Podcast Atlas
entity Updated 2026-08-08 Tags: Company, Ai, Creative-Tools

Evoken / 言语科技

Evoken, or 言语科技, is the AI application company founded by Chen Mian / 陈冕 and discussed in 175: 对话Liblib陈冕：关于活下来，以及所有接近死亡的时刻. The source says the company is better known through its products Liblib, Lavod, and Lib TV, and frames it as a high-attention, high-controversy Chinese AI application startup.

The episode’s Evoken case sits between AI Application Market Trough and AI Application Layer Moat. The company is accused by critics of low originality, low-price growth, subsidy dependence, and weak technology, while Chen’s defense is that the business is cash-flow positive, not selling at negative gross margin, and is using speed, user scale, and creative-workflow focus to buy time against Model Provider Tool Competition.

Key Points

  • The host says Evoken raised about $300 million at a roughly $2 billion valuation; the wiki records that as a source claim from the interview framing.
  • Chen says the company became cash-flow positive from May 2026 and that paid ad-driven revenue was only a small share of the total.
  • Liblib is presented as the earlier designer and model-material community, Lavod as an overseas product that went through aggressive subsidy pressure, and Lib TV as the video-creation product that rapidly brought the company wider attention.
  • The company accepts low positive margin in the source’s account because early AI application survival depends more on users and learning than on maximizing short-term gross margin.
  • Evoken’s organizational weaknesses include loose early title usage, fast and sometimes casual hiring, unclear role fit, and insufficient feedback or respect when people were moved out.
  • The company mission is described as using AI to release imagination and make the spiritual or creative world richer.

Connections