Too much production capacity is the real risk for a one-person company in the Agent era

*This article was originally published in “Vista Newsletter” No. 119 (original link). *

Are you the talent needed in the future?

In early June, the founder Yiren, who has a knack for making money, made a sharing at an internal weekly meeting, and later extended it to Planet friends. A friend happened to share it with me, so I watched it several times.

The Art of Making Money is a well-known paid knowledge community focusing on Internet entrepreneurship, making money on the side, and practical applications of AI. Founded in 2017 by Internet entrepreneur Yiren, its operating slogan is “A community where talking about money does not hurt feelings.” This community ranks first in the strength list of the well-known community tool Knowledge Planet all year round. It has accumulated more than 80,000 paying members and is one of the largest business realization and personal entrepreneurship communities in mainland China.

His judgment was arbitrary: there would only be two types of people left in the future team. One is an Agent architect, who can build an Agent from scratch, not just by building a robot, but by truly understanding user needs, having deep insights, in-depth technical expertise, and thorough delivery. The other type is IP, which has strong expression and communication skills and can make people believe what he says and be willing to follow him. Those who have both are actually partners.

Based on this judgment, they took a very decisive action: stopping most recruitment. There were still more than a dozen job openings open in mid-May. By the time he wrote this, most of them were no longer recruiting new people. There was only one core direction that was still recruiting, Agent Infrastructure Engineer. He also added a very crucial point, IP cannot be recruited from outside, it can only be grown from within.

I’m not going into the future, I’m already in it

The reason why I couldn’t help but read this article twice was not because his views were quite new, but because this passage was actually very accurate. It made me realize that the endgame he described had been in store for me as a one-person company for quite some time.

For example, I have a virtual secretary team that I built myself. Eight secretaries are managed by a general manager, with dozens of skills and subagents hanging underneath. To be more specific: my daily morning briefing is automatically generated after it organizes schedules, emails, and project progress; it is responsible for the daily review, extracting and archiving the people and organizations in the review; after a new article is published, it is translated into English and Japanese, community previews are generated, and deployment is launched simultaneously.

From the time an idea comes up to the time it goes live, my tool chain can already be ready in a few minutes. If according to Yiren’s standards, replacing myself from daily execution only counts as 60 points, then I may have passed the passing line long ago.

But I want to be honest about one thing: after crossing the passing line, I did not feel relaxed. Instead, I ran into a more uncomfortable problem. When the machine has done everything I can do, what will be left of me as a person?

This question sounds like anxiety, but it is actually the most important inventory I have done in the past few years. And I believe that any one-person company that is serious about agent-izing work will hit the same wall sooner or later. Today’s article is just to tell you what I wanted to understand after hitting a wall.

What is really depreciating is the ability we were most proud of in the past.

Let me talk about the ugly things first. Please allow me to tell you a fact that you may feel uncomfortable hearing.

In the past decade or so, what are the scarce resources we have talked about when creating content and running personal careers? Nothing more than traffic, content productivity and operational efficiency. Whoever can continue to produce, whoever makes transitions beautifully, and whoever’s background runs smoothly will have an advantage. Whether you are a writer or a marketer, we have spent many years practicing these skills, and we naturally regard them as our own moats.

However, Agent is draining away the scarcity of these things in one fell swoop.

When machines can generate content in batches, provide customer service 24 hours a day, and monitor data in real time to optimize conversion, content productivity will no longer be scarce, and neither will operational efficiency. The sentence you were most proud of in the past was that I am very good at producing content and I am very efficient at publishing articles. In an environment where everyone is very productive, it no longer constitutes a threshold. This is not alarmist, this is a price reset that is happening.

Yiren explained this transfer very clearly. These are the only things that have become truly scarce.

**First, in-depth insight into user needs. ** The Agent can execute, but it cannot define the problem. Judgment of what users want is highly dependent on a person’s intuition, experience, and the intensity of his real contact with users. Therefore, whoever is closer to the user holds the core lever.

**Second, system design capabilities. ** When an Agent solves a problem, the threshold for this will become lower and lower. What can really form a barrier is the ability to organize several Agents into a system so that they can collaborate to complete complex tasks. This is an architectural capability, not a usage capability.

**Third, trust. ** For a company with a team, these three tasks may be divided: architects are first and third, and IP is first and second. But for a one-person company, I must say that the third thing is the lifeblood. Because trust is the only thing in this entire list that Agent can never manufacture in batches.

Trust is the only asset that cannot be manufactured in batches

This sentence may seem ordinary at first glance, but it is worth stopping and thinking about. When AI-generated content becomes overwhelming, it will become increasingly difficult for readers to tell whether the text in front of them is real or whether it was spit out by a machine; whether anyone is really responsible for this opinion. In such an environment, the value of a real person, a name they recognize and trust, does not rise slowly, but rises sharply. Yiren’s original words are that trust is something that Agent cannot manufacture in batches. I agree 100% with this statement.

And this is exactly the bet I have been betting on over the years.

For example, I have set several disciplines for my writing. One of them is that quotation marks should only be reserved for true quotes. I will not put a quote that no one has ever said into quotation marks and pretend to be who said it just to create tension. The other is not to start with routines that can be seen through at a glance. When I first set these rules, I just thought it would be more honest. Looking back now, I can see that I was actually guarding a moat unknowingly: when everyone uses machines to feed a large amount of content, the more I hold on to that little bit of humanity and propriety, the higher my relative value becomes.

So, if you ask me to choose between the two types of people Yiren mentioned, my answer is very clear. I will build Agent, and this article itself is a product of the collaboration of my tool chain. But there are too many engineers in the market who know how to use Agent better than me, and there are so many that I can’t possibly make a difference with this. What is really scarce in the market and cannot be stolen by others is this person and the trust he has accumulated over the past twenty years. With this combination, I am the only one in the world.

For a one-man company, IP is never a choice of two roads. It is the path that you can only walk on your own and cannot be copied by others.

Sixty percent means pulling yourself away, and one hundred percent means coming back with a new identity.

In the message area of Shengcai, a friend said something, and I wrote it down as soon as I saw it.

He said that from zero to sixty is from Human in the loop to Human out of the loop; from sixty to one hundred is from Human out of the loop and back to Human in the loop. The same person is in the circle, but the role this person plays is completely different.

I think this is the most critical level of the whole thing, but unfortunately most people think they have reached the top when they stop at 60.

People who are sixty-sixty percent do what they do is hand over repetitive work and make themselves disappear from the process. This step is important, but it only frees you up and does not create new value in itself. One hundred percent of people will go back to the process, but this time, they will no longer touch the execution. He came back to make judgments: whether this question should be answered, where to stand, who this passage is addressed to, and which things that seemed worthy of doing should not be touched at all.

In other words, what machines take away is labor, but what remains should be taste and judgment. These two things are exactly what Agent cannot answer or take away. Agentization is never a goal, it is a means; the goal is to release one’s judgment and creativity from trivial matters and invest them in truly high-value areas.

Because of this, my view on efficiency is different from that of many people. Yiren said that the rhythm of the new era is that when you come to me to discuss, the product has been made, and the discussion is about the finished product, not the idea. I agree with this, speed is indeed a barrier. But when it comes to speed, what a one-person company needs to do is subtraction, not addition. Let me tell you why.

The biggest risk is not insufficient production capacity, but too strong production capacity

Here, I want to diverge from Yiren’s discussion. This is also what I want to leave you most when writing this article.

He gave two suggestions to friends working on projects. First, stop spending a lot of time on project selection. Make it first and throw it into the market, and let real feedback tell you the answer, because your analysis cannot be more accurate than market feedback. Second, your competitors are also accelerating. For the same idea, others may go online in three days. Speed ​​itself is a barrier.

For a person who executes many small projects at the same time and relies on a lot of trial and error to find opportunities, these two suggestions are correct. But for a one-person company, for a person who is betting everything on a single credible personality, there may be a deep trap hidden behind these two suggestions.

Because my core asset is not a certain project, but my credibility as a person.

For example, when my tool chain is fast enough to be able to launch several things a day, the biggest temptation is to regard what I can do wrong as what I should do, and throw everything out and try it out. But please keep in mind that every half-finished product and every inaccurate public statement dilutes not the data of a certain project, but the readers’ trust in my name. When the data of the project is lost, it can be updated next month; but if trust is something like this, once it breaks down, it often takes several years to recover it, and sometimes it cannot be recovered at all.

I feel deeply about it myself. I’ve been anxious before. When I saw a hot news item, I wanted to quickly catch up on it and publish an article to get traffic. Only later did I realize that I hadn’t thought through it at all. After that time, I truly understood that for a person who relies on trust, every word you send out is an endorsement of your personality. You cannot use the readers’ trust to do A/B testing.

So, I drew a very clear line for myself. For high-speed trial and error, please stay behind the scenes of products and tools, where you can iterate as many times as you like, and no one will believe you any less because of it. But when speaking out, you still need to think clearly before taking action. With AI and machines, I can run faster, but speed is not used to test randomly in front of readers. Speed ​​is used to think about things that should be thought about more thoroughly and deeper than before. Decision-making speed is important, but for a one-person company or a small team, the quality of decision-making always comes before speed.

Don’t completely remove yourself from the process

There is another boundary that I guard very tightly.

Yiren said that you need to completely replace yourself from your current job. You should not be in your job. The Agent has completed the delivery and you are not in the process. For most of the execution links, I agree with both hands. It doesn’t matter whether I am involved in itinerary arrangement, file archiving, format conversion or deployment. The sooner I get rid of it, the better.

But for a writer who relies on words and ideas, there is one thing I will not hand over, and that is writing itself.

Because writing is not just output for me, it is two things at the same time. It is the source of value. What readers want is what I think and say. It is also the material for my next article. Only in the process of writing will I be forced to come up with new ideas and discover what I really care about. If I automate all my writing, I will save effort in the short term, but in the long term, I will completely wipe out everything I want to write in the future. A person who no longer thinks with his own hands will soon have nothing to give to others.

So this boundary is actually very simple, so simple that it can be used as a mantra: if it is repeated, you can leave it to the agent; if you want to make a judgment and feel, please keep it to yourself. The hard thing is not to distinguish what is repetitive, but to resist the urge to automate everything and honestly admit that a little slowness is worth it.

A trusted personality, plus a group of Agent assistants

As I write this, the answer to what a one-person company should look like in the Agent era becomes clearer and clearer in my mind.

It’s not one person and a bunch of tools. That version is only sixty points, and will become increasingly worthless, because everyone has the tools and production capacity.

What it really looks like is a trusted personality with a group of Agent assistants. Tools are responsible for production capacity and liberating people from repetition; and people are responsible for that thing that machines can never do, making another person willing to believe in you, willing to put time and money in your hands, and willing to follow you for a while. Yiren left a question for his team at the end of his sharing. I think it is equally sharp for every one-person company or small team. He asked: If all your execution work is taken over by Agent tomorrow, what else can you contribute to users, the team, and the business? If you can’t answer it, this is the question you need to solve most now.

Let me take this question a little closer to the direction of a one-person company and give it to you who are reading this article:

If all the things that can be done by machines for you tomorrow are done by machines… then, what is left in you that is worthy of others’ trust?

Well, the sooner you think about this issue clearly, the sooner you will know where you should really spend your time.

---