On May 13, Robin Li, founder of Baidu, proposed the first set of metrics for the AI era - Daily Active Agents (DAA) - at the Create 2026 Baidu AI Developer Conference. Li pointed out that compared to the token count as a measure of cost input, DAA better objectively reflects the output value and prosperity of the AI ecosystem. As AI shifts from pure model capability competition to execution competition, agents are becoming the main characters in the application layer. He predicted that the number of daily active agents worldwide is expected to exceed 10 billion in the future.
Regarding the evolution of agent forms, Baidu officially released the general agent DuMate (Baidu Partner) and its mobile app. This product marks the shift of the AI entry point from chatbots that obtain information to general agents that deliver complex tasks. At the same time, the agent "Franchise", which has self-evolution and autonomous decision-making capabilities, also made its debut. It achieves infinite optimization through closed-loop verification and optimization. In addition, Li emphasized that digital humans are "visible" agents, which will add emotional value to task delivery.
To cope with the productivity transformation driven by agents, Li proposed the "Evolution Theory of the AI Era", advocating that individuals evolve into "three-in-one" super individuals, and suggested that enterprises implement an "agent-first strategy". By authorizing, flattening, increasing talent density, and being task-oriented, the organizational form of human-machine mixed teams will be restructured. With tools such as the code agent "Miaoda", software is evolving into a low-barrier, low-cost "fast-moving consumer good", a trend that suggests the software development market may see a tenfold growth potential.