ProtoPRO

A tool for non-technical software creators to create custom working apps in minutes.

Comment

Inspiration

Every software development tool that can build truly custom applications requires technical knowledge. I am not an engineer or developer, I am a recruiter. When trying to build the recruiting products my industry has been missing for a long time I found the workflow of copying code back and forth between GPT and an IDE cumbersome and left a lot of room for error. I also found out how feeding in certain context into the model at the right times improves its output by a very meaningful amount and opens up the ability for non-technicals like me to build extraordinary things. To make my life easier I decided to build a tool that abstracts away these problems in a way that makes sense for me as a non-technical. As I was building it I realized two things: 1) the time is finally here for the no-code promise to come true, and 2) it's going to take non-technical people to build software development tools for non-technical people.

What it does

  • Integrate the tools needed to enable software creators to build quickly (AI, up-to-date developer documentation, web preview, cloud IDE)
  • Package it up in a delightful way that offers creators a simple, fast, and seamless user experience
  • Build features that allow creators to seamlessly provide more helpful context to the AI to build complicated custom apps

How we built it

Prompting GPT and copying the code to replit. Once I had an extension that allowed gpt to read/write to my replit directly my developer productivity 10x'ed. No more copying and pasting, just prompting

Challenges I ran into

  • GPT slows down significantly around noon, slowing down developer speed
  • Hacking together a working product without building my own IDE infrastructure
  • Building a desktop app that in an environment GPT did not have access to
  • Trying to replicate what I built the "proper" way, collaborating with real engineers, that ultimately did not yield the desired result. We had to split up. I used none of the code they wrote.

Accomplishments that we're proud of

  • The fact that I have a working app that seriously makes developing apps easier for people who can't write a print hello world statement (I still don't know the syntax)

What we learned

  • Building things the proper way is great but not for a hackathon. You've got to build a product that works first and foremost. Nobody will care about the backend code.
  • There is a platform risk when GPT slows down. My superpowers are tied to this product and I must diversify intelligence sources once possible. Hoping Bard extensions come out soon!

What's next for ProtoPRO

  • Build out the authentication logic so that I can distribute this app to everyone
  • Build more logic that quietly feeds the appropriate context into the models on behalf of the user so they get better results without them having to think about it. Like magic.
  • ProtoPRO will play a crucial role as it will be my main tool for developing recruiting apps
Share this project:

Updates