Building flexible digital companions for the next generation of cyber life.

ZuiTi is a research-driven organization focused on creating highly adaptive digital companions. We explore how expressive character systems, large language models, memory, skills, and privacy-aware interaction design can come together to make digital companions feel more natural, capable, and trustworthy.

Our work is organized around five core areas:

1. Live2D Models & Animation Design
   Designing expressive virtual characters with fluid, emotionally responsive animation.

2. LLM-Driven Action Orchestration
   Translating language, context, and intent into meaningful companion actions.

3. Memory System
   Building long-term, context-aware memory to support continuity, personalization, and deeper interaction.

4. Skills System
   Developing an evolving skill framework that allows companions to learn, adapt, and expand their capabilities over time.

5. Privacy & Permission Protection
   Designing safety-first permission models so users remain in control of their data, context, and companion behavior.

We are exploring multiple forms for digital companions:

• Pet Projection: Embodied companion experiences that can exist beyond a traditional screen.
• Web: Accessible browser-based interaction and experimentation.
• App: Mobile-first companion experiences for everyday use.
• Desktop: Persistent, productivity-aware companions for personal computing environments.

We believe digital companions should be expressive, useful, respectful, and continuously evolving. ZuiTi aims to bridge character design, AI systems, and human-centered interaction to bring cyber life closer to reality.