Plan smarter. Focus deeper. Build sustainable productivity.
FlowMind is a full-stack productivity platform for students and working professionals that combines task management, focus support, habits, intelligent scheduling, time management, productivity analytics, wellbeing-aware tools, explainable recommendations, and machine-learning task-risk prediction in one integrated workspace.
Stable Release: v1.0.1
Modern productivity often requires switching between separate task managers, calendars, focus timers, habit trackers, time trackers, and wellbeing tools. FlowMind was developed to reduce that fragmentation by bringing these capabilities together and adding adaptive, explainable decision support.
The system is implemented as a real full-stack application with a Next.js frontend, FastAPI backend, PostgreSQL database, and an integrated V4 machine-learning task-risk ensemble.
- Task Management - CRUD, priorities, lists, categories, tags, subtasks, recurring tasks, reminders, favorites, board/list views, search, smart date filters, Eisenhower Matrix, calendar view, analytics, and import/export.
- Habit Tracking - recurring habits, check-ins, streaks, progress insights, recovery support, and habit-management analytics.
- Focus Sessions - Pomodoro-style sessions, configurable durations, linked tasks, breaks, session history, daily goals, streaks, adaptive focus recommendations, and post-session reflections.
- Smart Scheduling - month/week/day/agenda/timeline views, event CRUD, drag-and-drop planning, resizing, reminders, task-linked calendar blocks, conflict awareness, and explainable smart-schedule suggestions.
- Goals & Time Management - weekly goals, time tracking, time budgeting, work categories, activity timeline, and productivity planning.
- Productivity Analytics - live dashboard, productivity score, analytics hub, deep-work analytics, yearly productivity heatmap, weekly review, and personal-pattern insights.
- V4 Task-Risk Prediction - predicts the probability that a task will be completed before its deadline.
- Explainable Risk Levels - Low, Medium, and High task-risk classifications with contributing factors.
- Recommendation Engine - explainable actions based on productivity context.
- Smart Scheduling - considers deadlines, priority, estimated effort, workload, planning preferences, and task-risk context.
- Weekly AI Coach - summarizes patterns and provides practical productivity guidance.
- Personal Patterns - surfaces explainable trends from user activity.
FlowMind treats wellbeing signals as productivity context rather than medical diagnosis.
- Movement Break Coach
- 20-20-20 Eye Care
- Energy & Mental Fatigue Check-In
- Sleep Regularity
- Cognitive Load
- Distraction Log
- Anti-Procrastination Starter
- If-Then Planner
- Productivity Experiments
- Workload Warning
- Hydration & Meal Awareness
- Guided Recovery Breaks
- Life Balance
- Habit Recovery
- Secure registration and login
- Email verification
- Password recovery
- JWT access and refresh flow
- httpOnly authentication cookies
- Protected frontend routes and backend endpoints
- User-specific data isolation
- Light and dark themes
- Responsive desktop and mobile interface
- Browser notifications
- Installable Progressive Web App (PWA)
- Customizable workspace feature visibility
Dashboard
|
Tasks
|
Smart Schedule
|
Focus Sessions
|
Analytics
Looking to get started with FlowMind or explore its features in detail?
- 📖 FlowMind User Manual - Complete guide covering installation, account setup, workspace navigation, productivity features, AI task-risk prediction, Smart Scheduling, notifications, PWA installation, settings, and troubleshooting.
FlowMind includes a trained and explainable task-risk model that estimates whether a task is likely to be completed before its deadline.
The deployed model is a compact ensemble built from evaluated machine-learning approaches. Its final non-zero blend uses:
- CatBoost
- Logistic Regression
- HistGradientBoosting
The final evaluation used an untouched holdout containing unseen users, with zero user overlap between training and holdout data.
| Metric | Final Holdout Result |
|---|---|
| Accuracy | 78.89% |
| Balanced Accuracy | 78.75% |
| Macro F1 | 77.04% |
| ROC AUC | 86.69% |
| Risk Precision | 64.13% |
| Risk Recall | 78.34% |
| Risk F1 | 70.53% |
| Unseen-user overlap | 0 |
The production prediction provides:
Completion Probability
↓
Low / Medium / High Risk
↓
Explainable Contributing Factors
↓
Tasks + Dashboard + Smart Scheduling + Recommendations
The model is a productivity decision-support component, not a guarantee of future behaviour. The first model version was developed using a designed synthetic behavioural dataset because sufficient long-term real-user FlowMind data was not available during the project period.
FlowMind follows a Modular Layered Architecture inspired by Clean Architecture principles.
User
↓
Next.js / React Presentation Layer
↓
REST API
↓
FastAPI API / Controller Layer
↓
Service / Business Logic Layer
↓
AI / Analytics / Recommendation Layer
↓
Repository Layer
↓
SQLAlchemy ORM / Model Layer
↓
PostgreSQL Database
This separation keeps interface logic, business rules, AI processing, persistence, and database responsibilities maintainable and independently testable.
| Area | Technologies |
|---|---|
| Frontend | Next.js 16.3, React 19, TypeScript, Tailwind CSS, Framer Motion, Lucide React, next-themes |
| Backend | FastAPI, Python 3.12, Pydantic |
| Database | PostgreSQL, SQLAlchemy, Psycopg |
| AI / ML | Scikit-learn, CatBoost, HistGradientBoosting, Pandas, NumPy |
| Authentication | JWT, access/refresh tokens, httpOnly cookies, password hashing |
| Testing | Pytest, HTTPX, FastAPI TestClient, Vitest, Playwright, axe-core |
| Version Control | Git, GitHub |
| Development | Visual Studio Code, npm, pip |
flowmind/
├── frontend/
│ ├── app/ # Next.js routes and workspaces
│ ├── components/ # Reusable UI and feature components
│ ├── hooks/ # Frontend hooks
│ ├── lib/ # API client and utilities
│ ├── types/ # TypeScript models
│ └── public/ # Branding and PWA assets
│
├── backend/
│ ├── app/
│ │ ├── api/ # FastAPI routes
│ │ ├── services/ # Business logic
│ │ ├── repositories/ # Data-access logic
│ │ ├── models/ # SQLAlchemy models
│ │ ├── schemas/ # Pydantic schemas
│ │ ├── ai/ # ML prediction and explainability
│ │ ├── database/ # Database configuration
│ │ └── core/ # Security and application configuration
│ ├── tests/ # Backend automated tests
│ └── requirements.txt
│
├── database/
├── docs/
├── .github/
├── README.md
└── LICENSE
Install:
- Node.js
- Python 3.12+
- PostgreSQL
- Git
git clone https://github.com/Asas-Ahmed/flowmind.git
cd flowmindCreate and activate a virtual environment, then install dependencies.
cd backend
python -m venv .venvWindows:
.venv\Scripts\activatemacOS/Linux:
source .venv/bin/activateInstall packages:
pip install -r requirements.txtCreate backend/.env using backend/.env.example as the template:
DATABASE_URL=postgresql+psycopg://username:password@localhost:5432/flowmind
FRONTEND_URL=http://localhost:3000
SECRET_KEY=replace-with-a-long-random-secret
ALGORITHM=HS256
ACCESS_TOKEN_EXPIRE_MINUTES=30
REFRESH_TOKEN_EXPIRE_DAYS=7
PASSWORD_RESET_TOKEN_EXPIRE_MINUTES=15
EMAIL_VERIFICATION_TOKEN_EXPIRE_MINUTES=30
RESEND_API_KEY=re_your_api_key
EMAIL_FROM=FlowMind <onboarding@resend.dev>Never commit real secrets or API keys.
Run the backend:
uvicorn app.main:app --reloadBackend default:
http://localhost:8000
Open another terminal:
cd frontend
npm installFlowMind defaults to:
NEXT_PUBLIC_API_URL=http://localhost:8000
If a different backend URL is required, create frontend/.env.local:
NEXT_PUBLIC_API_URL=http://localhost:8000Run the frontend:
npm run devFrontend default:
http://localhost:3000
The final development stage included backend, frontend, integration, machine-learning, browser, responsive, accessibility, dependency, build, and user-acceptance testing.
| Verification Area | Result |
|---|---|
| Backend automated suite | 121 passed |
| Backend coverage | 80% |
| Frontend Vitest | 8 / 8 passed |
| Dependency audit | 0 vulnerabilities |
| Production build | Successful |
| TypeScript verification | Successful |
| User Acceptance Testing | 15 / 15 passed |
| Browser E2E suite | 122 passed, 43 failed |
| Cross-browser profiles | Chromium, Firefox, WebKit |
| Mobile profiles | Mobile Chrome, Mobile Safari |
| Accessibility | axe-based checks executed |
The remaining Playwright failures are documented as part of the project evaluation rather than hidden. They primarily form evidence for the known testing limitations and future improvement work.
FlowMind can be installed as a standalone Progressive Web App on supported devices and browsers.
The project includes:
- Web app manifest
- Application icons
- Service worker support
- Installable desktop/mobile experience
- Standalone application mode
FlowMind was developed as a final-year Software Engineering project and was supported by:
- Project proposal and feasibility analysis
- Software Requirements Specification
- Literature review
- User research questionnaire
- System architecture and UML/design diagrams
- Desktop and mobile wireframes
- Machine-learning experimentation and unseen-user evaluation
- Automated software testing
- User Acceptance Testing
- Technical, data, research, ethical, and project limitation analysis
The project evaluates FlowMind as a software artefact and decision-support system. It does not claim that short-term project evaluation proves long-term improvements in productivity, wellbeing, or behaviour change.
The project documentation includes:
- System Context Diagram
- Context-Level and Level-0 DFDs
- Intelligent Core Workflow
- Use Case Diagram
- System Architecture Diagram
- Entity Relationship Diagram
- Component Architecture
- Class Diagram
- Activity Diagrams
- Sequence Diagrams
- Desktop and mobile wireframes
- Final UI screenshots
- Testing and UAT evidence
These artefacts are maintained under the project documentation/evidence structure and support the final thesis and demonstration.
The current stable release refines the core FlowMind productivity experience, including:
- smart task date filtering;
- task-linked Focus Sessions;
- improved integration between tasks and deep work;
- continuation of previous Time Tracking activities;
- preserved Smart Scheduling, analytics, AI, PWA, and wellbeing functionality.
This release is the stable software artefact prepared for final project evaluation and demonstration.
- The first ML model is trained using synthetic behavioural data and requires future external validation with anonymised real-user data.
- AI outputs are productivity guidance and are not medical or psychological diagnosis.
- Long-term behaviour-change effectiveness was not established within the project period.
- Browser E2E/accessibility testing identified remaining issues that are documented for future improvement.
- Production-scale load and long-duration field testing remain future work.
Project: AI-Powered Adaptive Productivity and Life Management System to Enhance Productivity and Work-Life Balance for Students and Working Professionals
System: FlowMind
Module: CIS6035 - Development Project
Programme: BSc (Hons) Software Engineering