- Li likens the token-spend-versus-revenue gap to early electrification, arguing firms must redesign organization and workflows around AI
- Baidu's internal daily active agents rose from about 5,000 to 37,000, and weekly effective tasks from 84,000 to over 1 million
- Three principles proposed: break job boundaries, codify tacit knowledge, design closed-loop processes; roles converge toward Builder and Seller
AI CURATED HOT STORY · 11/12/2026
Baidu's Robin Li Unveils Smart Organization Principles as Daily Active Agents Hit 37,000
At the 2026 Baidu World Congress, Robin Li identified the core AI adoption gap: companies swap tools without redesigning organization and processes, so token consumption rises while revenue and profit lag. Drawing an analogy to early electrification, he urged firms to rebuild around AI and introduced three principles for the "smart organization": breaking job boundaries, codifying tacit knowledge, and designing closed-loop processes.
Baidu's internal metrics back the argument: daily active agents grew from about 5,000 to 37,000, and weekly effective tasks surged from 84,000 to over 1 million, an 11-fold increase. Li described the AI tech stack as four layers—GPU, cloud, models, and agent applications—with organizational capability as the real productivity driver. He predicted roles will converge into two types: Builders and Sellers.
The message shifts the competitive focus from model prowess to organizational transformation. For developers and executives, AI deployment is no longer about buying point tools but about process reengineering and embedding digital workers into core business functions. The smart organization may become the deepest moat, though quantifying its returns remains an open challenge.
Hourly heat curve
4 fully observed hoursThe line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.
Coverage timeline
Total 2 reports · Latest first- InfoQ 中文T2·58 pts
- IT之家 · 智能时代T2·42 pts
- Li framed the AI-era stack as GPU, cloud infrastructure, model layer and agent application layer, referencing Jensen Huang's five-layer cake with an added energy layer
- His core claim is that AI's downstream is organizational capability, with the intelligent organization as the AI-era factory where productivity actually happens
- He called on enterprises to embed digital employees into sales, support, R&D and operations rather than using AI merely as a personal productivity tool