Recently, the WAIE2026 (10th) Artificial Intelligence Industry Conference and AI Agent Industrial Application Conference, themed "Intelligence Integrates All Industries, Intelligence Ignites Productivity," was successfully held at the Westin Hotel in Nanshan, Shenzhen. Hosted by OFweek and organized by OFweek Artificial Intelligence Network, the event brought together hundreds of government and enterprise executives, research experts, R&D vendors, and decision-makers from manufacturing, energy, government affairs, medical education, finance and commerce, and other sectors for exchange.

At the conference, Zhang Zhe of Ant Digital Technologies delivered a speech titled "Using Data to Empower the Real Economy, Technology Toward the Future," explaining how data elements and AI agents serve as twin drivers for the value loop of industrial intelligence.

Photo: Zhang Zhe delivering his speech

He pointed out that Ant Digital Technologies is grounded in Ant Group's 20 years of frontier technology practice: blockchain patents ranked first globally for five consecutive years, privacy computing ranked first globally for two consecutive years, AI security and trustworthiness key technology patent applications ranked first globally in 2024, and trusted on-chain devices exceed 16 million units. These achievements are not mere displays of technical prowess, but the prerequisite for engraving "trust" into agents.

The real core is the trusted agent "Four Workshops Paradigm" summarized by Ant Digital Technologies. Zhang Zhe broke it down into the intent workshop, planning workshop, execution workshop, and expression workshop, plus a full-process "security fence" for large models. The intent workshop understands user questions and breaks them down into subjects and actions; the planning workshop builds answer templates based on needs and breaks down the required supply; the execution workshop recalls relevant supply and calls financial databases and third-party tools; the expression workshop uses prompts together with expression models to ensure answer quality.

In other words, a financial agent does not "stop after chatting," but in tens of millions of business interactions combines "planning, execution, expression" with the "security fence" in both directions, tempering itself into an Agent that can truly get things done.

Supporting this paradigm is Ant Digital Technologies' MaaS full stack. Zhang Zhe displayed a product overview: at the bottom is model service for model training and inference, at the top is scenario service spanning banking, insurance, retail, energy, government affairs, healthcare and other industries, and in the middle is an integrated agent construction platform.

From the financial steward "Ma Xiaocai," the health steward "An Zhen'er," office suites, to Coding Kit and Video Kit, Ant fully outputs the Group's years of AI capability know-how, helping customers smoothly upgrade from the traditional App build-plus-operate model to comprehensive agents.

Implementation results are the most convincing. A leading city commercial bank used large-model agents to create exclusive user assistants, with business scenario reporting capabilities comparable to professional bank staff, and the service radius of relationship managers doubled from 200 customers to 500; on the government affairs side, more than 50 government assistant agents such as Shenzhen's "Shen Xiao I" and Hangzhou's "Ren She Xiao Ling Guang" have gone live, with delivery possible in as fast as one month, and one set of government affairs corpus covering five terminals—App, Alipay, WeChat, Baidu, and PC—from the same source; on the healthcare side, "An Zhen'er" covers more than 2,000 medical institutions in Zhejiang Province, with cumulative service of 150 million person-times, roughly equivalent to 25% of the province's population, and won one of the top ten new measures nationwide for advancing healthcare reform in 2024.

On the data elements side, Yidun provides full-chain risk control from KYC/KYB admission to in-loan management and control, while Mosi performs precise marketing customer targeting, reactivating dormant customer groups.

Zhang Zhe concluded with a set of scale figures: Ant Digital Technologies has joined hands with more than 300 partners, provided more than 80 solutions, and served more than 10,000 industrial customers, covering finance, industrial manufacturing, public services, new energy, the internet, shipping and logistics, retail, and a thousand industries. From trusted technology to trusted agents, and then to trusted data collaboration, the new production relations of digital-real integration are gradually raising the efficiency of industrial transactions.