A video titled “Jonathan Blow – If AI Solved Coding, Why Is My Software Still the Same?” reached the top of Hacker News on 13 November 2026. It had 354 points and 479 comments when I looked. I couldn’t load the video’s description or transcript, so I won’t say what Blow argues in it. The title alone asks a fair question, and the comments show a lot of people want to answer it.
The claim behind the question
The claim is that AI has solved coding. Anyone can now describe an app and get working software in an afternoon. If that were the whole story, the software we use every day should look very different by now. Most of it looks and works much as it did.
I think the question has a plain answer. Writing code was never the bottleneck for most software that succeeds.
What actually changed
A lot changed, and I don’t want to wave it away. Vibe coding collapses the distance between thought and working output. You have an idea, and an hour later you have a prototype you can click through. That used to take a team and weeks.
It’s also real for personal use. A normal person can now build fairly competent software for their own needs. The risk is small because it only handles their own data. I replace many tools and scripts with ones I build myself this way.
I also expect consumer utility software to get hit hard. If a single-purpose utility is easy to build yourself, fewer people will pay for one.
What didn’t change
That is the only distance vibe coding collapses. The distance between a working product and a successful one is still there.
Take a clinic that has a working booking app built in a weekend. Does it fit how the front desk really takes calls? Do patients use it, or do they keep messaging on WhatsApp? Does anyone find the clinic when they search? None of those questions is answered by code.
Real-world testing still takes real users and real time. Their feedback still has to be read and acted on. Marketing and customer acquisition still decide whether anyone shows up. Faster building doesn’t touch any of that.
B2B software has an even bigger gap. Every serious build still needs back-and-forth on alignment, strategy, approach and security. AI can’t supply those without human context and experience, and you don’t get that from a single prompt. Context is the missing piece, and the quality of what you put in decides the quality of what comes out.
Why your software still looks the same
If building got cheap for everyone, building stops being the advantage. A competitor can copy your feature set in a weekend. What they can’t copy quickly is your customers, your reputation and what you’ve learned from watching people use the thing.
So the software on your screen looks the same because the hard parts of making software succeed were never typing. Teams that were slow at those parts are still slow. They just get to a wrong answer faster.
Where this touches what I build
With AeroChat, the part that matters most is local integration. Each market has its own payment processors and logistics partners, and global vendors tend to skip them. You can’t vibe code your way into those relationships, however fast the code gets written.
What a business should do
If you run a shop, a clinic or a service business, spend less effort on how fast something gets built. Fast building is cheap now and your competitors have it too.
Put the effort where the outcome is decided:
- Put a rough version in front of real customers early and watch what they do.
- Read the feedback, including the complaints, and change the brief before you change the code.
- Write down the context an AI can’t guess, such as your rules, your customers’ wording and your limits on refunds or bookings.
- Decide how new customers will find you before you build anything for them.
- Build your own small tools if they save you time, and keep the data they touch to your own.
Then measure the result in customers, not in how quickly the thing shipped.