After 2 Failed Apps and Years of Freelancing, I Finally Found My Idea
"I'm sick of working for other people's dreams."
That sentence, said out loud one evening in Montpellier after weeks of chasing freelance gigs, is what started Mozart.
A bit of context
I'm a self-taught developer. Discovered code at 14, learned everything on my own. For the last two years, I freelanced for small companies while travelling: Vietnam, Turkey, Portugal, Morocco, Thailand. Between stopovers, I tried launching my own apps.
Two of them shipped:
- A Chrome extension to automate social-media replies via LLM. Decent idea, mediocre execution, zero distribution.
- A sports platform around events in France. ~200 visitors/day, $0 revenue. Alive, but small.
The lesson? Distribution. I knew how to build. Not how to get it known.
After burning out, I dropped code for a few months. When I came back, LLMs had made a massive leap, and something clicked.
The realization
Developers aren't just the ones writing each line anymore. We're becoming the ones who orchestrate, direct, and arbitrate.
The bottleneck has shifted. It's no longer code production, it's coordination. The more agents you stack, the more productivity you gain, but only if you know how to pilot them.
Every AI giant pushes its own closed interface. ChatGPT has one. Claude has one. But none of them is designed to make several agents work together, across different models, on interconnected tasks, in parallel.
That void is where Mozart was born.
What Mozart is
A local-first cockpit for orchestrating AI coding agents. Three pillars:
- For developers first , built for technical profiles who work with files and code.
- Local-first & Markdown , your files live on your machine, synced via GitHub. Low latency, more privacy.
- Multi-agent & extensible , coordinate agents from different vendors, distribute tasks, aggregate results.
Where I am now
Building the MVP. Or rather an "MVP++", because I have this annoying tendency to over-polish, a real flaw when "talk to users" is the rule. Working on it.
Next step: beta with a few peers and early adopters.
I already know my pain point will be the same one that killed my last two projects: distribution. So I'm starting that work now, including writing here.
If any of this resonates, solo founders, agent tooling, the orchestration angle, I'd love to hear what you think. Especially: what would make you actually try a tool like this?
Full story (with the embarrassing details about Y Combinator dreams and crossing deserts): mozart.build/blog/hello-world
Hello, world. 🎻
The freelancing years before finding the right idea are underrated as a research phase. You end up talking to more real customers than most founders ever do — you just don't frame it that way at the time.
The pattern I've noticed: the ideas that stick usually come from a problem you personally hit repeatedly while freelancing, not from brainstorming in a vacuum. What was the moment you knew this one was different from the previous two?
This is a strong founder story, but I think the most important part is the category shift you named: coding is becoming less about producing code and more about orchestrating agents, context, files, tasks, and decisions across models.
That is a much sharper wedge than “AI coding tool.” The crowded part is code generation. The less crowded part is the control layer around multi-agent work: local files, Markdown, GitHub sync, privacy, and a cockpit for technical users who want to direct the system instead of being trapped inside one vendor UI.
One thing I’d think about early is the Mozart name. It is memorable, but it may pull the product toward a creative/metaphor brand while the actual product feels more like serious developer infrastructure. If this becomes the local-first orchestration layer for AI coding agents, Exirra .com would give it a cleaner infrastructure-grade identity.
This timeline is so real. The freelancing years feel like "wasted time" until you realize they were just market research in disguise you were watching people struggle with problems up close. What was the moment the idea clicked? Was it a client pain point you kept seeing over and over?