It's Magic
"It's magic." Those were the words from the audience during a presentation I gave for the Health Innovation School last week. During the final 15 minutes, I vibe-coded an app live with the audience, using a randomised, audience-generated prompt. The app was running on their smartphones within six or seven minutes.
I could hear the murmur ripple through the room when the first participants got it working. Someone laughed in disbelief. Another pulled out a second phone to show a colleague.
Now, I'm normally not a fan of live demos. Too many things can go wrong, especially when you let the audience randomise what you're building. But I was fairly confident I could pull it off. And honestly, that confidence itself feels like magic.
Why this matters for healthcare
Because it offers the possibility of giving agency back to the people who need it most.
Firstly, vibe-coding enables healthcare workers to build apps that help them with bureaucracy and repetitive work. Instead of waiting months for IT to deliver a generic solution, a nurse or administrator could describe what they need and have something functional the same afternoon.
Secondly, this approach creates personalised solutions instead of one-size-fits-all "solutions" that appeal to no one in reality. Every ward, every practice, every care team has its own quirks. Personal software can actually fit those quirks.
A market that keeps growing
This adaptability plugs directly into the wellness and fitness market, which is growing at 15–16% annually. A16Z Bio + Health recently highlighted a new segment they call "healthy MAUs": people who aren't acutely ill but want recurring, data-driven monitoring and pro-active health management.
Falling AI-driven care costs, new prevention-focused insurance models, and consumer willingness to pay for subscriptions are expected to make this preventive, engagement-heavy segment a major target for healthtech companies.
The body adapts, and so must our tools
I tend to think of the body as an adaptable organism rather than a machine. It's constantly reshaping itself to new environments and loads, with some adaptations (like metabolic shifts) taking far longer than others. That implies we continuously need to adapt our applications, or build new ones, depending on progress, changing environments, and altered aspirations.
The tools supporting this evolution are adapting too. Just look at the major announcements from the past two weeks:
- OpenAI launched ChatGPT Health (Jan 7) for consumer record syncing
- Anthropic released Claude for Healthcare (Jan 11) targeting enterprise workflows with HIPAA-ready connectors
- Google debuted MedGemma 1.5 (Jan 13), an open model for 3D medical imaging
Notice the pattern: all three prioritise administrative support over diagnosis. They're not trying to replace doctors. They're trying to free up time so healthcare workers can actually care.
What this means for you
If you're in healthcare innovation, the question isn't whether AI will change how care is delivered. It's whether you'll be building with these tools or waiting for someone else to build for you.
Try this: Take one recurring administrative pain point in your organisation. Describe it to an AI coding assistant. See what happens in 15 minutes. You might surprise yourself.