Before we jump into things, there’s a been a big point of contention lately in both the tech and financial communities.  There has been a lot of skepticism around whether or not AI will have any lasting impacts or will ever be able to justify its cost.  My honest belief is that anyone who thinks AI won’t change the world in ways we can’t even conceive right now hasn’t truly used current models to their full potential.  

To any doubters (especially non-technical doubters), I strongly encourage you to download Claude Code, watch some tutorials, and spend a few hours tinkering with it.  If you aren’t absolutely flabergasted, then I guess I owe you a drink!

Now, with that out of the way, let’s jump into things:

As many of us know, AI is changing the world in ways that were inconceivable just a few years ago.  The sector that’s seen one of the highest adoption rates has been coding/software engineering.  Up until very recently, AI code has largely been written off as “slop”, but recent improvements in the underlying models and their harnesses have led to huge breakthroughs in the software engineering space.  

The genesis of this breakthrough was Claude 4.5 Opus (and very recently, Opus 4.6, which is a considerably more capable model), and recent updates to the most popular harness for it, Claude Code, the CLI tool created by Anthropic last February.  

Although this model certainly is not a replacement for developers, I’ve spoken to quite a few developers, and it has been the “ah-ha” moment for many of them.  The release of 4.5 Opus has led more seasoned developers to take a more hands-off approach to development, acting more as a manager of Claude than someone who is actively writing software every day.

This development has been flying under the radar of Mr. Market since late November/early December, when Opus 4.5 was released, until very recently.  And chances are, it has flown under your radar as well, unless you keep up-to-date with all of the latest AI news.  However, this isn’t your typical AI hype.  There’s been a drastic increase in the amount of code being shipped.  If you don’t believe me, check out these graphs that a16z published showing the number of iOS apps being shipped: 

Although we saw a couple of names pop in a huge way with the very rapid adoption of Clawdbot (or Moltbot), an agentic system that allows you to control Claude Code and other AI tools from chat apps, like Telegram, WhatsApp, and Slack, we’ve since seen a HUGE sell off across the entire tech sector.  

While I think this sell-off is largely unwarranted, because many of these companies do have moats, there are a handful of names that likely won’t be hurt by AI - instead, they’ll be helped in a huge way.  Why's that?  Well, a large number of people building these apps are either non-technical or semi-technical.  Many of them have never built an entire app on their own, so they’ll be looking for suggestions.  

Who are they going to ask for suggestions?  The very AI tools that they’re using to build the apps.

Since people are more likely to take suggestions from their AI tools, we devised a relatively simple way to identify which tools are likely to benefit from the rise of vibe coding.

First, we asked Gemini 3.0 Flash to generate a list of 100 unique business ideas that could be vibe-coded.  Once we had our list, we then asked each of the leading AI models, GPT-5.2, Gemini 3.0 Pro, and Claude Sonnet 4.5 (we used Sonnet as it’s API costs are quite a bit cheaper than Opus and found in test runs that it generates similar responses) to come up with an end-to-end stack of what they would use to build out this idea.  We had our “AI cofounders” spec everything from the programming languages and hosting to be used, to how we should market our vibe-coded apps.

From there, we parsed out the recommended products and companies and organized them into a handful of charts below.  We’ve organized them by the number of recommendations per model and as an aggregate.  It’s interesting to see clear biases, with each model having “favorite” companies and tools.

Once we identified which companies/tools the models favored, we examined the download statistics for each company's popular software packages.  We did this using npmtrends.com. This site tracks npm downloads using a freely available API provided by node js.  

For the non-techies in the crowd, npm downloads are effectively Google Trends for a given tool's developer usage.  A spike in downloads of a certain package equates to a spike in usage.  We’ll have a handful of these charts for the individual company deep dives. 

For anyone looking to run this experiment themselves, doing so takes a good bit of time and racks up more than $15 in API costs, so you might as well subscribe to Sentinalysis Premium to just skip to the results, right?