CompNav · Open-source projects
Soul of Waifu
A Windows desktop AI character app where one character can join chats, voice conversations, RPG stories, and desktop companionship.
What it is
Soul of Waifu extends character chat across several connected experiences. Create or import characters for text and voice conversations, play tabletop-style roleplaying stories hosted by an AI game master, or keep a character on the Windows desktop as a companion. The project also includes persistent memory, character-asset management, and multiple model connections, making it a desktop suite for custom characters rather than a single chat window.
Core mechanisms and capabilities
Create or import characters and chat by text or voice with Live2D or VRM avatars; character data and conversation memory can carry across modes.
Soul Stage provides tabletop RPG sessions hosted by an AI game master, with dice rolls, skill checks, story objectives, and relationships that change with the narrative.
A desktop companion can stay on screen and, with user approval, interact with the mouse and keyboard, browse the web, or use tools; these capabilities require attention to permissions and confirmation prompts.
Editorial take
Soul of Waifu stands out by carrying characters across chat, RPG stories, and desktop-companion modes instead of treating each as a separate app. It provides Windows installers and local-model management, which may appeal to people who want the character experience on their own computer. Its broad feature set and hardware needs are also significant: local models, voice, and 3D assets use resources, desktop-agent features involve computer-access permissions, and hosted models may cost extra.
Who it suits
Suited to people who want to spend time with custom AI characters on Windows and are interested in roleplay stories, voice, desktop avatars, or local models.
Deployment and technical requirements
The project provides a Windows archive and installer script. After launch, import or create a character, then choose a hosted model or download a local GGUF model through the model hub. Hosted mode mainly requires provider credentials; local mode also requires model downloads and a suitable inference backend.
Runtime requirements
The stated minimum is Windows 10/11, 8 GB of RAM, and about 10 GB of free storage. Local models are recommended with at least 16 GB of RAM and a GPU with 6 GB or more of VRAM. Hosted models do not require a high-end GPU but still need API-key setup.
Soul of Waifu can use local GGUF models or connect to hosted services such as OpenAI, Claude, and Gemini. Local use requires downloading a model and choosing a quantization and inference backend that fit the hardware; hosted use requires provider credentials and may incur provider charges. The project's local-first description does not mean every model is preinstalled or will run smoothly on every computer.
Limitations and considerations
Windows is the primary supported platform; local-model speed and model choices depend on the computer's hardware. Desktop-agent features can interact with the operating system or run tools, so review their permission and approval flows carefully. The code uses GPL-3.0, while Live2D, VRM, voice, and other character assets may have separate licenses.
Official sources and further reading
Information checked: Oct 26, 2026.