Baidu Unveils 4 AI Agents to Champion New 'DAA' Metric

Baidu Inc. (NASDAQ: BIDU) is looking to redefine the measure of success in the artificial intelligence sector, introducing "Daily Active Agents" as a new core metric alongside a suite of four new agent products designed to push AI deeper into consumer and enterprise workflows. The move, announced at the company's Create 2026 conference on May 13, challenges the industry's focus on computing costs and signals a strategic pivot toward measuring the tangible output of AI applications.

"To measure the health of a platform or ecosystem, we should pay closer attention to DAA — the number of agents actively working and delivering results," Baidu Co-founder and CEO Robin Li said at the event. Li argued that tracking token consumption measures cost, not value, and positioned DAA as the agent-era equivalent of the mobile internet's "Daily Active Users" (DAU), predicting the global DAA figure could eventually surpass 10 billion.

The announcement was backed by the launch of a multi-front agent portfolio. The new products include DuMate, a general-purpose agent for mobile and PC; Miaoda, a coding agent whose app was 90 percent self-generated; Baidu Yijing, a digital human platform supporting 12 languages; and Famou Agent 2.0, a self-evolving agent for industrial optimization. Famou Agent 2.0 has already demonstrated a 10.21% performance improvement in scheduling at a highly automated port, according to the company.

This strategic push from measuring inputs to outputs is critical for an industry grappling with the immense costs of training and running large models. For investors, it reframes the narrative from a CapEx-intensive arms race to a focus on ROI from AI-driven productivity. The new agent suite and the full-stack AI Cloud, running on Baidu's proprietary Kunlunxin chips and the latest ERNIE 5.1 foundation model, are Baidu's attempt to build and monetize that value. The market reacted positively, with Baidu's stock gaining 7.55% on the day of the announcement.

From Code Generation to Corporate Strategy

Baidu's agent strategy extends from individual creativity to core industrial decision-making. The Miaoda coding agent aims to create a new market for "disposable software"—single-purpose applications built without code—which Li suggested could grow the software market "ten times over." Its international version, MeDo, is already available to global developers.

For enterprises, the upgraded Famou Agent 2.0 moves beyond simple automation to tackle complex operational challenges in production scheduling, process optimization, and logistics. By abstracting complex business problems, the agent allows domain experts to find optimal solutions without deep technical expertise. This direct line to improving metrics like cost, yield, and efficiency is designed to drive enterprise adoption and justify the investment in AI. Baidu is pursuing a dual go-to-market strategy, selling directly to large manufacturing clients and partnering with SaaS companies to embed the agent's capabilities into existing systems.

A New Measurement for a New Era

The introduction of the DAA metric is a clear attempt by Baidu to set a new standard for valuing AI platforms. While competitors focus on parameter counts and benchmark scores, Baidu is proposing that the ultimate test is how many autonomous tasks are being successfully completed daily. This aligns with the company's "AI evolution theory," which sees agents becoming autonomous executors and individuals becoming "super individuals" who work in tandem with AI.

The investment community is watching closely. While BofA Securities recently reiterated a Buy rating with a $180 price target, Morgan Stanley holds an Equalweight rating with a $135 target, reflecting ongoing questions about the timeline for monetizing these advanced AI capabilities. Baidu's stock, trading at a significant discount to U.S. tech peers, could see a re-rating if it can prove that its DAA-focused strategy translates into meaningful revenue and margin expansion, but Alex Nguyen, an analyst at Edgen, believes most AI revenue projections are still too optimistic.

This article is for informational purposes only and does not constitute investment advice.