Hi Reader,

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

Javier's Insights:

Got a Vibe Coded MVP? Here's How to Actually Get It Live

Vibe coding is great for getting started fast. But launching to real users is a different game entirely.

If I was dropped into a startup team that had vibe coded their MVP and was asked to get it live, here are the three things I'd focus on immediately.

1. Figure out the business side first.

Before worrying about polish, I'd ask: How are we getting users to test this? Who's doing sales? How are we collecting feedback fast?

I'd build a waitlist. Start marketing. Get the go-to-market figured out. Too many founders keep building in isolation without ever putting real users in front of their product — and that's where momentum dies.

2. Test it relentlessly.

Does authentication actually work? Do the workflows hold up with multiple users? What happens on edge cases?

There's a big difference between something that works for you and something that holds up in the real world. I'd spend real time stress-testing it with actual users in real scenarios before claiming it's ready.

3. Bring in someone technical to look under the hood.

If the founder is non-technical, this step isn't optional — it's essential. You need someone who knows AI products to do a proper assessment: What are the blind spots? Are there security issues? Scalability problems? Is the tech stack right? Are you running expensive models that'll blow up your costs at scale?

You need to know exactly how close your MVP is to handling hundreds or thousands of users before you try to scale it. That's where a strong technical partner like Hal9 comes in.

Vibe coding gets you started. Getting to production is about business clarity, rigorous testing, and knowing what you don't know.

Customer Success Story: Limber Health

From Prototype to FDA Approval: How Limber Health Built Clinical-Grade AI on Regular Smartphones

Limber Health, a Washington D.C.-based digital health company, set out to solve one of physical therapy's biggest problems: patients don't finish their treatment. Between access barriers, high costs, and demanding work schedules, too many people drop off before they've recovered.

Their vision was a hybrid-care model — one that complements in-person clinical care with AI-powered support patients could access at home. The catch? It had to be clinical-grade quality, run on ordinary mobile devices, and protect patient privacy throughout.

That's a serious technical challenge. And it's exactly the kind of problem Hal9 was built for.

Hal9's team of data scientists and engineers developed an AI-powered motion tracking system using image recognition and pose estimation models. The system tracks patient movement in 3D space using standard webcam and smartphone cameras — no special equipment required. It evaluates functional tests, counts exercise repetitions, validates positions, and provides real-time feedback during therapeutic exercises.

Crucially, all video processing happens locally on the user's device. No images or video are ever sent to a server — only anonymized performance data. Privacy wasn't an afterthought; it was engineered in from day one.

To make it work at clinical standards, Hal9 collaborated closely with Limber's physical therapy experts to document, test, and validate hundreds of exercises and functional tests. The result is a product that genuinely augments expert clinical care — not just a wellness app.

The outcomes speak for themselves. Limber has submitted one of the functional tests developed with Hal9 for FDA approval, positioning them as a leader in the digital health space.

"We thoroughly enjoyed collaborating with Hal9 as our trusted AI partner, from initial design to successful FDA approval." — Chief Medical Officer, Limber Health

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