Exploring Local-First AI with Evelyn: A Journey of Innovation
Hi LinkedIn Community! I’m Ted, and during my recovery from surgery, I decided to transform my passion for AI into a tangible project. Presenting Evelyn, an AI assistant designed to run entirely on my own machine.
Key Features of Evelyn:
- Fully operational on macOS (tested on Mac Mini M4 Pro/Macbook Pro M1)
- Utilizes Whisper for accurate transcription
- Connects to open-source LLMs via LM Studio
- Generates real-time speech with advanced fallback layers (XTTS → ElevenLabs → macOS TTS)
- Maintains a simple JSON-based memory across sessions
- Routes queries between local and external models seamlessly
This project is not a commercial product; it’s an early exploration of local-first AI in a scalable world. I’m eager for feedback on the technical approach and the potential of local-first models.
🔗 Watch the demo here: YouTube Demo
Let’s connect! I’m looking forward to hearing your thoughts and insights. Please share this post and join the discussion!
