App Engine

Inspiration

I was inspired by the new release of GPT agents to create this project. Tools such as auto GPT were also largely motivational.

What it does

Classes include the Product Owner, the Manager, and the Developers. Each 'Worker' is developed on their own thread in order to maintain a conversation with their supervisor. The organization then upholds the hierarchy Product Owner -> Manager -> Developer(s). User input (product requirements) are passed to the Product Owner. This class then parses the text fits a product backlog to these requirements using a predefined format. This product backlog is then pushed to the manager(s) who create lists of tasks based on files that must be edited. Developers and files edited have a one to one relationship. These developers are agent instances themselves using the code interpreter tool. They can then both parse code and write it out to files locally in order to accomplish tasks.

How we built it

I built this project using the OpenAI Agents API and python.

Challenges we ran into

The main challenges

Accomplishments that we're proud of

I enjoy how this application can produce random and different integrated applications with from the same input, truly showing the power of modern LLMs.