First, an apology. I wanted to get this to you last Tuesday, but a nasty virus had me out of commission all week. So this week you get a double: today's email, and tomorrow a second one with a step-by-step guide to building something like this yourself.
Today's email does two things. It gives you a free resource and the story behind it. And if it lands the way I hope, it gets you thinking about what you could build, because building is the best way to learn AI, and the best return you'll get on the time you put into it.
Your old projects might be a gold mine
From 2012 to 2014 I ran a small consulting service in Colombia called SIMPLE. We worked with companies to simplify their legal documents and the way they talked to their users. The idea was to bring some clarity into people's lives. It was a short run. The project was ahead of its time and the incentives weren't there, so I let it go. But I kept the method, and I kept working on it for years after, because it has always been a passion of mine. I just never knew what to do with it.
This year, with the models we've had since Opus 4.5, I decided to look at SIMPLE again. I believed the models at this stage could help me build something meaningful with it. I have a few things like this from the past that always seemed very hard to bring forward. The new models changed the question for me: could I rethink what SIMPLE is, with what I know now, and make something useful out of it?
What I realized is that the hard part had always been convincing companies to do the right thing and simplify their contracts. With AI, I didn't need to convince any one company. I could give people the capacity to simplify their own documents, and I could give it to everyone, for free. Something useful, and a way of being of service.
SIMPLE, live today
The service is live at simple-lab.org. You can use it right now.
You upload a document that confuses you. An insurance letter, a lease, a tax notice, a 401(k) summary. For now, in English and Spanish. You get it back by email as one clear page: what it is, what it means, what to do. Within 24 hours, usually in 3 to 6. Then your file is deleted. No account, no cost.
It's a working prototype, and it's fully usable. If you have a document that confuses you, try it and reply to tell me what you thought. That feedback is what I'm after right now.
I attached a short slideshow that explains SIMPLE in seven pages. If you know someone drowning in paperwork they don't understand, forward it to them; that's exactly what it's for. And if you'd like to collaborate with SIMPLE, or sponsor it to help keep it free, reply and tell me.
Building is the best way to learn AI
The method behind SIMPLE took years to build, and for nine years I couldn't get it out of the drawer. Turning it into a working service took days of focused work. That is the asymmetry AI brings: things that used to take years, or never happened at all, now take days of concentrated effort.
Building it has taught me so much, more than many other things I've done since I started working in AI. Doing a full build, end to end, is how you learn what a service like this needs. It's also a great way to dive into the deep end with Claude Code or a similar tool. It can be intimidating, and there are many steps, and it is worth it.
A lot of us are sitting on a gold mine of ideas and past work that is just there, waiting for the right moment. With where the technology is today, that moment might be now. You might be sitting on a great service, or a passion project, and you can get it out into the world in a lot less time than you think.
If you want to go look at what's in your drawer and you'd like some help bringing it out, book a call and we'll scope it together: cal.com/santiago-restrepo/15min.
Santiago