Inspiration
In the Nvidia agentic AI tutorials I saw how agents were able to search the web on their own and be able to reason and draw conclusions from what they found and I wanted to build something similar.
What it does
It's a web based application where the user will enter what company/stock their looking at and Freddy will search for the current price, how its changed in the past, and what events are currently affecting its price.
How we built it
I used Nvidia and Tavily API for the models and internet searches respectively. For the application part I used Claude to help me write the HTML and JavaScript files to run in browser.
Challenges we ran into
The agent didn't use the stock the user selected and would instead choose a stock at random.
Accomplishments that we're proud of
I'm proud of the fact I was able to build a working project in 24 hours.
What we learned
I learned about the basics of using agentic AI in software. I came into this weekend knowing almost nothing of the implementation and development of AI in software. I learned about agent work flows and the ReAct pattern.
What's next for Freddy Financial
There's still a lot of bugs that need to be fixed and the agent still likes to misbehave sometimes, but one of my future goals is for my project to display graphs and be able to research multiple companies at once or even an entire industry.
Log in or sign up for Devpost to join the conversation.