Doctor builds an AI app for medical documentation without writing code and finishes top three at Anthropic’s hackathon

A practising cardiologist has shown how quickly ideas can now turn into working software. Michał Nedoszytko built an AI-powered medical app in just seven days without writing traditional code and finished in the top three at a major hackathon run by Anthropic. The competition attracted around 13,000 applications from across the world. Nedoszytko’s result has become a clear example of how AI-assisted development is lowering barriers to software creation and allowing domain experts, not just engineers, to ship real products.

How a doctor built an AI app without coding

The tool, called postvisit.ai, is designed to help doctors generate structured post-visit summaries after patient consultations. It can take inputs such as patient history, clinical notes and test data, then draft follow-up plans and assessments. The goal is to reduce the administrative workload on clinicians so they can spend more time with patients rather than documentation.
The app is positioned as a clinical support tool. Final review and medical judgment remain with the doctor, keeping it aligned with current medical and regulatory expectations.
Nedoszytko did not build the app using conventional programming methods. Instead, he relied on AI-assisted development tools that convert natural-language instructions into working software. This approach, often described as “vibe coding”, allows builders to focus on workflows, logic and outcomes rather than syntax.
As a cardiologist, Nedoszytko provided the clinical insight. He defined how post-visit documentation should look, what details matter most and where errors commonly occur. The AI handled much of the technical execution, dramatically reducing development time.

Inside Anthropic’s hackathon

Anthropic’s hackathon was unusually large and competitive. Around 13,000 applications were submitted, with only a few hundred participants selected. Over roughly six days, participants shipped hundreds of functioning products, many built by people without formal software engineering backgrounds.
Anthropic has highlighted the event as evidence that AI is changing who gets to build software, enabling professionals from fields such as medicine, education and law to prototype tools directly for their own industries.
Postvisit.ai stood out because it addressed a clear, real-world problem. Medical documentation is a major contributor to clinician burnout, and the app offered a focused solution to a widely recognised issue. The combination of deep domain knowledge and rapid AI-assisted development helped the project outperform more complex but less practical entries.

What this signals for the future

The success of postvisit.ai points to a broader shift in how software is created. For years, turning an idea into a product required engineers to translate concepts into code. AI-assisted tools are now narrowing that gap and allowing subject-matter experts to build directly.
This does not remove the need for engineers, but it reshapes their role and expands who can participate in building technology. In healthcare, it opens new possibilities for clinicians to design tools that reflect real clinical workflows.
Despite the attention, postvisit.ai remains an early-stage prototype. Any real-world clinical use would require further testing, validation and compliance with medical data protection and privacy regulations. Hackathon recognition signals potential rather than immediate deployment.

A small project with a big message

By building a working AI app in a week and finishing in the top three of a global competition, Michał Nedoszytko has shown how quickly the rules are changing. As AI lowers technical barriers, those who understand problems most deeply may increasingly be the ones who build the solutions.