest, doesn't feel closed at all. But punctuality was never really what this blog was for. What matters is that this is the point where I stop writing about the next feature and try to look at the whole shape of the thing.
So: five years, give or take. Here's what they actually looked like.
1. From one freelancer to a studio with a real product
Before there was a studio, there was GuidAran — a tourism app I started building in December 2019, alone, for a valley of about 10,000 people. That's the actual origin of Maladeta Studio, not the incorporation date. The studio was an attempt, in February 2021, to make that kind of work sustainable rather than a plan to build an empire.
For the first couple of years, sustainable meant bespoke. NaturPark, the civic tech work on parking and mobility, JoRecicli for waste management — all of it was building what someone else needed, on their timeline, to their spec. I wrote about the trade-offs of that honestly in the "year one" post: rural, small, dependent on whoever walked through the door with a brief. 2022 added MeteoTrail's checkpoint-level weather predictions and SnowTeach, a two-sided marketplace we launched without knowing if either side would show up.
2023 is where the shape changes. Lueira wasn't a client project — it was the first time we decided what to build before anyone asked us to. A booking engine with real-time availability and multi-tenancy from day one, built as a bet rather than a brief. Everything since has compounded on top of that bet: the WhatsApp AI agent in 2024, Verifactu compliance absorbed directly into the product, Lueira Team for instructors in the field. By 2025, Lueira was running across five sports verticals, and the question SnowTeach had left open since 2022 — does a marketplace model actually work for us — finally got answered by generalizing it into RocRoi, KayakGrandTour, and NaturaAdventour.
Five years ago I was one person building one app for one valley on a freelance contract. Today there's a product with real, paying customers who never think about us at all — which is, I think, the actual sign that it worked.
2. The throughline we didn't plan
If you read back through five years of these posts, one argument keeps resurfacing in different clothes: a small studio embedded in a specific place can absorb complexity — legal, regulatory, technical, linguistic — that would otherwise land on people who have no business becoming experts in it. A ski instructor shouldn't have to understand Verifactu to issue an invoice. A valley of 35,000 Aranese speakers shouldn't need a large AI lab's permission or roadmap to get language tools that work for them. Val d'Aran by UTMB shouldn't have to build its own weather infrastructure from scratch and hope it holds at 3 a.m. on a mountain pass.
Here's the honest part: I did not write that thesis in the founding post. Go back and read it — it isn't there. What's there is a person trying to make freelance work in a rural valley financially viable. The throughline showed up later, case by case, absorbed into products because it kept being the thing that made them worth building, until eventually it was retroactively obvious enough that I could write a paragraph like the one above and have it sound intentional.
It wasn't. I think that's worth saying plainly, because "we knew all along" is a much better story than "we noticed a pattern in our own decisions after the fact," and only one of those is true.
3. What hasn't gotten easier
The founding post and the year-one post both said the same uncomfortable thing out loud: hiring in a valley of 10,000 people, most of whom are not software engineers, is genuinely hard. Five years later, that has not changed. If anything it's a sharper problem now, because we're running more surface area — a mature product, an emerging infrastructure layer, a new B2G line, an experimental geospatial project — with a structurally small local talent pool that hasn't grown to match.
The rural-talent hiring platform we started in 2025 is, in a slightly uncomfortable way, us trying to solve for other people a problem we haven't solved for ourselves. It's still unnamed and still unproven. I don't have a tidy update on it. It exists because the problem is real, not because we've cracked it.
4. How much is too much
Here is everything currently running at once, honestly listed: Lueira, a mature product across five verticals. An MCP-first pattern spreading across products — Ringover, the marketplace catalog. A standalone communications engine handling WhatsApp and email delivery. Weather and tide-prediction microservices. nautaran, a brand-new B2G GIS platform for municipal infrastructure that didn't exist a year ago. aran-twin, an ambitious geospatial digital twin of the valley that is still early and, frankly, still unfinished. And the Aranese-language AI work — Condò, the Salamandra corpus, the public dictionary app — which has nothing to do with revenue and everything to do with being a studio that lives here.
That is a lot of surface area for a studio our size, hiring from a pool that hasn't grown. I can make a coherent argument that it's not actually scattered — weather, geospatial, place, language, all of it draws on the same core expertise we've built over five years, applied in different directions. I can also look at that list and see the early shape of overextension, the kind that doesn't announce itself until something breaks. I genuinely don't know which of those two readings is correct. I suspect I won't know for another year or two, and I'd rather say that plainly than pretend the list above is a strategy instead of a list.
So that's five years, roughly, six months late. There's no bow to put on this. Lueira needs the next release. aran-twin needs someone to keep pushing it past "early." The hiring problem is still the hiring problem. This blog will keep going, because the studio keeps going, and I still don't know what the next post in this series will have to admit.