Picking the wrong programming language for your developer product can hurt its success. 📉 Looking at this chart, I don’t understand why you don’t focus on Python and/or TypeScript when building your developer product. It are also the languages most starting developers and LLMs excel in. The language you choose for your product, like a CLI or SDK, can decide how many people use it. A new, trendy language might sound cool, but few developers rely on it daily. Most stick with what they learned early on. We’ve seen startups make this mistake by relying on things like Reason on Elixir and failing to both attract developers as users and as employees. When you need something like Rust for efficiency, it’s perfectly fine to use that under the covers But go with popular languages like TypeScript and Python on the surface to reach more developers. As your product grows, add support for others. Bigger audience, better results.
Tushar Dwivedi
5mo
Source of this data? Also, if there IS a genuine source, I wonder how they collected the data on languages used in the products they analysed. And if they compared the actual ratio of successful products in a language, over total products built in that language. Everyone and their aunt today writes code in TS & Python, because that's what the vibe coding tools these days are working with. So the sheer number of things being built in these products might skew the numbers, if not accounted for, the ratio of success/total. And did I mention, an image without a source could be from anything (even an AI prompt) In the era of AI tools writing a lot of our code, I have felt that while TypeScript really shines in controlling hallucination to an extent, Python is doing a pretty bad job. We need more strongly typed languages to call out all the silly things that AI tools are doing in our codebases (maybe not all, but many get exposed if you can just build the code beforehand). But then, this is my experience from my own work (mostly in Python) and side projects (which I have moved from Python to Golang, for the most part, for this very reason)
Shyam Chawla
5mo
Agree with this. Most dev tools struggle not because the tech is bad, but because the language choice shrinks the potential user base. TS and Python aren’t “trendy,” they’re where developers actually live day-to-day. If you want adoption, meet people where they already are.
Dan Marshall
5mo
Choosing the right tool for the job is more important than choosing the tool for the tradesman. You can always upskill devs to your Lingua Franca, and you probably want to employ people who are keen to learn, but choosing a framework or language that is a bad fit requires a rewrite. AI makes working in a foreign programming language almost trivial now too. For example, I’m working in the JVM stack using Kotlin for the first time in my career and I’m pretty much as productive as before in .NET. I literally just pop my C# code into my agent and it converts it for me.
David Nebinger
5mo
Roy the “Top five languages” based purely on number of users isn’t a meaningful metric for choosing Python or TypeScript for a developer product. Popularity ≠ suitability. Using this metric, someone might feel justified building their next desktop application in TypeScript because “Hey, lots of people use it.” That logic falls apart quickly once you look at context. The real decision criteria should be: ✅ Pick the right tool for the job ✅ Leverage the skills your existing team already has ✅ Deliver a successful, maintainable product for your actual use case The biggest failures come from choosing a language because it’s popular in other ecosystems with completely different requirements. Trendy languages don’t make good products. Good language/problem fit does.
Are you sure you are not biased as a speaker for JavaScript, TypeScript etc? This is so wrong in many ways. Yes LLMs don't excel at languages without a lot of examples on stack overflow. But the original JavaScript left pad "library" had several bugs. C# or Java are fine for many things and you can find reasonable amounts of developers for those. For some things you might need COBOL Forth or APL. Sure there are people willing to learn those languages. Personal thoughts. I did work with Python I didn't like the language. Same for JavaScript. I do think TypeScript is cool: you can run doom in the type system. But that is not a reason for business use. Java with Spring boot & Spring AI provides an easy to use solution for working with LLMs. No you cannot vibe code your solution; I tried for fun, it was entertaining. With a short tutorial you are up and running in an hour. IMHO the code was easier to read than what I saw in Python.Yes I am biased too. There are a few other usable libraries as well.
I'll just say "I agree" but don't just look at #1 and 2. Look below and in specialties. C# and Java are pulling huge numbers here for enterprise backends and this graph is based on public repos in GitHub. Most business code isn't public. Other languages like Go and Rust are used for key systems like compilers or infrastructure. Python leads in data science and JS/TS the web, but behind both are other languages with high usage. None of these are a bad choice to learn. But don't learn Rust to write a web frontend or TS to write a compiler.