NeuroBloom

The first agent detects an emotion and runs the journaling agent to do.cument what's going on. The user can add more detail if they see fit
Journal entries themselves are saved on a txt file, the journal entries in the frontend are stratchpad ideas for the user to use with AI.
The final agent can run and open applications as needed to help the user stay on task. Finished tasks are logged in the journal scratchpad.

Inspiration

I was inspired by my personal lack of organization and anxiety about everything I needed done. A tool that could help set up that sort of system with me would be a gamechanger.

What it does

Using A2A(Agent to agent) AI frameworks, these agents detect the user's emotion, draft up journal ideas in a scratchpad, and encourage the user to start their tasks by automating the opening of applications and files the user may need.

How I built it

Using Gemini API for the agentic interface and FastAPI for the python backend server, it fits seamlessly with the beautiful React-JS frontend with tailwind css

Challenges I ran into

Learning how to use A2A, since this is a pretty new technology, I was in the dark about a lot of things and needed to figure them out as I went. Along with this, bridging the FastAPI data to react proved another major issue, one that I was able to somewhat fix after several iterations of debugging

Accomplishments that we're proud of

Creating a system of multiple agents that are capable of running eachother without human interaction(although a human interaction method is needed)

What I learned

It's better to have a complete fully fledged plan before jumping right in, because I was lost in many details executing this project, which in turn may have made it stray from the initial image I had for it