entity Updated 2026-07-09 Tags: Founder, Ai, Product, Social

Tristan

Tristan is the founder of Natural Selection / 自然选择 interviewed in 135. 和自然选择创始人Tristan聊,Elys、赛博分身、灵魂、Context的获取与流动和AI社交网络. The source presents his product path as a long-running exploration of emotional value and loneliness: audio content, stranger social products, romance games, EVE, and Elys all become ways to help people feel accompanied or make better connections.

His distinctive claim is that the AI product shift is not only stronger models or new interfaces, but Proactive Agents acting with enough Context Engineering to face the world on the user’s behalf. He sees Subjectivity As AI Asset as the scarce layer: people and creators with clear values, taste, style, past work, and self-understanding can give agents better direction as intelligence becomes more widely available.

这可能才是 AI 陪伴真正该有的样子|对谈刷屏产品 EVE 创始人 Tristan sharpens the EVE side of his thesis. Tristan defines AI companionship as “super alignment” with the user’s life, then works backward into AI Companion Active Memory, emotional post-training, real-world temporal awareness, independent personality, and game progression as the product mechanisms needed to make a companion feel stable rather than merely chatty.

Key Points

  • He describes EVE as the product where prior game, content, character, and AI ideas converged into a companion-like “person.”
  • He describes Elys as a larger network opportunity: a “field” where Cyber Avatars pre-interact and discover valuable real-person connections.
  • He frames Context Flywheel as the product loop that can overcome both personal context cold start and network cold start.
  • He argues that ChatGPT and Doubao are serious competitors, but also that social-network rules, graphs, protocols, and trust cannot be supplied by a model alone.
  • He treats Character AI-style session duration as insufficient evidence of companionship, because content consumption and relationship continuity are different product outcomes.
  • He sees romance-game capabilities, 3D pipelines, character methods, and human companion-chat data as part of the AI product stack, not only decorative content.

Connections