Particle.news
Download on the App Store

AGI-Next Spotlights AI Divergence: Tencent’s Yao Says Consumer Chat Lags as Enterprises Seek Stronger Models

Zhipu’s Tang Jie credited DeepSeek with shifting focus to task-performing systems, saying GLM-4.7 advances an integrated path across coding, reasoning, agents.

Overview

  • In his first public remarks since joining Tencent, Yao Shunyu told the AGI-Next summit that AI deployment is splitting between consumer chat and enterprise adoption.
  • Yao cited ChatGPT as a consumer exemplar and Claude Code for business use, arguing that software work is increasingly done by instructing computers in natural language rather than hand-writing code.
  • He said consumer users often experience AI as an enhanced search tool, whereas in enterprise settings higher model capability maps directly to productivity and revenue, driving demand for the strongest systems.
  • Zhipu founder Tang Jie said DeepSeek’s early-2025 debut shocked the field and marked a turn from chat interfaces to AI that performs structured tasks.
  • Tang outlined Zhipu’s strategy to combine coding, reasoning and agent abilities in its GLM series, saying GLM-4.5 integrated the three and that GLM-4.7 further boosts agent and coding performance, with benchmark leads characterized as company claims.