Code as Commodity: observations since I hunted ChatGPT in 2022

Thanks for sharing this, @chrismessina ! It really resonates.

The point that commoditization doesn’t just cheapen code, but changes what’s worth building, feels spot on. When code becomes abundant, the scarce resources shift to judgment, intent, and accountability. The core question stops being “can we build this?” and becomes “can we trust it in the real world?”

That shift is something we’ve been running into firsthand while working on QualGent. As LLMs collapse the cost of writing code, the bottleneck moves downstream to proving behavior: does the app actually work across real devices, real users, and real edge cases, continuously?

AI agents generating code is quickly becoming table stakes. Systems that can observe reality, verify outcomes, and close the loop with accountability feel like the next durable layer. In a world where everyone can ship, the real advantage belongs to teams that can ship with confidence.

Appreciate you articulating this so clearly, it captures a transition many of us are feeling but still learning how to name.

@chrismessina
The new economics of AI force every founder to face a brutal question before they write a single line of code: "Am I just building a 'me-too' product?"

Code has been cheap for a long time; the real cost has always been the talent required to execute with quality. In a world of infinite brushes, the question isn't whether you can paint—it's whether you’re a painter or just a paintbrush operator.

Thanks for sharing this - it is an interesting thought.

The question that I have is: when do we reach the point where it is cheaper to self-develop everything through AI (thus making paid software obsolete) or will we just never reach that point. Right now, what I see is lots of interesting new concepts popping up that are implemented because the effort in doing so has decreased, just like you mention in your post.

That said, what I am now finding out about myself: because of AI am I actually willing to pay more for services than I ever had before. I have more subscriptions to useful tools and am less conscious about it because in my mind they are replacing actual coworkers/freelancers I would have had to hire before. So for now this tells me that at least for the moment, there could be potentially more money in the market of SaaS services not less - albeit going hand in hand with a decrease in actual workforce demand.

this really resonates. the flood of AI products is wild - I'm seeing it firsthand.

launched a desktop tool for reddit scraping a few months ago ( Reddit Wappkit ). nothing fancy, just solves a specific problem I had. but the interesting thing? most "competitors" now are just ChatGPT wrappers with pretty UIs.

the ones that survive seem to be the ones solving actual workflows, not just slapping AI on everything. like you said - code is cheap now, but understanding the problem isn't.

curious if you think solo builders have an advantage here vs well-funded teams? feels like speed matters more than resources now.

"Unlocking previously uneconomic uses" is the most interesting part here. I've found myself generating "disposable software"—scripts for one-off tasks that I would never have paid for or spent time coding manually. But if SaaS is dying because the software itself is cheap, where does the value capture move to? Is it purely in proprietary data, or do we see a shift back to service-heavy models where the AI is just the backend implementation detail?

Hunting ChatGPT in 2022 was a hinge moment ,because it collapsed the cost of experimentation in tech and in other sectors too.

When software is so cheap to create, what signals do you now look for to separate noise from durability?

Thought provoking read indeed.