Local, customizable, open-sourced role-play app built for VisionOS.

My recent experiences with emotional chatbots revealed a gap between the comfort they offer and the privacy they compromise. Despite their ability to meet emotional needs, data security in cloud-based solutions was never explicitly confirmed. Local alternatives, while secure, fell short in delivering an immersive experience due to their text-only interfaces.

Inspired to enhance chatbot interactions, I envisioned integrating 3D human models for a more life-like experience. VisionOS is chosen as the platform for its spatial computing capabilities, envisioning a chatbot that feels like conversing with an ideal companion in any scenario, whether walking or traveling, amidst multiple active windows. Despite VisionOS's niche status, its forseeable growth under Apple marks the potential of this project.

The aim is to craft a local, private chatbot beyond typical digital companions, using a customizable 3D avatar with user-defined personality that interacts in real-world environments, such as walking on the street besides you. Envision your ideal VTuber being able to accompany and converse with you in any setting about any topic with responses tailored to your preferred tone and style. It works even when you are on an airplane.

If you're passionate about privacy, emotional AI, or 3D modeling and are excited by the idea of creating a more immersive and secure digital companion, I invite you to join this innovative project by either emailing [email protected] or submitting a pull request. Together, we can redefine the future of personal chatbots.

In short, this project aims to achieve:

1. Local, customized role-play chatbot for privacy and internet independency.
2. Personalized 3D human model that interacts with the environment to add immersiveness.
3. Digital companion that appears at any setting, no matter what you are doing and where you are.

• Local llama.cpp for VisionOS.
• Customized character card + prompt engineering.
• First-gen UI with photo + chat.
• Customized 3D Human Generator.
• 3D Human with mouth changes when speaking.
• Second-gen UI with 3D model + chat.
• 3D Human with interaction to the surroundings.

... More to come