Inspiration 📒

Every day, I'll carry a mini notebook with me. This contains various ideas for coding projects I would use for later, whether it would be architecture, database structures, or even code snippets for certain functions.

The idea of having that compacted down to a single app that can organize all of those ideas would be a huge productivity boost for me, along with any other developer who's constantly away from their computer who has those, "spur of the moment" ideas.

What it does 🪙

It allows developers to generate code snippets using GPT-3's API, and to store / organize those code snippets into collections for use in later coding sessions. The end users can also buy tokens for use with each code generation, with each generation using up one token.

How we built it 🛠️

My teammate and I started planning out the low fidelity wireframe to demonstrate and confirm that the feature set, outlined in the project brief, would be conformable to a phone app. From there, the project was built up using the MVVM architecture to set up an easily updatable and flexible UI system.

The overall architecture uses Firebase as the backend for managing users and their responses; along with utilizing OpenAI's API to make calls to GPT-3 with a library known as OpenAIKit.

Challenges we ran into 🏃

The main challenge for me, personally, was fully utilizing a new paradigm of UI elements for such a unique app. SwiftUI can be certainly tricky when it comes to arranging certain UI elements, and this project was no exception to the rule.

Accomplishments that we're proud of 🥳

The main accomplishment is making a fully fledged app using high-fidelity UI design trends, that fully utilizes a backend system for a user account system; along with utilizing GPT-3 to its full advantage when it comes to code completions.

What we learned 🤔

My main take away from the project was that nothing in life comes easy, especially with this project. A lot of hard work was used to fully implement a fully fledged UI, along with a fully working backend system for tracking collections and responses.

What's next for ChatGPT ⏩

Outside of the hackathon, we will begin the process of cleaning up the codebase for further implementations and future development, so that CI / CD can be fully implemented.

As well, more features, such as code compilation and execution, will be integrated into the app, so that developers can also run the needed code while on the go.

Built With

Share this project:

Updates

posted an update

We had a blast bringing DevGPT to life! Marcus and I put a lot of effort, creativity, and last but not least, our heart and soul into creating it. While we made mistakes, we also learned a lot. I would like to give a big shout out to Assembly AI for organizing this hackathon!

After this hackathon, we're aiming to add lots of cool features to DevGPT and transform it into a full-fledged AI assistant for developers. We're now on a mission to make developers' lives easier and more productive. Oftentimes, innovative and groundbreaking ideas fall out of their heads and never make it to the world. We would like to help them develop these ideas into reality.

Log in or sign up for Devpost to join the conversation.