Two days
I built an app and released it on Google Play. In total, it took me about two days of work. Yes, scan the QR code, you’ll be able to install it.
I am not a full-time mobile developer. I am a service management and digital strategy consulting director. I had an idea, I had an AI assistant, and I had a weekend. That was enough.
Five years ago, the same idea would have stayed in a notebook. Not because it was a bad idea. Because the bar to turn it into real software was too high for people from cost, time and skills aspects.
The old wall
For most of software history, ideas were cheap and building was expensive. The gap between them was a wall, and most ideas died at its foot.
To cross it, you needed some combination of:
- Skills. Years of learning a language, a framework, a deployment pipeline.
- Money. Enough to hire developers, or enough savings to stop working and learn.
- Time. Months of evenings, with the idea going stale while you fought the tooling.
- Access. Knowing someone technical who believed in your idea enough to build it.
So the people who shipped software were mostly the people who already had those things. The nurse with a better rostering idea, the tradie with a quoting tool in their head, the teacher who knew exactly what their class needed: they waited, or they gave up.
What actually changed
AI did not make building software free. It moved the hard part.
The old hard part was how, e.g., syntax, libraries, build errors, store submission rules. AI now handles most of that. When my build failed, I pasted the error and got a fix. When I needed a database layer, I described it and got one. When the Play Store asked for a privacy policy and content rating, I was walked through it.
The new hard part is what and why: knowing the problem, deciding what to build, saying no to features, and judging whether the result is any good.
That is a huge shift. The scarce skill used to be writing code. Now it is having clear intent and good judgement.
Where this leads us
When the cost of making something drops close to zero, the world fills up with it. We have seen this before. Cheap printing gave us pamphlets and newspapers. Cheap cameras gave us YouTube. Cheap publishing gave us blogs and social media. Software is next.
Here are five shifts I expect.
- A flood of small, personal software. Most new apps will not be startups. They will be tools built for one team, one family, one niche. Software for an audience of ten will become normal, the way a spreadsheet for one person is normal today.
- Domain experts become builders. The best new products will come from people who live with the problem, not from people who are good at code. The physio, the logistics coordinator, the school admin. They no longer need to explain their idea to a developer. They can just build it.
- Distribution becomes the new wall. If anyone can build, building stops being the advantage. Attention, trust and reach become the bottleneck. App stores, search and social feeds will get more crowded. The winners will be the ones who can get found and stay used.
- Taste and judgement become the premium skill. When ten people can build the same app in a weekend, the difference is in the details. Is it simple? Does it solve the real problem? Does it respect the user? Good product sense will matter more than ever.
- Professional developers move up the stack. Developers do not disappear. They shift to the hard problems AI still struggles with: architecture, security, scale, integration, and fixing what the weekend builders break. Their work looks more like engineering and less like typing.
The catch
A low bar to build is not a low bar to build well. Some things did not get easier.
- Security and privacy. AI can write code that works and still leaks data. Someone who never learned the basics may not know what to check.
- Maintenance. Shipping version 1.0 takes a weekend. Keeping it alive through OS updates, store policy changes and user bugs takes years.
- Ownership. If you cannot read your own code, you are trusting the AI with everything. That is fine for a hobby. It is risky for a business.
- Noise. More software means more low-quality software. Users will get more cautious, and trust will be harder to earn.
Inside organisations, this becomes a governance question. When any employee can build and ship a tool, shadow IT turns into shadow software. The answer is not to ban it. It is to set clear guardrails on data, security and support, so people can build safely.
So what should you do?
If you have an idea that has been sitting in a notebook, the excuse is gone. Build a small version this weekend. You will learn more from shipping it than from planning it.
If you lead a team, assume your people are already building. Give them guardrails, not roadblocks.
If you are a developer, lean into the hard parts. Judgement, architecture and security are now your edge.
The question is no longer can we build it. It is should we, and will anyone care. That is a better question to be stuck on. It means more ideas finally get a fair chance to meet the world.