Everyone keeps saying AI will replace developers.
I think it is proving something completely different.
💡 What Actually Happened at Anthropic's Hackathon
Anthropic recently wrapped its Built with Claude Code hackathon. More than 13,000 people applied. Five hundred were selected, given API credits, and handed one week to build something meaningful.
The winners were not who most people expected.
- A personal injury attorney — frustrated with the months-long permit correction cycle in California's housing market.
- An interventional cardiologist — exhausted by repetitive patient documentation eating hours out of every clinical day.
- An electronic musician — imagining instruments and tools that had never existed before.
- A roads and infrastructure specialist — watching systemic inefficiencies repeat themselves on every project.
- And yes, one software engineer.
Four out of five winners came from outside traditional software development.
Let that sit for a moment.
The Permit That Takes Months (But Shouldn't)
California has issued more than 429,000 Accessory Dwelling Unit permits since 2018. Over 90 percent return with corrections. Each round costs builders weeks — sometimes months.
The attorney who won built CrossBeam: a tool that reads a permit application, identifies the likely correction points before submission, and turns a multi-month feedback loop into a matter of minutes.
No senior engineering team. No product roadmap. No VC pitch deck. Just someone who lived inside that frustrating process every day — and finally had the tools to fix it themselves.
The Real Shift Nobody Is Talking About
Here is what I think most people are missing in the "AI vs developer" debate.
The conversation is framed wrong. It should not be: Will AI replace developers?
It should be: Who has always had the best ideas — and why couldn't they build before?
For decades, the people closest to real problems — in law, medicine, infrastructure, music, education, manufacturing — have had ideas. Many of them could articulate exactly what needed to exist. But they could not build it. Software development was locked behind years of technical barriers, specialised tooling, and massive development costs.
The result? Ideas died on whiteboards. Problems went unsolved because the problem-owner and the builder were rarely the same person.
That wall is breaking.
🤖 Domain Expertise + AI Assistance = The New Superpower
“Anthropic recently wrapped its *Built with Claude Code* hackathon.”
Here is the dynamic that is quietly reshaping the industry:
Knowing the problem deeply is now worth more than knowing the syntax perfectly.
A cardiologist who understands exactly what repetitive documentation costs in clinical time — that insight is genuinely rare. It takes years of practice to accumulate. A software engineer without that context might build a technically correct system that completely misses the real friction.
When AI handles the translation layer between intent and implementation, the bottleneck shifts. It moves away from "can you write code?" and toward "do you understand the problem well enough to direct a solution?"
That is a different kind of skill. And it is one that millions of domain experts already have.
This Is Not About "No-Code"
I want to be precise about something, because there is a popular but slightly off-target narrative here.
This is not the no-code story. No-code tools have been around for years, and while they democratised some building, they were always constrained by the platform's limits. You could build what the tool allowed you to build.
What is different now is genuine software creation guided by natural language. A cardiologist using Claude Code is not dragging blocks around a canvas — they are directing the construction of a real application, with real logic, real data handling, real edge cases.
It is closer to having a highly capable technical co-founder who happens to be available at 2am and does not need equity.
The ceiling is much higher. And the learning curve, while not zero, is navigable by someone who understands what they are trying to build.
💡 What This Means for Developers
If you are a developer reading this, I want to be honest with you: this does not reduce the value of deep technical expertise. The cardiologist still needed to understand enough to direct the build. The attorney still needed to test against real permit documents. The musician still needed to think through instrument logic.
What it does change is the exclusive nature of technical gatekeeping.
In the past, having a complex idea without coding skills meant you needed to hire, fundraise, or wait. Now the gap between vision and prototype is dramatically narrower.
For developers, the message is this: the people who thrive next will not just write code — they will combine technical literacy with deep understanding of a domain problem. The hybrid will be more powerful than either alone.
🤝 The People Who Were Always Closest to the Solutions
Think about who has been sitting on ideas for years.
- The operations manager who sees the same workflow break every quarter but cannot get IT to prioritise it.
- The teacher who has tried every classroom tool and knows exactly what none of them get right.
- The healthcare worker who documents the same inefficiency in every shift report.
- The logistics coordinator who has drawn the better system on a napkin a hundred times.
These people are not waiting for someone to build for them anymore.
The biggest moat in software may no longer be the ability to write code alone. It may be understanding real human problems better than anyone else — and now finally being able to act on that understanding directly.
💡 What the Hackathon Actually Proved
Claude Code itself started as a hackathon project exactly one year before this event. Today it supports thousands of builders.
The winners of this particular hackathon were not people who learned to code in a week. They were people who brought years of domain expertise to a problem they already understood intimately — and used AI tooling to close the gap between knowledge and execution.
That is the shift. And it is not small.
We are entering an era where:
→ People who deeply understand workflows will finally be able to build the tools that serve them.
→ People obsessed with solving inefficiencies in their field will no longer be blocked by technical barriers.
→ People who have spent years saying "I wish this existed" will be able to make it exist.
A Personal Note
I spend a lot of time thinking about who gets to build — and who is left waiting.
The traditional path to product creation was: have an idea → find a technical co-founder → raise money → build → ship. Each step was a filter. Most ideas never made it through.
What excites me about this moment is not that coding is dead. It is that the filter is being removed. The person with the deepest understanding of the problem is now capable of being the builder too.
That changes the quality of what gets built. And it changes who gets to participate in building the future.
If you have been sitting on an idea — whether you are a lawyer, a nurse, a teacher, a musician, or anything else — the timing has never been better.
The wall is breaking. Build something.
Have a domain expertise you have been itching to apply? I would love to hear what you are working on — drop me a note through the contact page or connect on LinkedIn.
Comments
Loading comments...