-
-
Functionality of the app
Inspiration
Frustractions with OpenAI's Atlas, and Perplexities Comet. Using tools such as these, we have always noticed this lack of actually seeing what is going on in the browser. We wante dot implement overshoot into this by connected the web extension with overshoot's vlm analysis to report what is going on in the browser.
What it does
Same functionality as Comet and Atlas, basically we used overshoot to utilize the vision language model to interpret the browser and provide better context for gemini to decide on.
How we built it
Started off of a simple google side panel extension nd built it into a chatbot with a vlm attached to it thanks to overshoot.
Challenges we ran into
Overshoot was not working during the majority of the even due to networking issues on the api part.
Accomplishments that we're proud of
Executing an idea and learning to deal with major issues such as the overshoot api not working. First time working with web extensions so having to explore and try to implement a web extension was fun.
What we learned
Despite looking into all the possibility of the code being wrong, sometimes the issue is not the code but with the infrastructure.
What's next for Baz.AI
Parallelism with browser tasks and then integration with other llm models and experimentation with better context delivery to the LLM.
Log in or sign up for Devpost to join the conversation.