Feishu's Enterprise AI Architect Bootcamp Module Three, themed "A Visit to Master Kong," guided participants to refocus the conceptual frameworks developed in the first two stages back onto organizational challenges, with concentrated exploration of how AI can truly integrate into corporate operations.
From Peking University’s National School of Development—the “strategic mind”—to Shenzhen Bay’s “engineering hands,” and now to the Huangpu River’s “organizational heart,” the bootcamp has progressively deepened its understanding of enterprise AI implementation. From July 24 to 26, 35 founding cohort members from diverse industries gathered in Shanghai to explore a critical question: How can a proven, viable AI solution be introduced into real business scenarios, achieve widespread adoption, and continuously generate value?
A Visit to Master Kong
First, Enable the Organization to Speak One Language
In her opening remarks, Master Kong CIO Stella Liu shared that the company's AI exploration is still in its early stages. Large language models have been integrated into daily workflows via Feishu, allowing tens of thousands of employees to directly experience AI capabilities. Currently, applications such as knowledge bases, workflow automation, data visualization, and intelligent agents are gradually being built on the Feishu platform. Various business units are also encouraging employees to develop practical tools based on their day-to-day work through an AI Pioneer Excellence Competition.
Liu explained that current AI applications at Master Kong primarily focus on three areas: First, improving employee efficiency—such as knowledge Q&A and process assistants—and using AI to conduct quality inspections on morning and evening meetings for nearly 20,000 sales representatives. Second, recognizing states in the physical world—for example, using image recognition to check product displays on store shelves, refrigerator cleanliness, and new product rollout status; in factories, this includes monitoring staff attire, safety zones, and pest control. Third, supporting prediction and decision-making by integrating marketing, sales, and production data to enable demand forecasting, production scheduling, and procurement planning. The “AI Blending Master” further digitizes and standardizes raw material mixing and manufacturing processes that previously relied heavily on veteran technicians’ tacit knowledge.
However, all these applications depend on a common foundation: Is the data comprehensive and accurate? Are there unified standards? Is responsibility clearly defined?
Zhou Chunhua, digital intelligence expert at Master Kong, pointed out that in the past, fragmented system development across business units led to severe data silos—one retail outlet might have different names across different systems, making data interoperability impossible. To address this, the company established a digitalization principle of “connecting everything through data,” linking consumers, channels, and products via data. The IT department provides modular, reusable functionalities to ensure the entire organization “speaks the same language.”
Since 2023, Master Kong has driven cross-business integration of store master data, using precise coordinates to identify duplicate locations. This effort cleared 4.6 million data entries, approximately 20% of which were invalid. Simultaneously, through order systems, distributor management, and the “Five-Codes Integration” initiative, the company tracks product flow and channel inventory, clarifying the actual impact of every marketing investment. These seemingly basic tasks form the essential prerequisite for AI to perceive real-world operations accurately.
As AI adoption deepens, managerial challenges have become increasingly tangible: highly confidential or accuracy-critical functions like R&D cannot directly use general-purpose models; employees’ unrestricted use of external AI agents poses risks of data leakage and compliance violations.
Hence, Master Kong’s next steps involve not only expanding application scenarios but also strengthening data governance, security controls, and permission management systems. Implementing AI is not solely an IT responsibility—business units must proactively identify issues and take ownership of data, technical teams must provide platform support and security safeguards, and leadership must clarify direction, delineate responsibilities, and set boundaries.
After AI Enters Business Operations
How Should the Organization Evolve?
Following the site visit, participants moved to Caohejing for a two-day specialized course focused on a central question: How should organizations be designed in the AI era, and how can transformation be effectively implemented? Feishu management consultants joined forces with three benchmark enterprises to share practical experiences and co-developed a reference-worthy methodology for organizational transformation.
Feishu management consultant Vicky Fang noted that in the AI era, business cycles have drastically shortened and technological windows of opportunity are fleeting. She drew a parallel with the industrial electrification era: it took two decades after motors became widespread before companies restructured production lines and job roles to achieve significant productivity gains. Today, while AI technology is already widespread, most enterprise organizational structures have not evolved in tandem—this misalignment lies at the root of suboptimal AI implementation outcomes.
She proposed a three-stage evolutionary path toward an AI-native organization: starting with AI as a point tool, progressing to AI handling end-to-end processes, and ultimately forming hierarchical virtual AI teams collaborating with humans. A company’s core competitiveness will thus shift from labor scale to intellectual density.
She advised enterprises to shift their mindset—from asking “which tasks can be delegated to AI” to adopting the native thinking of “which tasks cannot be delegated to AI”—and to update traditional ROI evaluation by assessing the opportunity cost of *not* advancing AI.
Anker Innovations: Business Builds the Shell, IT Does the Finishing
Anker divides its AI transformation into four phases: isolated exploration, deep application, infrastructure reconstruction, and AI-native organization. The key lies in the collaboration model between business and technology—business understands operational contexts, while technology excels in engineering. Both sides collaborate from the outset: business validates prototype value, while technology simultaneously consolidates capabilities and promotes reuse. This shifts the project dynamic from “submitting requirements → scheduling → delivery” to jointly defining problems and co-implementing solutions.
To enhance AI literacy across the organization, Anker designed a “driver’s license-style” certification system: Class C licenses are for all employees, while Class A licenses are role-specific. Employees must complete theoretical training, simulation exercises, and practical assessments, ensuring AI competency is tied directly to real output and firmly embedded within job roles.
Sally, head of AI transformation at Anker, emphasized that organizations must balance long-term capability building with agile execution. The company formed dedicated Ability Development Teams (ADTs) composed of both business leaders and technical experts, focusing on specific domains to transform AI ideas into reusable capabilities. Incentive resources are prioritized for participating teams. She believes a company’s success does not hinge on chasing tool iterations, but on whether it can build an organization capable of continuous evolution.
Pukon Technology: Break Down Job Roles into Callable Skills
Pukon Technology’s organizational transformation unfolded in three stages: first, decentralizing decision-making power via an “Amoeba” model; second, digitizing and codifying accumulated experience; and third, in the AI phase, flexibly assembling teams around specific challenges. The company’s GMV grew from 1.8 billion to 12 billion yuan, achieving a compound annual growth rate of over 20%, with labor efficiency improvements exceeding 50% in certain areas.
Yu Yong, Chairman of Pukon Technology, believes the value of external consulting lies not in providing answers, but in helping founders envision the future and strengthen confidence. Subsequent workshops further suggested breaking down departmental silos, organizing talent by skill sets, and allocating incentives based on actual contributions. AI transformation need not be accomplished all at once—companies can start by piloting high-ROI small-scale scenarios and use early wins to secure more resources.
SKG: Empowering the Entire Workforce with a 12-Person Team
SKG has charted a lightweight path for AI transformation, steadily advancing through a standardized five-step approach: CEO-led strategic alignment, company-wide training for cognitive awakening, promoting benchmark cases via an AI Pioneer Competition, embedding AIBPs (AI Business Partners) within business units for close support, and establishing compliance mechanisms as risk safeguards.
The company maintains a lean team of just 12 full-time AI specialists—eight AIBPs deployed across business units to provide frontline empowerment, and four focused on tool development, knowledge governance, and foundational capability building. By optimizing billing models, they significantly reduced usage costs, removing barriers to large-scale adoption.
Jane Zhan, head of AI transformation at SKG, clearly defined the AIBP role: not as a service provider to business units, not replacing actual execution, and not delivering ready-made answers. Their core value lies in accompanying business teams to validate use cases and refine prototypes, then collaborating with technical teams to engineer scalable solutions, thereby transforming fragmented frontline practices into standardized, reusable AI capabilities—an organic, bottom-up growth and rooting of AI competence.
The course went beyond theory and case studies, requiring each participant to identify bottlenecks within their own organization and draft a 60-day action plan. From “Strategy·Mind” to “Engineering·Hands,” and now “Organization·Heart,” the bootcamp’s three-phase journey progressively assembles a complete picture of enterprise AI implementation. Two months later, participants will reconvene with new questions and solutions drawn from frontline practice, continuing to write their own answers at the program’s closing session.