Mira - Your Mental Mirror

AI-powered real-time productivity monitoring agent that tracks your mood and screen activity, and provides intelligent feedback and nudges to help stay on track.

Comment

Inspiration

Frustration with traditional productivity tools that only track time without understanding context. We wanted to create an intelligent system that actually analyzes what you're doing and provides meaningful insights and nudges you, not just give raw data.

What it does

Mira is an AI-powered productivity monitor that can talk to you, capture your screen activity, analyze your mood and work patterns using computer vision and AI, and provides personalized insights and gentle nudges to optimize your workflow. It combines real-time screen monitoring with intelligent analysis to help you understand and improve your productivity habits.

Detailed workflow

Mira's Multi-Agent Workflow

Screen Capture Agent Captures screen activity and performs AI analysis. Takes screenshots, processes images to base64, sends to Gemini AI for productivity assessment, validates response with Pydantic, and stores analysis in MongoDB with timestamp and user preferences.

Orchestrator Agent Coordinates all agents and makes intelligent decisions. Manages monitoring cycles, interprets analysis results, compares productivity scores against configurable thresholds, triggers notifications when needed, and saves nudge data to database via Flask API.

Notification Agent Delivers personalized productivity nudges. Receives notification requests from Orchestrator, sends desktop notifications using Plyer library, handles 10-second timeout display, and provides graceful error handling for cross-platform compatibility.

Complete Workflow Cycle Orchestrator initiates monitoring every 60 seconds. Screen Capture Agent captures current screen and AI analyzes with Gemini. Productivity score calculated on 1-10 scale. Orchestrator evaluates against threshold and decides to send notification or continue monitoring. Notification Agent delivers personalized nudge if needed. Data saved to MongoDB and Flask API. Cycle repeats after configured interval.

Voice Agent

The Voice Agent provides natural language interaction with Mira through speech-to-text capabilities. It captures audio input from the user's microphone, transcribes the speech to text, analyzes the emotional content and context, and generates personalized productivity nudges based on the user's verbal expressions. The agent processes voice recordings in real-time, maintains a history of voice inputs with timestamps and emotional states, and integrates with the main productivity monitoring system to provide contextual recommendations. It currently uses mock transcription and AI processing for demonstration purposes, but is designed to integrate with actual speech recognition and natural language processing services for production use.

Key Integration Points Real-time WebSocket communication between Python backend and React frontend. MongoDB for persistent storage of analyses and nudges. Google Gemini API for intelligent content analysis. React dashboard for real-time productivity visualization. Flask backend for data management and agent coordination.

How we built it

Multi-agent AI system with React frontend and Python backend. We built: Screen Capture Agent: Captures and processes screen images Orchestrator Agent: Coordinates analysis and decision-making using Google's Gemini AI Notification Agent: Sends personalized productivity nudges React Dashboard: Real-time visualization of productivity metrics Flask API: Backend communication between agents and frontend

Challenges we ran into

AI model integration complexity especially with the voice agent, and creating non-intrusive notifications.

Accomplishments that we're proud of

Successfully created an intelligent productivity system that actually understands context, not just tracks time.

What we learned

Multi-agent AI architectures can effectively coordinate complex real-time tasks, and computer vision + LLM integration enables sophisticated productivity analysis. We learned how Google's Gemini API can be leveraged for contextual productivity insights. We also learned that Flask-SocketIO provides excellent real-time communication between Python backend and React frontend for live dashboard updates.

What's next for Mira - Your Mental Mirror

Enhanced AI capabilities, mobile app, and deeper productivity insights. We plan to add more sophisticated analysis features, develop a mobile companion app, integrate with calendar systems, and create more advanced productivity coaching based on user behavior patterns.

Built With

Share this project:

Updates