Pythagora - First complete AI developer that builds apps by talking to you

TL;DR Pythagora is not a coding assistant or an autocomplete. It’s a true AI developer that takes your requirements, asks you questions, considers what technologies to use, and creates the development plan. Then, it breaks the plan into tasks, writes code, reads the CLI, fixes bugs, iterates, reviews the code, and asks you questions to make sure it’s on the right track.

🛤 OUR JOURNEY

Senko and Zvonimir met at a startup event in 2016. when Senko was looking for someone who would lead a side project that he started. Zvonimir accepted the challenge, we turned that project into a company, and 5 years later, in 2021, we sold the company to Miro.

Leon and Zvonimir have been best friends since starting college in 2009. After the Miro acquisition, we started learning ML together and playing with different ideas. In 2023, we created our first open-source project for generating automated tests, and it quickly gained 1.4k Github stars. After talking to users and observing how AI is evolving, we decided to pivot to GPT Pilot, which became one of the fastest-growing open-source projects, now with more than 22k Github stars.

While we worked on the GPT Pilot, Senko explored a similar idea, so finally, in October 2023, we decided to join forces and build Pythagora together.

❌ The Problem

Ok, let’s state the obvious - there’s still a huge demand for software development, which is still expensive. On the other hand, we have LLMs that are extremely good at writing code. They can write code snippets like a senior developer but if you ask them to build the full app, it will do a worse job than a junior developer. This is because building production-ready apps is not only writing code. It’s writing code, running it, debugging, iterating, changing requirements, refactoring, and many other mechanisms that developers do every day.

While LLMs are getting smarter every single day, they will always be answering machines—they won’t have the mechanics that allow humans to iteratively progress on building an app.

The good thing is that all these mechanisms are deterministic and repetitive.

🌟 Our Solution

Pythagora is a blueprint of what app development looks like. First, it takes your requirements and asks you questions to clarify how you want your app to look. Then, it thinks about what technologies should be used for your specific app and creates the development plan. Then, it breaks the plan into tasks, writes code, reads the CLI, fixes bugs, iterates, reviews the code, and refactors it. Before making each commit, it asks you questions to make sure that the app works exactly as you want. Finally, it proceeds this way to the next task until the entire app is finished.

Many AI copilots help developers write code faster, but Pythagora goes beyond the current developer experience — we’re reimagining how software development is done.

🙏 Asks

If you know any company that’s looking to build internal tools, it would be amazing if you could connect us. I’d be happy to demo Pythagora to them.

Also, it would be amazing if you could share this launch on Linkedin or Twitter.

Finally, join us on Discord!