slAIds

Inspiration

Nowadays, AI generated content is changing the market, peoples ways of working and accessing information. We as a team, thought that it would be interesting to come up with a project integrating several of this technologies for a purpose. In this case we finally decided to create a Web App that generates a structured presentation, with AI generated text and images, from a simple title and a little description.

What it does

With this project you can deploy a Web App to generate presentations about any subject, with variable length, structured in a FrontPage with the title of the presentation, an index and the number of slides selected. It also supports user management, signUp, login and viewing the historic of the presentations generated. We have also a beautiful frontend, to generate the presentations and seeing them directly, slide by slide, once they are generated.

How we built it

To build this project we used a backend in java and spring with a PostgreSQL database. This backend connects to a dockerized python service that conect to the reversed engineered ChatGPT API, on top of which we built an API for the purpose of our WebApp. For the image generation we use a dockerized version of StableDiffusion that exposes an API that we call from the backend.

In the backend we also use Apache POI, to generate the presentation and saving it in .pptx and .pdf.

For the frontend we used Next, css, sass built and designed by our more expert frontend developers. We tried to build a modern and minimalist fronted,and very easy and intuitive to use. With a very polished animations and work in the transitions.

Challenges we ran into

One of the challenges we found was to integrate all this different technologies in a single project, and projects already built by other developers. We had to read a lot of documentation and sometimes rely on trial and error. Another difficulty we step into was conecting all the different parts of the system we developed separatedly. But, he hardest challenge we faced, was to build all of this in such a limited amount of time, and with the pressure of the deadline.

Accomplishments that we're proud of

We are proud of having such a complex system working, in this limited amount of time.

What we learned

We learned a lot of new technologies, libraries and technics of prompt generation. Also it gave us a lot of knowledge about good practices for teamwork.

What's next for slAIds

First of all there are a lot of parts of this system we should improve, and a lot of error handling left to do =). Also we would like to add new features, like image replacing, slide regeneration, introducing the index by the user instead of generating it via AI,etc. And one of the features we want to improve is the prompt generation to generate better images for the presentations.