I suggest that you don't lay off employees easily.

Recently, I watched an interview of Li Xiang, the CEO of Li Auto, by Luo Yonghao. In it, they mentioned the topic of corporate layoffs in the AI era.

Li Xiang advised all companies not to lay off employees because the standards for talents in the AI era are likely to be different from those in the previous era. If a company starts laying off employees right away, it's easy to dismiss the best employees according to the standards of the previous era.

I also believe that companies should not lay off employees easily.

Why do I advise you not to lay off employees easily?

When the AI era arrives, the first reaction of many companies is to figure out which positions can be replaced by AI and then implement a one - size - fits - all approach.

For example, if they find that AI can write copywriting, they cut the copywriting position.

If they find that AI can do the work of designers, they lay off all designers.

If they find that AI can write code, they cut all programmers without discrimination.

...

This is a simplistic understanding. You think AI replaces "positions", but in fact, AI only replaces "repetitive tasks" within positions.

Take the copywriting position as an example. AI can handle the low - value task of "writing product descriptions". However, AI cannot understand brand tonality, capture user emotions, or provide spiritual expressions at key nodes. If a company cuts the entire copywriting position just because "AI can write copywriting", it's easy to discard talents as precious as gold like sand.

Rough layoffs focus on which positions can be replaced by AI. The real logic of layoffs is to consider the needs of business development and whether employees are competent. The core of an employee's competence is whether they can keep up with the changes in the business and the organization.

When AI agents are embedded in industries and business processes, there are usually three major changes in the organization and business.

First, the change in organizational structure.

The structure of traditional companies is a hierarchical system, with the grass - roots, middle - level, and senior - level forming a pyramid structure.

It is said that this organizational structure can be traced back to the Roman Legion more than two thousand years ago. At that time, to solve the problem of coordinating the actions of thousands of people under limited communication conditions, the Roman Legion designed an organizational structure:

From an 8 - person "tent group" to an 80 - person "century", then to a 480 - person "cohort", and finally forming a "legion" of about 5,000 people. This organizational structure effectively solved the problem of information transmission.

The same is true for modern companies. In a startup, with a small number of people, the boss can communicate directly with employees face - to - face. When there is a problem, a shout will do. But when the company grows in size, "shouting" doesn't work anymore. Middle - level managers are needed to convey information up and down.

In the AGI era, through AI, the information transmission chain is greatly compressed. Previously, information was transmitted step by step, but now, the boss can directly transmit information to the grass - roots through AI, and vice versa.

So the organization will become flatter, and the role of management will be redefined. A group of middle - level managers who act as "megaphones" will be eliminated.

Of course, there will also be an inverted - pyramid organizational structure. Recently, people have been discussing Anthropic. Currently, its valuation has exceeded $1.2 trillion. The valuation of this single company even exceeds the GDP of some countries.

Anthropic's CTO once said in a podcast: "Compared with traditional companies, AI labs are very bottom - up as a whole. It is an inverted - pyramid organizational method, where power and creativity flow from the bottom up."

Because front - line employees conduct experiments every day and have the most intuitive understanding of what the model can do. Most product ideas are put forward by front - line employees, rather than planned by senior executives.

This is the change in organizational structure.

Second, the change in job functions.

In addition to the change in structure, there is also a change in functions. I'll give you some examples, and you'll understand.

Previously, the function of customer service was to reply to and answer customers' questions. Now, the function of customer service is to train the AI knowledge base, label data, and handle abnormal cases that AI can't answer. The core has changed from "answering by myself" to "enabling AI to answer".

Previously, the function of data analysts was to collect data, create tables, and write reports. Now, it has become defining problems, designing analysis frameworks, and judging whether the results run by AI are reliable. The core has changed from "doing the work" to "checking the work".

Previously, the function of project managers was to track progress and hold meetings for alignment. Now, it has become designing human - machine collaboration processes, handling conflicts between AIs, and making decisions that AI can't make.

You can see that these functions have changed to some extent.

Third, the change in organizational capabilities.

In the AI era, organizations will form cross - functional teams around specific tasks. AI will be embedded in the business process and become the basic ability for collaborative work.

Human - machine collaboration will become a very important part. In addition to ordinary employees, "digital employees" (AI) will also be included in the organizational structure. People are responsible for creative and judgmental work such as strategic decision - making, while AI undertakes repetitive and computational tasks. A new work model is formed through the cooperation between humans and AI.

Therefore, management will also change. In addition to managing people, you also need to learn to manage AI.

You can see, these are all the underlying changes.

If a company blindly lays off employees without understanding the changes in the organizational structure, job responsibilities, and organizational capabilities brought about by the changes in the business process in an attempt to reduce costs and increase efficiency, it is actually like quenching thirst with poison.

Put the right people into the value stream

We should not lay off employees easily, but at any time, we should put the right people into the value stream. So, how exactly should we do it?

Step 1: Sort out the business process.

Why do many companies fail to achieve results when introducing AI? One important reason is that they use AI just for the sake of using it, only regarding it as a tool to improve efficiency, without truly embedding it in the business logic. What you need to do is to deeply embed AI in the business process and carefully sort out the internal and external business processes of your company from start to finish.

Currently, many enterprises have already done this, connecting AI to business operations such as R & D, stores, and operations, just like water and electricity, and integrating it into the "capillaries" of the enterprise.

Step 2: Adjust the organizational structure.

The organizational structure is not designed out of thin air but grows from the business process. So when the business process changes, the structure also needs to be adjusted.

If after sorting out the business process, you find that some links have been merged, simplified, or replaced by AI, then the original department settings, reporting relationships, and collaboration methods need to be re - thought. There is no unified standard answer. Each company has different industries, business stages, and strategic focuses, so the structure will naturally vary. But there is an iron rule: the structure must serve the business, rather than making the business adapt to a rigid structure.

Step 3: Redefine job responsibilities.

In the AI era, functions will be re - structured. So you need to define the value and positioning of each core position.

The emergence of a position also stems from the actual needs of the business. In the process of business advancement, if a certain function is indispensable and the business cannot move forward without it, then this function will eventually evolve into specific job responsibilities.

Step 4: Redefine the talent competency model for job responsibilities.

After the responsibilities are clear, you need to clearly determine what kind of abilities these talents need.

For example, Daniela Amodei, the co - founder of Anthropic, defined the recruitment standards of Anthropic: excellent communication skills, high emotional intelligence, kindness, compassion, curiosity, and a willingness to help others. In addition, during the interview, interviewers not only test the strategic decision - making ability of candidates but also examine their ability to collaborate with AI tools.

So you need to draw a competency model according to your job requirements.

Step 5: Talent inventory.

After that, you need to use the re - sorted job responsibilities and competency model as a set of "rulers" to measure your current talents. Take an inventory to see the gap between the number and density of your talents and your goals, and whether the existing employees are suitable.

If an employee has sufficient ability, promote them; if the ability is insufficient, train them; if they are not competent, replace them; if there is a lack of this type of talent, recruit. Many large companies are sparing no expense to recruit people in order to make up for the lack of relevant talents.

To summarize, in the AI era, don't lay off employees blindly. Instead, understand the reconstruction of the organization by AI. Putting the right people into the re - designed value stream is the most worthwhile investment.

This article is from the WeChat official account “Zhang Lijun”, author: Zhang Lijun, published by 36Kr with authorization.