Don't Turn Your Business Into Outsourcing

It is well known that the number of people who know what AI is is far smaller than the number of people who do not know what AI is. The number of people who can use AI well is also far smaller than the number of people who are anxious because of AI.

So in theory, people who truly know what AI can do should have a massive asymmetric advantage when providing services to those who do not know.

But the fact is often the opposite. Many software companies or teams still make their to B business a living hell. I have heard of quite a few contractors who clearly started out selling high-end things like AI transformation, digital transformation, and intelligentization, but in the end turned themselves into labor outsourcing.

They do whatever the client wants. Today they work on a knowledge base, tomorrow they tinker with AI customer service. The day after tomorrow, the client sees that digital employees look very sharp, and immediately starts researching digital employees. In the end, when the client’s boss asks, “How many people can I lay off next month?” the answer is, “I don’t know.” And they are immediately shown the door.

Why does this happen?

Because from the very beginning, you and your client are not understanding the same AI. The AI you think of is context management, agentic loops, and process reengineering. The AI your client thinks of is an omnipotent Q&A machine, a top-tier technology that can replace manpower when business is bad.

When you tell your client that you are good at AI, you and your client have actually not established any consensus. You are talking about capability; your client hears results.

Do not say you know AI very well. Say you know very well how to solve one specific business problem. Because only when you participate in defining the business problem and defining what result is to be achieved do you have bargaining power and a voice.

Otherwise, if the problem is entirely defined by the client and you are only responsible for implementation, you can only be a mover of human resources, an execution outsourcing firm wearing a high-tech coat.


Sell results, not capabilities.

The first step of doing business is to discuss results first, then solutions.

I think everyone doing software services should learn from consulting firms. Mature consulting firms design their own sales models. They never tell clients what they know; they tell clients how much cost they can help reduce, how much profit they can help increase, and what specific problems they can solve.

In other words, they do marketing, not sales.

sales_or_marketing.jpg

This is the most critical first step in starting a to B business: peddle your value proposition.

As said earlier, if you want bargaining power, you must participate in defining the final delivery. You cannot tell your client that you are good at fine-tuning and have strong orchestration skills in prompts and flows. Even if you say your GitHub has 10k stars, it is useless.

Your capabilities have nothing to do with your client. They know you are very AI, they know your technical ability is strong, and then what?

Showing off capabilities only satisfies your own vanity. What your client needs is your “care.”

“Do you want your performance to double?”

“Do you want to cut operating costs by 30% while improving efficiency by 50%?”

Echo the desire in your client’s heart, instead of stuffing them with a capability they do not know what to use for.

You should ask your client: What is this year’s growth rate? Are costs out of control? Is this year’s profit in line with expectations? Which part is swallowing the most money? What problems have already affected business results? Then tell them which part of this you can help them solve.

This is what you should truly sell: landing a clear business goal. It is also what people who are truly responsible for business results care about. So get the boss to accept a business goal related to your capabilities, and only then can you truly participate in defining the business and gain bargaining power and a voice.


Make management your force.

ChatGPT_Image_2026%E5%B9%B49%E6%9C%8817%E6%97%A5_17_47_48.png

The boss nods. You gain the boss’s authorization and support, start going in to do research and interviews, and gradually understand the client company’s business processes and various implementation details.

At this time, the HR director suddenly pulls you aside and asks, “Can this agent replace some of my subordinates, so I can relax a bit and manage fewer people?” How do you answer?

If you happily tell her, “Yes, I am here to help every department reduce costs and increase efficiency,” then congratulations. You may have personally planted a huge landmine for yourself. You are doomed.

The company’s goal is profit, but every department must first solve its own survival problem. Budget, headcount, scope of responsibility, and organizational status together determine why a department should continue to exist. So to judge a department, do not look at what it does every day. Look at what it relies on to prove its value.

Take HR, for example. Recruitment is just routine work. How many people were hired and how many offers were sent are only actions. What truly justifies the department’s existence is whether it can solve the company’s problems in manpower allocation, organizational efficiency, and talent supply.

If you foolishly tell the HR director, “After this feature goes live, you will not need to manage so many people,” what she hears is not “easier,” but that she has become unimportant.

She seems to be curious about your work, but in fact she is confirming your purpose. Management’s first consideration is whether, after this thing is done, they will become more important or more redundant.

The boss supporting you only means the project can be done. After you truly enter, data, processes, hidden rules buried in the business, and historical problems all depend on management’s cooperation. So do not easily put management on your opposite side.

This is also why it was said earlier that you must participate in defining the problem and the result. If even the problem is defined by the boss for you, when you enter, you will most likely only do dirty work for the boss.

But there are many ways to achieve the goal. Most of the time, what you should truly do is use your technical capabilities to enhance existing management’s ability to solve business problems.

For example, help the HR director connect business departments’ project goals, progress, and manpower needs, discover position gaps earlier, and improve manpower allocation, recruitment priorities, and talent selection. Or help the marketing department discover market opportunities faster, complete data cleaning and aggregation, and see channel performance, lead quality, and ad delivery efficiency more clearly.

Help management cut out time-wasting transactional work, so they can focus more on the key links that affect business outcomes. The better your project is done, the stronger their problem-solving ability becomes, and naturally the more willing they are to help you drive the project’s success.


Reject all business unrelated to commitments.

ChatGPT_Image_2026%E5%B9%B49%E6%9C%8817%E6%97%A5_17_47_53.png

At this point in the project, you have helped HR with business data interoperability and also connected the marketing director’s channel data. You should have become a popular person in the client company. But at this time, if HR wants you to also make an employee profile, do you do it? If the marketing director hopes you can also make an automated weekly report, what then?

My answer is: do not do it.

What you sell to the client has never been technical capability, but a clear business result. Since you promised a result from the beginning, your time, budget, and attention should serve this work goal.

Employee profiles, automated weekly reports, and other extra work may have value, but they may not serve the result you originally promised. And every additional requirement you take on consumes part of the resources originally meant to fulfill your commitment.

These chores not only make the project’s evaluation criteria increasingly vague, your role will also slide step by step from “the person who solves business problems” back to “the outsourcing that does whatever the boss wants.” If in the end the result you truly promised is not achieved, no one will accept your delivery just because you did another hundred or so extra features for them.

There is an old term in project management called scope creep.

When a project starts, the goal, scope, budget, time, and acceptance criteria are originally a set of things that match each other. Every new requirement you “conveniently” accept is equivalent to secretly expanding the project scope once without modifying the goal, increasing the budget, or extending the time. The more you do, the more meaningless the original project baseline becomes.

So proper project management, although it does not forbid clients from changing requirements, requires every requirement change to re-evaluate the goal the requirement serves, the added cost, whether the change affects the original delivery, and whether it affects the final acceptance criteria.

If it is worth doing, go through the process. Resolutely reject business unrelated to commitments.

You may not have complete project management capability, but you should have basic project management awareness.

Rejecting all business unrelated to commitments is not giving the client poor service. It is protecting your own delivery. The client pays for the result you promised, not for your superpowers.


Do not do outsourcing.

ChatGPT_Image_2026%E5%B9%B49%E6%9C%8817%E6%97%A5_17_47_57.png

Seeing this, you will realize that everything discussed earlier actually comes from your initial commitment to the client.

Why can’t you handle the boss? Because you never made a valuable commitment to the boss in the first place. Why do you get screwed in various ways by management? Because you lost control of the project, and in the end you can only become the boss’s white gloves.

Starting from a mistake, you can only be that outsourcing firm that sells labor, suffers painfully, and cannot make money.

Sometimes when I see some technical teams doing to B or FDE for the first time, and their “blood and tears accusations” summed up at great cost, I find it quite interesting. For them, it is an eye-opening experience. For other mature to B industries, it has long been commonplace.

There are thousands of forms of to B. They make themselves into people-serving labor outsourcing, relying entirely on selling work hours and headcount to support growth. Mature professional services organize their value around business goals. Under the goal, methods, resources, and people are merely means to achieve results.

This is why I advise you not to do outsourcing.

If the AI, agent, automation, workflow orchestration, and context management capabilities you possess can ultimately only provide people, work hours, and execution delivery according to the client’s requirements, then these AI technical capabilities are no different from CRUD.

You have merely put a high-tech coat on labor outsourcing, and literally demoted yourself to a mover of human resources.

Let's talk

Tell me what you think