Hi Reader,

Welcome to Hal9 Weekly — insights from our founder Javier on turning AI ideas into real products.

Javier's Insights:

Vibe Coding Is Step One. Not Step Ten.

The worst advice I keep hearing in AI right now: "Just vibe code it and you're done."

You open Cursor, Claude Code, Replit, or Lovable. Type a prompt. Watch your app appear. Ship it.

That's not the finish line. That's step one out of a hundred.

Social media is pushing this idea that if you can generate something that runs, you've built a product. You haven't. You've built the first draft of the first version of the thing you'll actually ship six months from now.

The bar didn't get lower because vibe coding got easier. It got higher.

When everyone can generate a working app in an afternoon, "working" isn't a differentiator anymore. What matters now is whether you actually understand your customer. Whether your product is secure. Whether someone will trust it with their money. Whether it solves a real problem better than the ten other apps generated this week.

Vibe coding is a starting line — not a finish line.

If someone tells you that you can one-shot your company into existence with a single prompt, they're either selling you something or they've never actually tried to get paying customers.

The easy part got easier. The hard part got harder.

What's the hardest part of your build right now — the code, or everything else?

Customer Success Story: Limber Health

From Prototype to FDA Submission — AI-Powered Physical Therapy That Works on Your Phone

Limber Health is tackling one of healthcare's most persistent problems: patients who start physical therapy but never finish it. Based in Washington, DC, their team of sports medicine doctors and physical therapists built a hybrid care model that extends in-clinic treatment into the home — without requiring expensive equipment or compromising patient privacy.

The vision was ambitious: clinical-grade motion tracking and exercise feedback, running entirely on a standard smartphone camera.

The Challenge: Build an AI system capable of evaluating patient movement with the precision of an expert clinician — using only the camera already in a patient's pocket, while keeping all video data on-device to protect privacy. Then validate it against hundreds of real exercises and functional tests, with a medical team demanding clinical-grade accuracy.

Hal9 helped Limber Health:

🏃 Build an AI motion tracking runtime using pose estimation and image recognition to assess patient movement in 3D space — via standard webcams and smartphone cameras

🔒 Process all video locally on the user's device, sending only performance data to the server — never images or video — protecting patient privacy by design

📋 Document, test, and validate hundreds of exercises and functional tests alongside Limber's physical therapy experts

📱 Deploy across both browser and mobile apps (iOS and Android), making the technology accessible to any patient

🏛️ Support Limber's FDA approval submission for one of the functional tests developed through the project

The result? Limber is now pursuing FDA approval and positioned as a leader in digital health — with an AI system that doesn't just track movement, but gives patients real-time qualitative feedback that keeps them engaged and on track with their recovery.

Let’s Talk

Got an AI idea/building something and want to chat about it?

Book a free consultation with Javier — he’ll help you validate it before you build, so you don’t lose time chasing the wrong version.

Talk soon,

—The Hal9 Team