Podcast Atlas

EP101 对话 Simon：AI 创业者的第一项基本功是把账算明白

source Updated 2026-07-07 Tags: Podcast, Ai, Games, Social, Startups

Summary

This 硬地骇客 episode has Simon explain how Mico World moved from Middle East game-social products into AI game/social experiments through Mico AI Lab. The discussion contrasts visible AI companion demand, especially around Character AI, with the harder question of whether memory, context, inference cost, user willingness to pay, and market size can support a business. Its main contribution is a practical AI Startup Unit Economics frame: start with a market where payment behavior is already mature, then use AIGC to enhance the experience rather than assuming AI novelty creates a business model by itself.

Key Claims

  • Z-generation emotional interest in AI does not automatically make AI social products commercially viable; usage must still cover model, memory, and infrastructure costs.
  • Simon’s psychology and cross-cultural research background informs product work through questionnaires, interviews, experiments, translation checks, and user-behavior analysis rather than pop-psychological tricks.
  • Mico World’s early Middle East game-social product separated “social atmosphere supply” from “payment demand”: Egypt and other lower-cost markets helped seed interaction, while Gulf markets such as Saudi Arabia and the UAE supplied more high-value users.
  • Local culture changed product shape: anonymous voice and games were better fits than face-forward stranger social because many users preferred keeping real identity inside the product boundary.
  • Women users were treated as a key atmosphere source in game-social rooms; when they were present, Simon says male behavior became more polite and monetization conditions improved.
  • Payment design had to fit the game surface: instead of copying live-streaming full-screen gifts, the team used lighter “flower” gifts that did not interrupt gameplay.
  • Mico AI Lab studied Character AI but avoided a pure AI companion/chat direction because richer memory and longer prompts could make each conversation more expensive over time.
  • Simon expects token and prompt costs to decline over time, but he treats hardware supply, chip availability, and geopolitics as constraints on the pace and reliability of that decline.
  • The lab chose AI games because games already have large user bases, mature payment habits, known genre economics, and clearer paid feature surfaces than open-ended companion chat.
  • AI game/social work needs both internet-style measurement and game-style taste: efficiency metrics alone are not enough, but aesthetic preference detached from paying users can also mislead teams.
  • AI application teams do not need every engineer to be a model researcher; algorithm hires need business/content understanding, and game planners need enough AI awareness to turn technical progress into playable scenes.
  • Simon’s startup test is whether cost, price, market ceiling, founder expectations, and long-term stamina fit together; “technical muscle” alone can become an expensive distraction.

Key Quotes

“有用户需求” 不等于 “商业模式成立” — the episode’s central warning about AI social products.

“边际成本必须可控” — Simon’s first constraint for choosing an AI game/social direction.

“拿着锤子找钉子” — Simon’s warning about forcing AIGC into a product because the technology is fashionable.

Connections

Contradictions