AI Legacy Code Solution

Inspiration

Sooner or later someone had to come up with a way to refactor and upgrade code written by old school developers. We present a way to do that.

What it does

Given a directory of legacy code files, based on an interactive way of communicating with the user, our project is able to perform a variety of tasks involving refactoring, upgrading or even converting code to use other technologies.

How we built it

In short, we deployed language models to perform agent tasks. We simulated the work of a small crew that consists of Solution Architect, Tester and Software Engineer. The agents communicate with both the user and the language models seek to satisfy the requests of the user.

Challenges we ran into

Main challenges we went through involved API requests. For instance, finding a way to include a Language Model into crewAI or dealing with Cross Origin Requests to manage the back-end, front-end and language model requests.

Accomplishments that we're proud of

Considering how hopeful we were of solving the issues with the requests and wasting quite some time, we still managed to come with a great project and that's a great accomplishment for us.

What we learned

We learned how to be patient.

What's next for AI Legacy Code Solution

There is always room for improvement. Better organisation, better deployment, using more advanced tools and integrating it into companies with Legacy Code.