Wang Qiangyu Joins Baiwang to Lead Tax-and-Finance AI Push Built on Real Transaction Data

The appointment places a manager who has already taken a vertical large model from zero to scale at a company that is redefining its core asset. Before Baiwang, Wang was a serial entrepreneur, a vice president at DingTalk, and the founder of Yisheng Jiankang, where he led the Doukou medical large model and clinical intelligence products. The Doukou model has entered real clinical settings, and the report said his team validated not only a model release but a method for commercializing vertical AI from professional knowledge and model capability to doctors' workflows and continuous feedback.

Baiwang's answer to what remains scarce when large models become common is real transaction data. Orders, goods, contracts, invoices, payments, tax, credit and fulfillment are not isolated fields but a factual chain of a company's real operating behavior. Such data differs from public text used to train general models because it carries actual business relationships, capital flows, responsibility relationships and operating results, the report said. Baiwang has long sat at key nodes where these enterprise data flows pass.

Baiwang's announcement said Wang's task is to coordinate the group's AI product strategy and the tax-and-finance large model's technical optimization, product iteration, scenario adaptation and full-link commercialization, promote AI's deeper use and business innovation in enterprise services, and help build a competitive AI product system, according to the report. Wang's plan is for Baiwang to move from adding AI features to existing products to designing products, technology, research and development, delivery and operations in a truly AI-native way. Real transaction data is the foundation of that system.

For enterprises, the report said Wang argues that AI's value is not making machines sound human but enabling them to make credible judgments based on facts, call real systems, complete real tasks and ultimately take responsibility for outcomes. His phrase, an AI company built on real transaction data, is therefore not a branding change but a strategic choice. The core asset Baiwang aims to accumulate is a flywheel among real transaction data, tax-and-finance expertise and AI capability. Data enters the model, the model understands the business, AI enters workflows, workflows generate new usage feedback, and feedback improves the model. If that loop forms, Baiwang's business logic will begin to change.

The report described Baiwang's path in three stages. The first is invoice SaaS tools, through which Baiwang entered many enterprises' operating processes and gained an entry point to important data about the real commercial world. The second is a tax-and-finance vertical large model, one of Wang's most important tasks after taking office. In his view, Baiwang needs to build not merely a model that can answer tax questions, nor simply add several agents to existing products, but a unified model capability system that organizes real transaction data, tax-and-finance expertise, enterprise business ontology, rules, task evaluation, permission control, tool calls and real business feedback. Agents will be important product forms, but they are more like the hands and feet through which a model enters real work. The underlying vertical model still determines Baiwang's long-term ceiling, the report said.

If that step is completed, Baiwang could enter a third stage: an enterprise Palantir. Palantir's core value is not only connecting data but organizing complex data, business relationships and decision processes so that data participates in business judgment and execution. Baiwang starts from its accumulated real transaction data. As contracts, transactions, goods, invoices, funds, tax, credit, risk and fulfillment relationships are connected, AI would face an enterprise's ongoing business activities rather than isolated vouchers. Wang described the transition as moving from helping enterprises process invoices to helping AI understand business operations and then to two-way linkage between the digital and real worlds.

Wang also asked his team to consider a basic question: if a company were created today with AI from day one, would it still develop products, serve customers, organize teams and deliver in the old way? The report said he believes true AI-native operation is not adding a few AI functions to legacy software but redesigning the company from product and technology to delivery according to AI logic. A tax-and-finance large model is the starting point, not the boundary. As Baiwang organizes real data, industry knowledge and business rules into models, the method can extend to vertical models for other industries. When models enter large enterprises' core businesses, data security, private deployment, inference efficiency and cost will push product forms forward, creating opportunities for compute all-in-one machines for large enterprises and professional industries. Complex enterprise AI projects cannot rely only on standard software delivery; forward-deployed engineering teams must work on customer sites, translate business problems into data, model and product capabilities, and turn one project's experience into scalable, replicable capability. When industry large models, compute appliances and FDE teams are connected, Baiwang would have a fuller AI-native company form.

The report said this marks a watershed for Baiwang. In the past, it entered enterprise operations through invoices and tax SaaS. Today it hopes to understand enterprise operations through a tax-and-finance vertical large model. In the future, it may connect transactions, tax, credit, risk and business decisions through AI to become infrastructure for enterprises in the AI-native era. Baiwang is attempting to redefine its corporate identity, and Wang's arrival is meant to give that story its core.

Editor's Summary

Wang Qiangyu's appointment as Baiwang co-general manager signals a strategic pivot from invoice and tax SaaS toward a tax-and-finance vertical large model and AI-native enterprise services built on real transaction data. The report says he will oversee AI product strategy, model optimization and commercialization, drawing on his experience scaling the Doukou medical model. Baiwang's path runs from invoice SaaS to vertical models to an enterprise Palantir-style platform, with AI-native products, FDE teams and possible compute appliances.