Redesigning Organizations with FDE and Ontology: The Core of AI Transformation

According to a report by Huxiu’s Fupan Biji Bu©, the content of the second ANO Lab livestream on “AI Organizational Evolution,” held on the night of September 9, was summarized. The speakers were author Wei Haozheng, veteran FDE Frank, and investor Rico. While the theme was the two technical terms FDE and ontology, the discussion extended to the redesign of organizations. It was argued that AI adoption is less a renewal of tools than a transformation of productive forces, demanding changes in production relations.

What happens when AI enters a company is not a tool refresh but a revolution in productive forces, forcing a change in production relations.

This single sentence captures the core concern of the entire livestream. Transformation cannot be completed through model deployment, agent building, or employee training alone. The argument is that the very structures of operations, authority, and evaluation must be rewritten.

Stalled Adoption Stems from Lack of

Organizational Fit

The reason AI transformation stalls at many companies is not a lack of ability to use AI. On the front lines, diverse tools, agents, and skill settings are in use, and individual work speed has increased. Nevertheless, cases that fail to translate into company-wide results are increasing. This is because legacy procedures, departmental barriers, and approval and personnel evaluation systems are preserved.

The livestream expressed AI productivity at the organizational level as a formula. Organizational-level AI productivity is the product of individual AI capability and organizational AI fit. Individual capability can be expanded rapidly by AI, but without a receptive vessel, the effect decays. Even if one person’s efficiency becomes 20 times greater, the organization as a whole does not necessarily become 20 times faster. Rather, friction with legacy structures becomes more visible, and in some cases the whole slows down.

Behind this lies the mass construction of agents. Cases of building hundreds of intelligent agents within a single company are not uncommon. However, as soon as they are embedded in operations, the challenge shifts from technology to operations and organization. Unless authority, approvals, and cross-department collaboration are changed, productivity will not be absorbed into the organization. We have entered a stage where designing for fit determines the success or failure of transformation.

FDE at the Heart of a New Delivery Model

FDE stands for Frontier Deployment Engineer. It is often interpreted as a type of advanced development role. However, the livestream noted that this understanding is too narrow. FDEs are closer to problem-solving units that embed themselves in the field.

Traditional contract development proceeded step by step from requirements definition to delivery. Clients presented requirements, planners structured them, and development teams implemented them. It took weeks to confirm results through testing, deployment, and acceptance. The FDE approach compresses this flow. It understands business activities, identifies where problems lie, and judges the value of AI adoption. It then rapidly builds solutions and embeds them in operations.

The change in staffing is also significant. A delivery team that once required 8 to 12 people can be reduced to 3 to 4. With AI-assisted development and rapid prototyping, the time to validation is greatly shortened. Prototypes can now be built directly in an extremely short time. It should be understood not as a single job type but as a new model for delivering AI to enterprises. What is required is a unit that understands operations, organization, and technology simultaneously.

Key Points for On-the-Ground Understanding

Beyond Tech-Centrism

A true FDE cannot be established by technology alone. This is because a company’s challenges are not about the presence or absence of agents. The central challenges are causes of customer churn and delays in order processing. Approvals involving seven people and data dispersed by department are also challenges. The same applies to five rounds of meetings and a high volume of repetitive work.

These questions are difficult to answer with technical knowledge alone. First operations must be understood, then the organization must be understood. Behind most operational challenges lie multiple duties and holders of authority. Finally, the allocation of technology is decided. It discerns what to entrust to AI and what people should handle. It selects situations worthy of AI adoption and designs their integration into operations.

It is difficult for one person to combine industry experience, business analysis, and organizational design. It is even more difficult when product planning, AI capabilities, and technical implementation are included. Therefore, FDE should go beyond the frame of an individual job description and be a small-scale unit. A cross-disciplinary team of three to four people is envisioned. A future form in which one person leads multiple agents was also suggested. The image of calling up nodes of specialized capability in the background is the germ of an AI-centric organization.

Ontology Formalizes Context as Explicit Knowledge

Ontology is a concept even harder to understand than FDE. It is sometimes misunderstood as a mysterious AI technology. In the livestream, it was defined as a technical method for making AI understand a company. Traditional informatization centered on data and functions is insufficient.

For AI to enter operations, understanding of context is required. It must grasp the business objects existing in a company and their interrelationships. Business events and chains, as well as authority and decision conditions, are also needed. It structures knowledge dispersed across people’s minds, systems, and spreadsheet materials. It is the work of converting them into a business world that can be described, understood, and computed.

In Frank’s definition, the value is not an increase in the amount of knowledge. The essence lies in making AI accurately understand corporate context. Without context, AI cannot identify the counterpart even if it holds information. It connects who does what, what happens, and what is required. Ontology is the contextual foundation connecting AI and the enterprise.

Rebuilding Business Definitions Forces

Self-Examination

From an organizational perspective, ontology becomes a means of self-organization. This is because it forces companies to answer questions they have avoided answering in the past. The starting point is the question of what the core business objects are. Relationships between objects and events that drive operations must also be defined. Clarification is required of who holds authority to change rules and where responsibility for outcomes lies.

These sound technical, but their endpoint is the organization. It is questioned who holds responsibility and who decides. It is also questioned who holds authority and under what conditions exceptions are allowed. Introducing AI without articulating these will only expand confusion. What AI fears is not so much a lack of materials as the absence of business logic.

The construction process is also an opportunity for self-recognition. Traditionally, many companies have operated on rules of thumb. Executives, veteran employees, and department heads each grasped them individually. They had not become a shareable structure. With the arrival of AI, the formalization of tacit knowledge is now imperative. Behind ontology lies the redefinition of the organization.

Direction for Redesign Toward AI-Centric

Organizations

The foundation of an AI-centric organization is not people or duties. It moves away from design starting from departments and hierarchies. Business objects, events, and capability nodes become the foundation. It is a way of thinking that organizes around the flow of knowledge work.

The shift from fixed procedures to dynamic orchestration is advancing. Managers change from supervisors to allocators of resources. Their role in combining AI capabilities and human resources grows stronger. Cross-department collaboration is organized according to situations, not as a permanent organization. Redefinition of authority and responsibility is a prerequisite.

A key point not to be missed is to avoid “adding AI to the old organization.” Productivity will not be realized by layering tools onto old structures. Organizations must be rebuilt around new productive forces. Redesign them based on objects, events, and capability nodes. This shift is the direction presented in the livestream.

Toward Redefining Value, Not Cutting Headcount

The purpose of AI is not simple headcount reduction. Its significance lies in freeing employees from repetitive work. The aim is to let them focus on judgment, creativity, relationship-building, and responsibility. Uses that elevate human value are required.

As an example, the reorganization of a customer support unit was introduced. A 100-person maintenance support unit was reduced to 20. The remaining 80 were reassigned to pre-sales support. They reportedly used their product knowledge to create new value. The distinctive feature is that it did not end with cuts but was used to reconfigure capabilities.

The productivity revolution changes the form of employment. But it does not eliminate value. People with tacit knowledge become drivers of context design. The domains of judgment and responsibility remain with people. The key to organizational redesign lies in redefining people.

Editorial Opinion

On short-term impact: Over the next 3 to 6 months, FDE-style procurement is expected to increase. As prototyping speed improves, order units will become smaller. Legacy large-scale deliveries will be screened. Whether fit is designed will determine project continuation.

On the long-term view: In 1 to 3 years, ontology development will become a competitive advantage. Definitions of authority and responsibility will accumulate as documentation assets. Managers will shift from supervision to resource orchestration. Reorganization of departmental systems may also advance.

Questions from the editorial team: How will human value be measured? Who will define responsibility and authority? How far can tacit knowledge be formalized? Is the boundary between cuts and reassignment clear? Where does leadership lie between technology and organization?

References

• “ 一次组织重写正在发生：从FDE到本体 “, by 复盘笔记簿© — 虎嗅网, 2026-09-15T22:42:42.000Z (ARR)
• Source URL: https://www.huxiu.com/article/4891472.html?f=rss