Alembic - AI powered Workspace

Project Overview

I built Alembic and Alee, a desktop application that functions as a secure, AI workspace. My goal was to create a powerful tool for note-taking, reminders, research, MCP support, transcriptions, brainstorming and chat, leveraging the capabilities of modern AI models and the new Realtime API from OpenAI. The application is installed as a native Windows app packaged into Tauri, connecting to a cloud-based managed and secure database. The backend is a mixture of FastAPI and Agno. The app was developed with a Python backend using the FastAPI framework, which communicates with a frontend built using Svelte and Tauri and stores everything in a secure Supabase database. This combination allows for a fast, responsive, and cross-platform desktop application that will soon have a mobile app as well.

Technical Details

• Backend: The core is a Python application using FastAPI and Agno, with a cloud-based managed database - Supabase. It manages everything from user authentication, data storage, vector embedings, and personalized settings.
• Frontend: The user interface is built with Svelte and styled with TailwindCSS. It's packaged as a desktop application using Tauri, a Rust-based framework, which ensures high performance and a small application footprint.
• AI Integration: The AI system in deeply infused withing the app, taking actions for the user with simple chat commands or by speaking to the human-like voice agent. It includes services for text generation, real-time audio transcription, and creating text embeddings for semantic search within notes and documents.

Key Features

• AI Voice Assistant - Alee: A SOTA human-like voice assistant that can be used to interact with the app and take actions for the user.
• Note-Taking: A rich text editor for creating, managing, and categorizing notes.
• Voice Transcription: Real-time and simple audio transcription capabilities.
• Reminders: A reminder system that the AI model can use to set reminders for the user.
• Cross-Platform: Built with Tauri to run on Windows, macOS, and Linux.
• Customizable Settings: This is still in progress.

Challenges & Solutions

One of the main challenges was learning how to create the relationship between the fast Voice Assistant and the backend AI agents. I didn't have prior experience with Tauri, but I managed to learn the basics quickly and efficiently with the help of AI.

Current Progress

The application is fully functional with all core modules integrated and working together.

• Authentication: Secure user login.
• Database: Supabase database for storing user data and notes.
• Chat Module: Manages conversations and interacts with AI models.
• Notes Module: Full CRUD (Create, Read, Update, Delete) operations for notes with categorization.
• Transcription Services: Speak into the Mic and get real-time audio transcription.
• Backend Server: The FastAPI backend is complete, with all necessary API endpoints to support the frontend.
• AI Chat models have full CRUD operations enabled and can be used to create, read, update, and delete notes, reminders and documents per user instructions.

Future Development

• Expand the capabilities of the Voice Assistant and the backend AI Agents with MCP support.
• Enhance the note-taking and document editor with more powerful formatting options.
• Expand the Reminders module with alerts and native Windows notifications
• Create a cohesive memory layer for the AI assistant.
• Create AI automations - let the AI do the work for you on a schedule.

I'm still working on the future development ideas, but I'm very excited about the progress I've made so far. I'm also very happy with the current state of the app, and I'm looking forward to sharing it with you and making it available for download. I'll soon share the app on it's own dedicated website.