Gastos Compartidos, 2025-today
For more than nine years I built, fixed and maintained spreadsheets to manage the shared expenses of co-parenting. Every month meant entering costs, splitting them by rules that kept changing, closing the cycle and explaining the balance without the conversation turning into an argument. The spreadsheets worked, but every improvement cost a weekend.
With AI I could multiply by ten what I was able to build as a hobbyist, and finish shaping the application those spreadsheets were asking for. Gastos Compartidos (gastoscompartidos.ai) is a platform for coordinating expenses between two people: record with a photo of the receipt, split by your own rules (halves, percentages or per category), automate recurring costs, close the month with a clear balance, and resolve differences through bounded actions, approve, observe or propose, instead of an open chat. It is built for families sharing custody, and it also serves couples, flatmates, trips and the professionals who accompany those cases.
I designed, orchestrated and built it end to end: web frontend, backend and services, iOS and Android apps, MCP servers so that the sheets can be operated from Claude or ChatGPT, agents and skills. It is the product where I took furthest the way I work today, agents as participants in development rather than assistants, and the one that came out of a problem I lived at home.
Products for race organizers, 2025-today
As a participant I know the problems of the people who organize races, and these products are my way of giving something back to that world: a better experience for the people who sign up and less work for the people who organize. Today I build small, concrete products for them: one makes it easier to manage participants’ medical certificates; another takes load off support staff with autonomous agents that answer the most common questions from registered participants. With organizers I know I am running their first proofs of concept, built with agents working largely unattended and with little supervision. I do it as a hobby, without expecting anything in return: it is rewarding to keep building and collaborating with the sports-event ecosystem. I build them in my free time, with Claude Code, MCP servers and spec-driven development.
Medalio, 2020-2022
Medalio was digital ticketing and collectible assets for sport: when a fan bought a ticket, an app opened an experience before entering the stadium, and when the match ended the fan received pieces generated from the result, as a digital memory of the event. I incubated it at Founder Institute, which connected us with leagues in the United States, Europe and Central America; a European accelerator selected us to build proofs of concept with brands and clubs, and the most complete pilot ran in the United States with a football league, where I set up an LLC. The platform, on React, Supabase and Polygon, deployed on Google Cloud with Google credits and grants, I built entirely myself.
It did not scale for two reasons: the sales cycle to a league, long and with many stakeholders for a one-person startup, and timing, because the market for digital collectibles collapsed in that period. In parallel came the opportunity to join Globant. I kept it going for a while and closed it in December 2022.
Evenance, 2019-2020
Evenance was a platform for finding participatory sports events across the country. We incubated it at Founder Institute in 2019 and it received Google grants to build on Google Cloud. That same year I took it to SLUSH, in Finland, where we pitched and presented it to hundreds of investors, and I came back with connections and ties to the Finnish startup and innovation ecosystem. We ran a pilot and an MVP, and did not get to scale it. It was my first formal incubation and the first time I worked on amateur sport from the other side, the organizer’s.
AgTech and IoT, 2017-2018
Digirodeo was a livestock traceability and health app built around IoT vaccination devices, which I was invited to join as co-founder and head of technology. It had to work in the field, with devices used by people who do not write software. In the same period I built prototypes and pilots of IoT applied to monitoring and automating poultry farms: sensors feeding a learning system, digitization of the manual management processes, and a gradual transition to automated processes. Both projects were presented to acceleration programs and received seed funding. It was my first hands-on work in AgTech and IoT.
What repeats
They all started the same way: a problem close to me, at home or at a race, and the wish to solve it with the technology of the moment. Some scaled and some did not, and from each one I took something into the next.