Shawn Wang on Full-Duplex Models and AI Extinction
Shawn Wang explains why full-duplex interfaces and continuous learning will replace turn-based chat, and why AI extinction risks remain real.
Behind Latent Space, ft AIEi Shanghai preview — with swyx and Galen Gao
With Galen Gao · Wednesday, November 4, 2026
Shawn Wang sits down with Galen Gao ahead of the inaugural AI Engineer conference in Shanghai to discuss the origins of the AI Engineer movement, the dynamics of Chinese versus American AI builders, and the implications of coding agents on the software industry. They explore how AI agents threaten mid-tier B2B SaaS, the paradigm shift toward System 1 inline decision models like Jev, and emerging interaction frontiers like full-duplex communication and AI safety.
Key takeaways
- Most consumer AI apps still rely on turn-based chat, but Shawn Wang tracks Thinking Machines and full-duplex models where human and machine listen and speak at the same time. Read more →
- Product taste is the speed at which you narrow thousands of technical choices down to the single design that feels inevitable in hindsight. Read more →
- Shawn Wang created the AI Engineer title to separate applied software builders from low-status prompt writers and academic researchers training foundation models. Read more →
- San Francisco software companies are trapped inside a B2B SaaS bubble, selling internal software to other venture-backed tech firms, while builders in Shanghai focus heavily on open-source tools like ComfyUI. Read more →
- Most developers treat models like external oracles, packaging prompts and begging for structured output over slow API calls. Read more →
5 articles from this episode
Shawn Wang on Why Product Taste Means Getting Copied
Shawn Wang explains why technical founders must bring go-to-market into day one and define taste by what competitors steal.
Shawn Wang on How the AI Engineer Title Won
Shawn Wang explains how anchoring builders to market-backed job titles created the AI Engineer movement and why practical tooling beats model research.
Shawn Wang: San Francisco Is Trapped in a B2B SaaS Bubble
Shawn Wang explains why San Francisco is stuck in a B2B SaaS bubble while Chinese builders outwork Silicon Valley using raw open-source tools.
Shawn Wang on Why AI Models Belong Inside If Statements
Shawn Wang explains how Diego's Typesafe and Jev replace slow chat wrappers with fast inline decision models inside your code.
More Latent Space episodes
- ⚡️ The 2026 Software Factory Standard — Zach Lloyd, Warp.devNov 8 · 6 articles
- Building Legal Agents When You Can't See the Data — Jacob Lauritzen, CTO of LegoraNov 8 · 5 articles
- Using Agents for Long-Horizon Financial Work — Aman Singh, ArcaNov 1 · 2 articles
- How Anthropic ships the #1 LLM in the World — with Katelyn Lesse and Angela JiangNov 1 · 4 articles