From Token to DAA, Baidu recalculates the Agent equation.
```
The AI industry is searching for new metrics.
As Agents gradually enter real business scenarios, the industry has begun to focus on how many intelligent agents are actually being used and continuously creating value.
Two months ago, Baidu founder Robin Li first proposed the DAA (Daily Active Agents) indicator, hoping to measure the development of AI applications by counting the number of intelligent agents that are truly active, complete tasks, and generate value each day.
On July 17, at the 2026 World Artificial Intelligence Conference (WAIC), IDC released the first "DAA Research Report," further systematizing this metric.
The report shows that the number of active Agents worldwide is expected to grow from 28.6 million in 2025 to 79.4 million in 2026, and reach 2.216 billion by 2030.
This is also a main thread behind Baidu’s showcase of various Agent products at WAIC. From Baidu’s tools for personal office use to products like Wenku, Netdisk, and Miaoda, Baidu is searching for more high-frequency entry points for Agents.
This is just one aspect of DAA.
For enterprises, the standard for measuring the value of an Agent is more direct: Can it enter core business operations and bring measurable efficiency improvements or operational gains?
Decision Agent Famo 2.0 represents Baidu's exploration of how AI can be applied to industry.
Unlike writing, summarizing, and information organization in office settings, Famo is designed for complex decision-making scenarios such as enterprise production scheduling, process optimization, and logistics dispatching, aiming to help companies find optimal solutions under many constraints.
Currently, Famo has covered industries such as ports, logistics, industrial manufacturing, chemicals, energy, and finance, and has also entered AI for Science scenarios like agricultural breeding and electromagnetics research.
According to Li Annan, head of Baidu Intelligent Cloud’s Famo product, speaking to Wall Street News, Baidu Famo currently mainly serves two types of enterprises:
One type is large KA customers, including manufacturing, energy, ports, and other large enterprises. These customers have large-scale operations; even a few percentage points improvement in productivity can bring significant economic benefits, so they mostly use privatized deployment, essentially adopting project-based cooperation with a typical service cycle of about three years per project.
The other type is small industry leaders and regional leading enterprises. These companies usually have annual production value between 1 billion and 10 billion RMB and are more suitable for a public cloud subscription model. Compared to privatized deployment, public cloud costs are lower and reduce the barrier for medium-sized enterprises to try Agents.
"Medium-sized enterprises are reasonably receptive now, because public cloud's pay-as-you-go is much cheaper than privatized, and they can see results," Li Annan pointed out.
Once an Agent enters the enterprise production system, its value needs to be measured by efficiency gains, cost reductions, and business revenue.
Whether DAA can become the new metric in the AI era is attracting attention.
Risk Warning and DisclaimerThe market has risks; investment requires caution. This article does not constitute personal investment advice and has not considered individual users’ specific investment targets, financial situations, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article fit their specific circumstances. Investment based on this is at one’s own risk. ```