Not in a lab. Not as a bet on a trend. Inside a working software company, to fix something that was slowing us down every day.
The origin
Born inside a 50+ developer company.
Myzelium started at Three Astronauts, a custom software company with more than 50 developers building real products for real clients. Like everyone, we adopted AI and expected to get faster. We did not get as fast as the tools promised, and we set out to understand why. The answer became Myzelium: a living source of truth that keeps a whole team, and all its AI, aligned and building from the right context. We ran it on our own projects, every day, before it was ever a product. We have delivered work for Fortune 500 companies using Myzelium. Three Astronauts is still where Myzelium is proven. It is the deepest kind of dogfooding: we do not sell something we do not run ourselves.
Three founders, one problem they lived before they solved it.
Alejandro Cordoba Borja
Co-founder and CEO
Founder and CEO of Three Astronauts, the software company where Myzelium was born. He has spent his career turning technology into business outcomes, with a track record of major ROI on software projects for companies ranging from startups to the Fortune 500. Educated at Universidad de los Andes.
Daniel Soto Rey
Co-founder and CTO
CTO, senior staff engineer, and machine learning mentor at Three Astronauts, and co-founder of the banking platform Cowally. He works at the intersection of AI and production software, and leads the engineering and research behind Myzelium's living neural network.
Juan Esteban Chaparro Machete
Co-founder and COO
Operator at the intersection of technology and business. At Three Astronauts he built the operational frameworks and product practices that keep a 50+ developer company running, the same real-world discipline that shapes how Myzelium works.
The research partnership
Grounded in peer-reviewed science.
Myzelium's core ideas are not just battle-tested, they are published. We developed our approach in partnership with Universidad de los Andes, one of the top five universities in South America, and the work passed peer review at two leading international software-engineering venues, ASE 2026 in Munich and ICSME 2026 in Bologna. Both papers are available to read.
Selected for the IEEE Technology Summit 2026 AI Showcase in Budapest, October 2026.
Belief
What we believe.
We think the industry has been optimizing the wrong thing. AI made writing code fast, but the real constraint on building software was never typing, it was alignment. Myzelium exists to fix that. That belief is the heart of everything we build.