Studio.com Flagship Creator field report
35 Years of Code, 40 Years of Violin, Nine Weeks on Studio.com. Here's What Held Up.
Studio.com said an AI app takes six hours. I gave it nine weeks. Three music apps survived. They launch tomorrow. In one of them, your instrument is the controller - literally.
Experiment summary
A CTO, violinist, and early Studio.com creator stress-tested the platform for nine weeks.
- Role
- Studio.com Flagship Creator
- Experiment length
- Nine weeks
- Studio framing
- Six-hour AI app build
- Onboarding
- June 27, 2026
- Apps launching
- Three
- Developer background
- Roughly 35 years programming
- Music background
- Roughly 40 years violin
I'm a CTO who builds AI systems daily, and I've been writing software since I was a young kid. Studio didn't recruit me for that. It recruited me because I'm a violinist with 300+ YouTube videos, a seat in the first YouTube Symphony Orchestra at Carnegie Hall in 2009, and a channel I mostly walked away from in 2013 to focus on technology.
Thirteen years later, the boomerang came back. I spent nine weeks deliberately looking for the platform's edges. Three apps survived. Eight didn't. Discarding ideas fast is the ROI.
I haven't made a dime yet. We find out tomorrow. But the R&D value and the return to music have already happened.
The three apps
None of them were the apps I planned.
Music Practice RPG
Your instrument is the controller, and your practice today changes the world tomorrow.
Violin Pitch Builder
Daily intonation challenges disguised as village tasks using the same RPG engine, with an authored curriculum for beginners that expands using pedagogical pillars for intermediate and advanced players.
Violin for Parents
Building confidence in music parents between lessons, without overriding the teacher.
Findings
Five things leaders should take from this
1. The AI kept revealing capabilities I didn't know it had.
Week one: it could write directly to the app blueprint, rewrite onboarding and marketing pages, and run tests. Week two: it built me a working metronome and a live AI chat widget. Later it generated full audio podcasts on its own, by accident. I found these by trying things I assumed wouldn't work. Most creators won't push this hard. The ceiling is much higher than the six-hour framing suggests.
2. My distrust of AI almost killed the best product.
Music Practice RPG was bad at first, so I did what a year of adversarial AI testing taught me: I over-specified everything. It got worse. The version that worked used a governing contract: a skeletal spine of what must stay true, and real freedom inside it. Beta testers' recordings and choices started shaping tomorrow's story. My daughter carried one run 47 days through Für Elise. A character I've voiced for 40 years said something I'd never written, and it fit. The lesson wasn't "trust AI more." It was get precise about what you trust it to do.
3. I fired the AI from the job I built the app for.
Pitch Builder was supposed to detect pitch. After hours of controlled testing, the AI was right most of the time and wrong some of the time. I can catch that when it writes code. A violin student can't. So my pedagogical pillars and canonical recordings became the source of musical truth, and the AI was reassigned to what it's actually good at: expanding the path between them and making a world worth returning to. Authority stayed with the human. That was a design decision, not a limitation.
4. Day one tells you nothing. Day 47 does.
AI apps look spectacular in a demo. Memory, continuity, and narrative drift only reveal themselves over weeks. Studio made building experiments cheap. It did not make longitudinal evidence cheap. Anyone evaluating creator apps should budget for that.
5. "Can AI do this?" is the wrong question for music.
Testing Violin for Parents, I realized the AI had my problem as a parent: it could produce advice I agreed with, and that still didn't mean it should interrupt what the teacher was building. Music teachers already hold authority. The interesting question is when AI should step in, not whether it can.
For Studio: the rough edges, honestly
Deterministic-variable bugs, image problems, audio concatenation failures, Cantonese limitations, test-state isolation issues, and behaviors that only became understandable under sustained stress. Many are being fixed. I note them because this is a snapshot of a fast-moving system, and I've seen that pattern before with React Native: the requirement decides the boundary, not the hype.
For YouTube: the work escaped the platform
Building these apps forced me to catalog hundreds of old videos into a searchable technique library, formalize a pitch curriculum, refresh my channel, and start recording again. Studio found me through YouTube. Then it sent me back.
For music teachers: I built these so you stay in charge
Every design decision above pushed judgment toward the student, the parent, and the teacher. The AI generates motivation and continuity. It does not grade your student's playing; it complements your professional teaching instead.
Eyes wide open
Building an app is now dramatically cheaper. Finding something people will repeatedly pay for is not. Of my 300 videos, maybe a dozen took off, and I never predicted which. The odds still favor the house.
I'd consider the nine weeks worthwhile even if all three apps make nothing, because two ROIs already happened: the R&D and the return to music.
Tomorrow may add a third - or not. I honestly have no idea what to expect.