Remi
How I built Remi's platform on my own, with AI
I built Remi's platform myself, with help from several AI models. It took me about seven days with almost no sleep, working around twenty hours a day, and by the end of those first seven days I already had a lot done.

- I built Remi's platform myself, with help from several AI models.
- It took me about seven days with almost no sleep, around twenty hours a day, and by the end of those days I already had a lot done.
- The hardest part has been grounding the models in real studies and datasets, so I can measure how well they work.
- I built the product in parallel with the legal side because legal was going to take a while and I'm convinced of the idea.
I built Remi's platform one hundred percent myself, with help from artificial intelligence. I used several AI models. It took me about seven days with almost no sleep. I was basically working twenty hours a day to get it done.
By the end of those first seven days I already had a lot done. I keep working on it every day, little by little, and working on stability.
How did I build Remi's platform?
Remi is the chronic care membership I'm building in Mexico for people with diabetes, high blood pressure or weight issues. I explained it in What is Remi. How I chose it is in Collections or diabetes.
I built the platform myself. There was no development team next to me. It was me, with several AI models.
It was about seven days, at a pace of almost twenty hours a day. When I say seven days, I mean a week of nearly nonstop work. On October 6 the core architecture was in place. I built the app, the admin and the system for doctors, and I bought the domain. I kept going. By October 22 almost all the apps and the whole system were done.
What was ready, and what connects today
Those three pieces are the core. The app is what the person uses. The admin is how things get run behind the scenes. The system for doctors is where they work.
Today a lot of things connect to that platform. A scale, a blood pressure monitor and a glucose meter, which I already covered in the three devices now connect to the app. Also a wearable band, earbuds and over-the-counter products, among others.
I won't go into each one here. What matters is this: the platform already takes them in.
It isn't standing still either. I keep working on it every day. Part of that work is stability: making sure what already exists holds up, not just adding something new.
The hardest part: grounding the models in real studies
The hardest part has been grounding every model in real studies, using data sources and datasets from different types of people with these conditions, so I can measure how well they work.
That work is ongoing. It's not an announcement of results. It's the least visible part, and the one that takes the most time: backing what I build with evidence, not with hunches.
Why did I build the product in parallel with legal?
People have asked me why I built everything before having the legal and regulatory side sorted out.
Because I knew legal was going to take a long time. I preferred to do it in parallel. And because I'm very convinced of the idea. I decided to start on the product right away while legal runs, at the same time, so I can be ready to launch.
If I stood still until legal was closed, the product would be delayed by the same amount of time. By building in parallel, when legal moves forward, the platform is already up and running.
That's not a launch date. I'm not announcing a day. Legal and regulatory are moving along on their own track: I covered it in Remi's legal side and in Remi and COFEPRIS.
The decision was clear: don't wait for legal to finish before starting to build. That's why today there's a platform I keep working on every day, and a legal process running at the same time.
Frequently asked questions
How did you build Remi's platform?
I built it myself, with help from several AI models. There was no development team next to me.
How long did it take you to build it?
It took me 20 hrs/day for 7 days, almost without sleep. By the end of those first seven days I already had a lot done, and I keep working on it every day, especially on stability.
What was the hardest part?
Grounding every model in real studies, using data sources and datasets from different types of people with these conditions, so I can measure how well they work. That's not an announcement of results.
Why did you build the product before sorting out the legal side?
Because I knew legal was going to take a long time. I preferred to do it in parallel. I'm very convinced of the idea, and this way I can be ready to launch. That's not a date.
Is Remi available yet?
Not yet. What is out there, and what I'm sharing here, is the work of building it: the [platform and devices](/blog/los-tres-dispositivos-de-remi-ya-se-conectan-a-la-app), the [legal side](/blog/lo-legal-de-remi-lo-que-me-dijo-el-abogado-antes-de-vender-una-sola-pastilla) and the [regulatory side](/blog/remi-y-cofepris).
Learn how to build a company by watching me build mine.
Every week: the decision, the number, the mistake and the win.