Flex – transpile natural language to a programming language
Flex [1] is a transpiler that converts a series of statements in natural language to either Python, Java or C++, and is trained using Rasa NLU [2], an open-source framework usually used for training chatbots and voice assistants. Flex is a project I made in college in 2018 along with two of my classmates, Gourav and Sanjay. The initial idea was to create a voice assistant you could speak to that wrote code for you, similar to GitHub's upcoming (and a lot more advanced) project "Hey, Github" [3]. We then had to reduce the project's scope to complete it in a single semester while writing…
In plain words
Flex is a transpiler that converts natural language statements into Python, Java, or C++ code. Built using Rasa NLU, an open-source framework typically used for chatbots and voice assistants, it takes text files written in plain language as input and generates valid executable programs. The tool was created as a college project designed with modularity to allow future expansion to additional programming languages.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Flex [1] is a transpiler that converts a series of statements in natural language to either Python, Java or C++, and is trained using Rasa NLU [2], an open-source framework usually used for training chatbots and voice assistants. Flex is a project I made in college in 2018 along with two of my classmates, Gourav and Sanjay. The initial idea was to create a voice assistant you could speak to that wrote code for you, similar to GitHub's upcoming (and a lot more advanced) project "Hey, Github" [3]. We then had to reduce the project's scope to complete it in a single semester while writing exams, preparing for job interviews and submitting reports and assignments. So we settled on creating a Python program that takes a file containing statements written in natural language as input and transpiles them to a valid Python/Java/C++ program. The project design is modular, so adding more languages and statement types is easy. The next step we had thought of was to hook it up with some UI with voice input to have some kind of voice-driven IDE, but then as college was over and we got jobs, the project got abandoned as we moved on to other things. Seeing the "Hey, GitHub" project on HackerNews today [4] reminded me of this project, so I just wanted to share it to inspire others to fork it and make something cool. Also looking forward to some feedback on how the project could be improved to make it more useful. [1]: https://github.com/Flex-lang/transpiler. Not to be confused with FLEX, a strictly-typed programming language with the same name (which was released first, so we should have used a different name!) [2]: https://github.com/rasahq/rasa [3]: https://githubnext.com/projects/hey-github/ [4]: https://news.ycombinator.com/item?id=33543946
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 27d ago · cactuscompute.com
Launched alongside, November 2022
the whole month →
- TC
I made a thing! In 2014, I was holding a stack of iPhones and thought to myself: "Hey, if I had each phone display a playing card, I could click a button and they'd shuffle themselves" I pared that idea all the way down to this: trading cards made of e-ink displays. Right now, each card costs me about $20 each, but with only a bit more scale, I think I can get that down to $10. In doing this project, I learned how to design electronics and circuit boards. I learned Rust and wrote my first driver, I upped my CAD skills, 3D printed, and did my first resin casting. I generated the images on the…
Dev tools · 2022 · wyldcard.io
