I AI coded an entire platform for my work. It works great for me. I also recognize that this is not something I want to make into a commercial product because it was so easy that there's just no value.

I think this might be more of an comment on software as a business than AI not coding good apps.

Does the data not support a 2X increase in packages?

Pre-ChatGPT, in ~2020, there were about 5,000 new packages per month. Starting in 2025 (the actual year agents took off), there is a clear uptick in packages that is consistently about 10,000 or 2X the pre-ChatGPT era.

In general, the rate of increase is on a clear exponential. So while we might not see a step change in productivity, there comes a point where the average developer is in fact 10X productive than before. It just doesn't feel so crazy because it can about in discrete 5% boosts.

I also disagree with the dataset being a good indicator of productivity. I wouldn't actually suspect the number of packages or the frequency of updates to track closely with productivity. My first order guess would that AI would actually be deflationary. Why spend the time to open source something that AI can gen up for anyone on a case by case basis specific to the project. it takes a certain level of dedication and passion for a person to open source a project and if the AI just made it for them, then they haven't actually made the investment of their time and effort to make them feel justified in publishing the package.

The metrics I would expect to go up are actually the size of codebases, the number of forks of projects that create hyper customized versions of tools and libraries, and other metrics like that.

Overall, I'd predict AI is deflationary on the number of products that exist. If AI removes the friction involved with just making a custom solution, then the amount of demand for middleman software should actually fall as products vertically integrate and reduce dependencies.

I have published 4 open source projects thanks to the productivity boost from AI. No apps though, just things I needed in my line of work.

But I have been absolutely flooded with trailers for new and upcoming indie games. And at least one indie developer has admitted that certain parts of their game had used the aide of AI.

I also noticed sometimes when I think of writing something, I ask AI first if it exists, and AI throws up some link and when I check the link it says "made with <some AI>".

So I'm not sure what author is trying to say here but I definitely feel like I am noticing a rise in software output due to AI.

But with that said, I also am noticing the burden of taking care of those open source projects. Sometimes it feels like I took on a 2nd job.

I think a lot of software is being produced with AI and going unnoticed, they don't all end up on the front page of HN for harassing developers.

Claude Code was released for general use in May 2025. It's only March.

Also using PyPI as a benchmark is incredibly myopic. Github's 2025 Octoverse[0] is more informative. In that report, you can see a clear inflection point in total users[1] and total open source contributions[2].

The report also notes:

> In 2025, 81.5% of contributions happened in private repositories, while 63% of all repositories were public

[0]: https://github.blog/news-insights/octoverse/octoverse-a-new-...

[1]: https://github.blog/wp-content/uploads/2025/10/octoverse-202...

[2]: https://github.blog/wp-content/uploads/2025/10/octoverse-202...

Please, be patient. Wrangling AI agents, writing and rewriting prompts, waiting for the start of another month because tokens ran out - there are so many challenges here, you cannot expect everyone to ship an app a day or something.

The thesis has it backwards. We will see fewer published/downloaded apps/packages as people rely on others less. I'm not sure we're quite there yet but I'm increasingly likely to spend a few minutes giving an LLM a chance to make a tool I need instead of sifting through sketchy and dodgy websites for some slightly obscure functionality. I use fewer ad-heavy sites that for converting a one text file format to another.

Personally, I see the paid or adware software market shrinking, not growing, as a testament to the success of LLMs in coding.

Is this the best way to measure this? I think the biggest adopters of AI coding has been companies who are building features on existing apps, not building new apps entirely. Wouldn't it make more sense looking at how quickly teams are able to build and ship within companies?

It seems like all tech executives are saying they are seeing big increases in productivity among engineering teams. Of course everyone says they're just [hyping, excusing layoffs, overhired in 2020, etc], but this would be the most relevant metric to look at I think.

Thoughts: 1. Some hype-types may have been effusive about AI-assisted coding since ChatGPT, but IMO the commonly agreed paradigm shift was claude code, and especially 4.5, very very recent. 2. Anchoring biases in reaction to hype is still letting one's perspective be defined by hype. Yes the cursor post is a joke, but leading with that is a strawman. This article does not aim to take it's subject seriously, IMO. 3. While I agree the hype is currently at comical levels, the utility of the current LLMs is obvious, and reasons for "skilled" usage not being easily quantifiable are also obvious.

IE, using agents to iterate through many possible approaches, spike out migrations, etc might save a project a year of misadventures, re-designs, etc, but that productivity gain _subtracts_ the intermediate versions that _didn't_ end up being shipped.

As others have mentioned, I think yak-shaving is now way more automated. IE, If I want to take a new terminal for a spin, throw together a devtool to help me think about a specific problem better, etc, I can do it with very low friction. So "personal" productivity is way higher.

The models are not well trained on bringing products to market.

And even “product engineers” often do not have experience going from zero to post sales support on a saas on their own.

It is a skill set of its own to make product decisions and not only release but stick with it after the thing is not immediately successful.

The ability to get some other idea going quickly with AI actually works against the habits needed to tough through the valley(s).

I won't make any claims as to the Python ecosystem and why there is no effect seen here (and I suppose no effect seen of the Internet on productivity) but one thing that is entirely normal for me now is that I never see the need to open-source anything. I also don't use many new open-source projects. I can usually command Claude Code to build a highly idiosyncratic thing of greater utility. The README.md is a good source of feature inspiration but there are many packages I simply don't bother using any more.

Besides, it's working for me. If it isn't working for others I don't want to convince them of anything. I do want to hear from other people for whom it's working, though, so I'm happy to share when things work for me.