-
-
user flow diagram
Inspiration
When we think of love, we think of Moms, the ultimate managers of our chaotic lives. However, as we grow up and move away, they can’t always be there to "schedule our next doctor’s appointment" or remind us to stay under budget. We created M.O.M. to fill that gap, providing a high-performance personal intelligence agent that manages your professional or daily life with the same care and precision as a mother, ensuring you stay productive, financially healthy, and environmentally conscious.
What it does
M.O.M. is a multimodal personal operations manager that handles three critical pillars of adulting: Management (Intelligence Feed): It uses MBZUAI K2-Think-v2 reasoning model to triage your inbox, separating critical meeting invites and boss updates from the "noise" of receipts and newsletters to provide a concise daily briefing. Optimization (Scheduling Agent): A conversational chat interface extracts tasks, times, and locations from natural language (e.g., "Yoga at 6pm in the Gym") and instantly pushes them to your structured calendar. Monitoring (Carbon & Spending): It monitors your environmental impact by calculating $CO_2$ emissions for every task you perform and visualizes your financial health through sleek, categorized breakdown charts.
How we built it
Frontend: React (Vite), Recharts (Data Viz), Tailwind CSS, Lucide Icons. Backend: FastAPI (Python), SQLAlchemy (ORM), Uvicorn. AI Engine: MBZUAI K2-Think-v2 (Reasoning Model). Database: SQLite (Relational Storage).
Challenges we ran into
As this was MY first hackathon the main challenge I encountered was trying to have multiple features. Even when staying up for more than 24 hours I kept wanted to add features that would ultimately hinder the applications main features from developing. That is when I decided to create the bare necessities I wanted the application to have and then adding more features with the extra time.
Accomplishments we are proud of
We are proud of how we incorporated the K2-Think-v2 model. Using prompt engineering we were able to create not only a email parser, but a conversation scheduler, and a data analyst. I believe a project such as this can be useful not only for personal use but for professional use aswell.
What we learned
Sanitizing LLM Output: We learned that you cannot always trust an AI to follow formatting rules perfectly; building "defensive" code to clean AI strings is essential for a stable UI. Relational Mapping: This project reinforced the importance of structured data. Shifting from combined strings to specific database columns was difficult but necessary for building scalable features like location tracking. Git Conflict Resolution: We mastered the "Reflog" and rebase workflows after an accidental git pull introduced merge conflicts in our branch.
What's next for M.O.M
M.O.M might seem like an application for personal use but it's capabilities span into enterprise as well. Imagine you run a company and you open M.O.M and you have all your important emails waiting for you on the dashboard already parsed.
We would love to add more personalization. For example, using OAuth to connect directly to a user's gmail account and parsing emails directly. Or allowing user's to set personal goals to save money or stay under certain carbon emissions.
We also planned on adding more interaction. For example, gamifying our application by adding a leaderboard and having user's and their friends compete earn eco-badges or community rewards.
Log in or sign up for Devpost to join the conversation.