Challenge: 12 months, 12 projects. My journey towards AI-assisted execution.

I recently realized something that gave me a bit of vertigo: I've been developing software for over 15 years. I've written code for companies around the world, solved complex architectures and seen dozens of technologies come and go. But looking back, my 'Personal Projects' folder looks like a graveyard of brilliant ideas that never saw the light of day.

It was always the same: I'd start with enthusiasm, get lost in the architecture, chase perfection, get tired of being a one-man band (design, backend, frontend, marketing...) and leave it half-finished.

That ends today.

I've decided to take on a challenge that terrifies and motivates me in equal parts: 12 months, 12 projects. And this time, I have an ally that changes everything: Artificial Intelligence.

The "Army of mini-mes": AI as a multiplier

I've been experimenting with tools like Claude, Copilot, Gemini and Codex for months. I don't have an absolute favorite yet —sometimes one hallucinates less with code and another writes better—, but the conclusion is indisputable: AI has given me superpowers.

It's like being 10 years old again and activating cheats in a video game. What used to take me weeks of research and boilerplate configuration now gets solved in hours or even minutes. I feel like I have an army of 'mini-mes' working 24/7, allowing me to focus on what really matters: delivering value and shipping the product.

Project #1: Rumbo — my financial health, at a glance

The original idea for kicking off this challenge was different: tappttoo, "the Duolingo for tattoo artists". It's still on the list — I'm fond of it, its month will come — but when I sat down to plan the first project I realized I had something more urgent in front of me: I'd been meaning to stop tracking my finances half in a spreadsheet, half in my head.

So project #1 is Rumbo: a personal financial health app. No spreadsheets, no checking five different bank apps — one place to see income, expenses, debts and savings, with a score that tells you at a glance how you're doing, and with data that never leaves my own Mac.

• 100% private data: local Postgres, no cloud provider involved.
• No black box: the score and recommendations are explained factor by factor, not a magic number.
• From zero to usable in 3 days: hexagonal architecture, tests on both layers, CI and even real bank statement import.

Every detail — what I built, the real bugs that came up testing it for real, and why there's no public demo — is in the dedicated Rumbo post, the first in the monthly 'Tech Sheet' series.

Why am I doing this publicly?

Documenting this process on my blog has three clear objectives:

• Accountability: If I publish it, I force myself to follow through. I don't want you asking me next month "what happened to project 2?".
• Personal Portfolio: After years working for others, I want my website to reflect what I'm capable of building on my own.
• Share the learning: I want to show how I use AI day-to-day. Not just the successes, but also when AI gets it wrong, the bugs that keep me up at night and how I solve them.

What can you expect each month?

Each month I'll publish a detailed analysis of the launched project:

• The Tech Stack: What languages and frameworks I used.
• AI Log: What AI tools helped me and what prompts were key.
• Lessons learned: The good, the bad and the ugly of launching a product in 30 days.

This is not a quest for perfection, it's an ode to execution. If you're a developer, entrepreneur or simply someone curious to see how far one person can go with AI's help, I invite you to join me on this journey.

Shall we begin?