Work · Ongoing

AI waifus and virtual assistants

Research and build digital companions and personal AI assistants that work not as a dumb chatbot but as an entity with memory, voice, character, an interface and a potential place in the daily life of a user. The work sits at the crossing of agentic systems, UX, multimodality and emotional interaction between a person and AI.

Year
2025
Role
Researched the architecture of virtual assistants and AI companions · Wrote long explainer pieces and build guides · Experimented with characters, memory, voice and visual embodiment
Client
Project: LOCI
Genres
AI · XR · Writing

The brief

Research and build digital companions and personal AI assistants that work not as a dumb chatbot but as an entity with memory, voice, character, an interface and a potential place in the daily life of a user.

Context

From the outside this topic can sound provocative or meme-like, but in essence it is a large research and applied circuit at the intersection of agentic systems, UX, multimodality and emotional interaction between a person and AI.

Translation between contexts matters here too: what one audience calls an AI waifu, another calls a virtual assistant, a personal assistant or a digital companion. The technology underneath is very nearly the same.

Solution

I did not just write about the topic — I assembled an architecture and a practical frame around it. Longreads and build guides for virtual assistants appeared, along with experiments with characters, voice, visualisation and agentic logic, and reflections on how such entities could live in an XR and WebXR environment.

According to the notes from NotebookLM, this circuit also involved more complex technological builds with memory, multimodality, voice and a visual layer. In this entry it should still be read as a direction and a practical R&D circuit rather than one finished product.

What I did personally

  • Researched the architecture of virtual assistants and AI companions
  • Wrote long explainer pieces and build guides
  • Experimented with characters, memory, voice and visual embodiment
  • Connected the theme of agentic systems with the XR and WebXR direction

Stack

  • OpenRouter
  • PydanticAI / ElizaOS
  • WebXR / VRM
  • voice and multimodal interfaces

Skills

  • AI agent design
  • UX of digital characters
  • packaging a complex topic into understandable content
  • researching new user scenarios for AI

Immediate focus

  • Separate published concepts from prototypes that were actually built
  • Add concrete links to videos, articles and demos
  • As the topic grows further, carve out a separate product layer on top of the case

Result

The case shows that I work with AI not only in applied business automation but also in a more complex human zone: memory, character, trust, an interface, a sense of presence. For the portfolio this is a strong research piece at the junction of technology, culture and new forms of interaction.

Original (Russian) →