Inspiration
We got our inspiration by looking at our daily routines and what parts of our routine could be improved, whether it be big part of our routine or small. We were talking about how Outlook can be bothersome because it notifies you about every email and as, students' we can't turn the notifications off if case we get one that's important. We wished we could just see the emails that mattered to us and that led to the idea of creating a filter for email notifications.
What it does
A person saves keywords in a sidebar inside Gmail. It then checks the inbox search in the lsat 30 days for the keywords. Grok read the keyword, the sender, the subject, and a short snippet, then write it in a sentence of its priority.
How we built it
Cursor was used to assist us with our project. We used to help make our sidebar and to help integrate our grok AI. Grok 4.7 runs during the check and writes the sentences. The bot is started locally with the Agent SDK, and a Cloudfire tunnels lets the email connect. Google Cloud holds the email permission. The app is in the testing mode, so only our demo accounts and connect and use the application. The gmail API reads the inbox and creates the colored labels. Google Apps Script runs the sidebar, saves each account's keywords, and shows the result. The sign-in is a typescript script and the sidebar itself is Javascript.
Challenges we ran into
We had to change our original idea from outlook to gmail since we didn't have personal outlook accounts. Github took us longer than we would have liked. We also had problems commit to the main repository on github. The sidebar also brought some challenges. Integrating the Grok AI was probably the most time consuming and difficult part of project. The sidebar can only run for about 30 seconds, so it can only check a few messages and when Grok was slow, we just return simple sentences. We did this so the bot didn't just crash.
Accomplishments that we're proud of
This is our first hackathon and neither of us really have that much experience. We wanted to do something that was slightly challenging which is why we chose the mlm category with cursor and grok bot. We were able to make a project that labels emails by key words and uses ai, which determines by a script, the importance of an email.
What we learned
Cursor and Grok helped to write our code. Grok is also the AI model we used in our project. We also became more familiar with Javascript and Typescript. Making this project with gmails taught about google cloud and app scripts.
What's next for Filtr
In the future, Filtr will be implemented in all messaging services and flooded inboxes will be a thing of the past.
