MemaTask

A smart task management app with AI-powered productivity tools. Create tasks with voice commands, work offline seamlessly, and sync across all devices in real-time. Your personal AI task assistant.

Comment

MemaTask — AI-Powered Task Management, Done Right

Inspiration

Managing tasks shouldn’t feel like a chore.

We were frustrated with task apps that felt disconnected—either too simple to be useful or so complex they became another burden. The spark came when we realized AI could bridge this gap.

We asked: What if your task manager actually understood your work and helped you prioritize intelligently?

MemaTask was born to make productivity feel effortless—natural input, smart insights, and a reliable offline experience.

What It Does

MemaTask is a smart task management app that combines AI intelligence with practical productivity features:

Voice-powered task creation Speak naturally and MemaTask understands. Example: “Remind me to call mom tomorrow at 5pm” → becomes a task with the correct due date and reminder.

AI daily summaries (Google Gemini) Intelligent overviews of your day highlighting what needs attention most.

Smart prioritization AI analyzes deadlines, dependencies, and usage patterns to suggest priority adjustments.

Task breakdown Big projects become manageable—AI can split tasks into clear subtasks.

Offline-first design Full functionality without internet. Changes sync automatically when you’re back online.

Real-time cross-device sync (Serverpod) Instant synchronization across devices using Serverpod.

Productivity analytics Track completion trends and receive personalized insights to improve your workflow.

How We Built It

Frontend

  • Flutter (Android, iOS, Web)
  • Provider (state management)
  • Hive (fast local storage)

Backend

  • Serverpod (Dart backend)
  • Auto-generated type-safe client code
  • Deployed on Serverpod Cloud
  • PostgreSQL database

AI Integration

Powered by the Google Gemini API with a robust fallback chain:

  • gemini-2.5-flash
  • gemini-2.5-flash-lite
  • gemma-3-27b
  • gemma-3-4b

Architecture Highlights

  • TaskManager: core business logic + offline queue management
  • ReminderEngine: priority-based smart notifications (with location awareness)
  • CacheManager: Hive-based local persistence for offline support
  • NetworkConnectivityService: detects online/offline state and syncs automatically
  • GeminiAIService: 1,400+ lines handling AI calls, retries, caching, and fallback logic

Challenges We Ran Into

Offline-first complexity Resolving conflicts when users edit tasks offline and later reconnect was tricky.

Our approach:

  • queued operations while offline
  • conflict detection on sync
  • last-write-wins as the default strategy (with a resolution path when needed)

AI rate limits We hit Gemini limits during development, so we implemented retries with exponential backoff (with a max wait cap) and automatic model fallback to keep the experience smooth.

Voice input parsing Natural language is messy. Turning “Call dentist next Tuesday afternoon” into structured data required careful prompt design, validation, and graceful fallback handling.

Cross-platform notifications Reliable notifications across Android, iOS, and Web required consistent Firebase Cloud Messaging setup and careful platform-specific testing.

Accomplishments We’re Proud Of

  • Zero-downtime offline experience—users can’t tell when they lose connection
  • Sub-second AI responses using optimized prompts and caching
  • Comprehensive test coverage (50+ test files across services, managers, models, and widgets)
  • Production deployment running on Serverpod Cloud
  • Clean architecture with clear separation of UI, business logic, and data layers
  • Graceful degradation: AI improves the app but is never required; the core app works without an API key

What We Learned

  • Serverpod is powerful—type-safe generation and built-in auth saved weeks of work
  • Offline-first needs upfront planning—retrofits are painful
  • Prompt engineering is an art—small changes can massively affect quality
  • Flutter is mature—great ecosystem for speech, notifications, and storage
  • Testing pays off—our test suite caught countless edge cases early

What’s Next for MemaTask

  • Collaborative tasks: share projects with real-time collaboration
  • Calendar integration: two-way sync with Google Calendar and Apple Calendar
  • Recurring task intelligence: AI learns patterns and suggests recurring tasks
  • Widget support: home screen widgets for quick capture and overview
  • Wearables: Apple Watch and Wear OS support
  • Advanced analytics: weekly/monthly productivity reports and goal tracking
  • Custom AI personas: choose how the AI communicates (professional, friendly, motivational)

Built With

Share this project:

Updates