A few months ago I posted about Dextr - an AI physio app I'm building to help with my recovery after carpal release surgery on both hands. Here's what's happened since: The platform changed - as much as I love building in Bubble, it is still a mostly manual process and I was struggling to make progress around client work. So I switched to Lovable! Once again I was blown away by what AI can build and the speed at which it does it. Within 20 minutes I had a working app with signup/login, an onboarding process, dashboard, warm up and hand stretching exercises with pain level reporting and logging. After a few iterations, I began to see how quickly you can burn through credits, and this coupled with some emerging limitations led me to switch again. I used up my remaining credits iterating on the UI because that is where Lovable really does shine - then downloaded the code and set Claude to work. I'm now building the app locally using Claude Code and Gemini - feeding screenshots to Gemini for design iteration, then building prompts for Claude to implement the changes. It's a surprisingly effective workflow. Right now the app handles warm-up stretches and tendon exercises with guided timers, pain level check-ins, and session logging. There's also an early AI layer that generates a short practice session for me - scales, arpeggios, chords, and finger dexterity exercises, complete with tablature and a built-in metronome. You can see it in action in the video. I'm not playing bass yet. But I'm stretching every day using the app, and that's the first step. I find it almost poetic that I'm using my new skills to aid in the recovery of the thing that ended my old career. It's a perfect case study in using AI to build tools for our daily lives - not a startup, not a SaaS side hustle, just a tool I needed that didn't exist. And the barriers to doing this are dropping month by month, week by week. The end goal hasn't changed: play bass again. 🤘 #buildinpublic #ai #lovable #claudecode #rehabilitation #careerchange #nocode