In a 72-hour sprint to the finish line, behold Verum!
Note: The video was very rushed, and I realized too late before the deadline that audio did not, in fact, record as expected.
Inspiration
I wanted a way for LLMs to essentially generate UI elements in real-time without actually having to deal with too much web programming. I wanted to figure out a way to leverage function calling along with easily-followable framework semantics (like grommet!) that an LLM would have an easy time replicating. I also have been needing a tool that lets me easily visualize what UI elements I want in an app. Prompting LLMs for web code is nice, but it would be nicer if there was a way to see what it produces in real time, and work on elements iteratively. Thus, Verum was born!
I also designed this thinking it would be a great complementary app to Google AI Studio :)
What it does
Uses Gemini's function calling to generate grommet-based UI components using natural language. Ask it to build a component and it will generate it for you.
How we built it
With lots of caffeine (React business in the front, Python party in the back)
Challenges we ran into
Time.
Accomplishments that we're proud of
Getting something submitted and built from scratch when I discovered this hackathon in less than a week
What we learned
How to build an app quickly that effectively utilizes function calling in large language models. Also, Gemini's capabilities for function calling is quite nice.
What's next for Verum
Cleaning it up so it can be used by the public! This was a great hackathon demonstration / portfolio builder, but it needs more work done before it's production ready. Stay tuned, this is just the beginning!
Log in or sign up for Devpost to join the conversation.