José Valim, creator of Elixir and founder of Dashbit, sits down with Victoria Melnikova to talk about how following curiosity over market trends led him to build one of the most loved programming languages. He shares why he made technical decisions over adoption-friendly ones, how decentralizing the community let Elixir reach domains he never imagined, and what he’s building with Tidewave — higher-level AI tools that understand web frameworks, not just code.
Companies using Elixir: Discord, Remote, Supabase, Fly.io, Apple, Toyota, BBC, PepsiCo, Mozilla.
What we talked about
Building a language from curiosity, not a business plan
José didn’t set out to create a programming language. He wanted to understand how languages work because he didn’t have a CS background and thought it would be useful. His first prototype was horrible, and he shelved it for eight months. When he came back to it, he asked his co-founders at Plataformatec if he could work on it part-time — with the explicit caveat that it might lead to nothing. His philosophy: if the learning and the fun are there regardless of the outcome, the risk is worth taking.
Why technical decisions should come before adoption
José chose to build Elixir on the Erlang VM — a battle-tested runtime nearly four decades old — because it was the best technical foundation, not because it was trendy. He says chasing adoption leads to bad decisions because adoption is a moving target. When Elixir launched, functional programming was a hot topic. Today it isn’t, but the technical decisions still hold up. His rule: “If all we have is opinions, I prefer mine.” Show developers the trade-offs, not a sales pitch.
Decentralizing Elixir: the only way to compete without Google’s resources
When Elixir emerged around 2012, Go had Google, Swift had Apple, and Rust had Mozilla — all with orders of magnitude more resources. José’s answer was radical decentralization. He designed the language to be extensible so communities could take it into web, ML, embedded, and data processing without depending on him. Today he has no idea how some of those domains work, and that’s the point. His job is to not get in the way.
Phoenix and the power of marketing to developers
José knew from his Ruby on Rails background that marketing matters — the “build it and they will come” mindset doesn’t work. He actively reached out to developers and CTOs, spoke at events, and attracted influential early adopters like Dave Thomas, who wrote a book on Elixir. Phoenix, the web framework built by Chris McCord, was a pivotal moment: it was the first framework that asked “what if we leverage everything this technology can do?” instead of just copying ideas from other ecosystems.
How Dashbit’s consulting feeds back into open source
Dashbit runs an Elixir development subscription where companies can ask questions about applying Elixir to their specific domains. But José doesn’t just answer — he updates the documentation instead, so the next person finds the answer already there. Client questions reveal friction points that turn into new open source projects, documentation improvements, or language features. The built-in code formatter, for example, came from teams fighting over code style.
Tidewave: AI tools that understand frameworks, not just code
José started Tidewave because he saw a gap nobody was filling: AI coding tools work inside the editor and understand code, but development is more than code. If a user experience is broken, pointing at a page element and asking “why is this happening?” is more natural than translating the problem into code terms for an AI. He wants higher-level tools specific to web development, game engines, and other verticals. He built Tidewave after Anthropic released MCP, which finally gave him the building blocks to do it himself. His vision: developers who work locally deserve the same great experience as cloud-hosted tools like Bolt.new and phoenix.new.
