slAIde | Devpost

Inspiration

With my teams recent experience in hackathons like ShellHacks and SaseHack, we all felt motivated to try to make a site using Google's ADK and A2A given the large influence that AI seemed to have at those Hackathons in regards to Challenges. As we discussed our ideas for an application of these tools, we landed on the idea of PDF and Presentation generation, agreeing that a site like this would save students and professionals hours.

What it does

Users can input a paragraph, story, summary, etc. and ask the agent to generate either a presentation or a PDF. With the click of a button (and a short wait), slAIde will produce a polished result for the user!

How we built it

First, we used Python to setup a Sequential agent, calling a series of agents to receive text, translate it to slide topics, form a theme for the slideshow, add bullet points to each topic, and finally, generate a fully functional HTML slideshow. As we did this, we also developed a front end for the application,

Challenges we ran into

Most of the challenges that we ran into had to do, in a big part, to the time constraint forced on us due to the nature of hackathons and with how we decided to push ourselves by experimenting with many different and new technologies. For example, the process of implementing A2A/ADK into our app took our team upwards of 24 hours due to the limited documentation available to us since it's such a new concept.

Accomplishments that we're proud of

IT WORKS!!!! but in all seriousness as a team we are seriously proud to have managed to implement not just ADK but also A2A into the generation of our output. We're also quite proud of how we managed to utilize the users microphone as an input for what the user wishes to generate.

What we learned

Some of the things we learned include how ADK and A2A work as a technology along with how to implement it effectively in order to cohesively complete a task together. Another new and interesting technology we learned to use and implement would be speech-recognition which we utilized as an input for generation. We also became more familiar with react and tailwind CSS, so much so that we were able to implement and design, in our opinion, a pretty nice user interface.

What's next for slAIde

Our team actually has many ideas we wish to implement into the app but were unable to due to the time constraint such as the ability to drag and drop mp3/mp4 files into a drop zone that would then transcribe and parse the text from the audio to then generate whatever the user wished for.