TOROS AI

TOROS AI is a smart desktop assistant that manages your routines, handles tasks, handles reminders and uses your computer and the internet for you. Meet TOROS AI, your ultimate virtual companion.

Comment

🚀 Inspiration

TOROS AI was inspired by a personal frustration with current AI assistants that fail to effectively manage routines and reminders. Despite their capabilities, most assistants fall short when it comes to sustained behavior reinforcement—especially in tasks like fitness and diet, which require consistency and time-awareness.

Inspired by the functionality and fluidity of Jarvis from Iron Man, I envisioned an assistant that not only converses intelligently, but also helps structure and maintain real-life habits on a desktop system—where much of our work and planning actually happens.

💡 What it does

TOROS AI is a powerful desktop-based AI assistant that enables natural conversations while offering intelligent features like:

  • Routine detection in both English and Turkish from user messages
  • Reminder scheduling with persistent, background-checked notifications
  • Real-time modal alerts without disrupting the workflow
  • Secure chat history and authentication using Supabase
  • Admin panel for testing, debugging, and managing core functionalities
  • All wrapped in a Tauri desktop app that can access native OS features

It’s not just a chatbot — it’s a productivity layer between you and your desktop.

🏗️ How we built it

  • Frontend: Built with React.js and TypeScript for a fast, modular UI
  • Backend (desktop layer): Powered by Tauri + Rust to interface directly with OS-level features and send background notifications
  • AI Conversations: Integrated with Perplexity's Sonar API for intelligent context-aware responses
  • Database & Auth: Handled by Supabase for reminders, user sessions, and chat history
  • Routine Detection: Implemented regex-based pattern matching in both English and Turkish, converting phrases to structured JSON
  • Notifications: Uses Tauri's emit and listen event system for real-time popups

🧩 Challenges we ran into

  • Multilingual routine detection: Balancing flexibility and accuracy across languages was tricky. Turkish grammar introduced edge cases we had to solve with smarter regex and time parsing logic.
  • Desktop notification synchronization: Building a background checker that doesn't affect performance while maintaining real-time accuracy took experimentation with Tauri’s event loop.
  • Time zone management: Ensuring reminders fire correctly regardless of user’s region required careful integration of timezone utilities.
  • Seamless UI: Crafting modal notifications that are helpful but non-intrusive needed thoughtful design iterations.

🏆 Accomplishments that we're proud of

  • Built a full-featured, cross-platform desktop AI assistant without relying on browser limitations
  • Created a truly usable routine/reminder system that's smarter and more reliable than anything I've used
  • Enabled multilingual support for natural task scheduling
  • Seamlessly combined a web-based UI with system-level functionality using Next.js + Tauri + Supabase + Rust
  • Built a modular, extensible foundation for future AI-enhanced productivity tools

📚 What we learned

  • How to build cross-platform desktop apps using Tauri and Rust alongside modern web technologies
  • How to design UX for persistent background systems like reminder checkers and notifications
  • The value of data modeling with JSONB in Supabase for flexible task structures
  • Practical AI prompt engineering for routine detection and task parsing
  • Importance of real-world usability — building tools we personally want to use creates better products
  • Sonar API doesn't support reminders but I manage to do that. Also little bit changed Sonar API response parser

🔮 What's next for TOROS AI

  • Speech-to-text and TTS integration for full voice command interaction
  • Calendar integration (Google/Outlook) for syncing routines and events
  • Mobile Companion App to receive notifications on-the-go
  • Plugin System so users can extend functionality
  • More AI models and custom agents tailored to user productivity styles
  • Team collaboration features for shared reminders, goals, and routines
  • Offline mode for privacy-first users
  • Habit analytics dashboard — understand patterns and productivity trends over time

Built With

Share this project:

Updates