The technology dividend is disappearing: domain expertise is the real moat

When OpenAI's GPT-4 can write better code than you, when Midjourney's drawing quality makes professional designers ashamed, when Claude can help you write logical business analysis reports-where is the moat of these professions as programmers, designers, and analysts? A recent article that sparked heated discussion on HackerNews gave the answer: No. The real moat has always been domain expertise, not the ability to write code itself.

This article comes from independent developer Bryan St. Brethorst, who uses his 15 years of experience in SaaS products as an example to explain why he would rather hire a developer who understands medical compliance but is average in programming than a person who is very skilled but knows nothing about the industry. This view received more than 1000 likes on HackerNews, triggering a rethink in the technology community about the fundamental question of "technical skills versus domain knowledge."

1. Event background and core controversy

In May 2026, Brethorst posted an article on his blog titled "Domain expertise has always been the real moat." The core argument of this article is that in software development, the real competitive advantage is never about how many programming languages you know and how many frameworks you master, but the depth of your understanding of a specific business area.

Brethorst gave a specific example. He once helped a medical SaaS company recruit developers and met two candidates during an interview. The first is a top computer science graduate who is proficient in Rust, Go, and TypeScript and has swiped 800 questions on LeetCode. The second is a graduate of a general state university with average programming skills, but has three years of working experience in the field of medical data processing. He is familiar with HIPAA compliance requirements and understands the particularities of medical data and medical workflow.

If you were CTO, who would you choose? Brethorst's answer is the second. His reason is that technical skills can be learned in 6 months, but professional knowledge in the medical field takes years to accumulate. What's more, when your product needs to deal with complex medical compliance scenarios, the "average" developer who understands HIPAA can see the compliance risks in the code at a glance, and the tech savvy may not even know it's a problem.

This view sparked heated debate on HackerNews. Proponents believe he is telling the truth, while critics believe he underestimates the importance of technical capabilities. But in any case, the discussion revealed a change that is happening: As AI tools lower the bar for technical skills, domain expertise is becoming a new scarce resource.

2. Why does domain professional knowledge become a moat

To understand why domain expertise has become more important now, we need to first understand what the concept of "moat" is. The moat originally referred to the canal outside the castle. It has been extended to the commercial field and refers to the lasting competitive advantage of a company in resisting attacks from competitors. Buffett's partner Charlie Munger likes to use a moat to evaluate the value of a company. He believes that a good company should have a wide and deep moat.

The traditional view is that the moat of technology companies comes from technological advantages-patented algorithms, proprietary data, and top engineers. But Brethorst pointed out that this view is losing ground in the AI era. When GPT-4 can pass medical examinations, write papers, and do data analysis, it is difficult to form a moat alone based on technical ability.

The core reason why domain expertise becomes a moat is that it has three characteristics that are difficult to replicate:

• Tacit Knowledge: Much of what domain experts know is difficult to express clearly in words. For example, a senior insurance claims clerk can see the abnormalities of a certain medical document at a glance, not because of what courses he has taken, but because of the intuition he has developed over the years in the industry. This kind of knowledge cannot be obtained through reading and must be accumulated through long-term practice.

• Trust Capital: In the B2B space, customers often prioritize whether the other party truly understands their business pain points when selecting suppliers. An implementation consultant who understands medical regulations can say,"What the director of your department cares most about is actually bed turnover." This insight can impress customers more than any technical solution.

• Network Effects: When you know enough about an area, you will attract more industry experts, customers, and partners around you. These relationships are themselves moats and will deepen over time.

To use a more straightforward metaphor: technical skills are hammers, and you can use hammers to show that you have basic abilities; but domain expertise is knowing where to hammer nails-everyone can use hammers, but knowing where to hammer is the real skill.

3. Why is this matter particularly important now

Brethorst's view resonates because it strikes a reality that is happening but many people are reluctant to admit: technical skills are devaluing.

Let's turn the time back to 2015. At that time, if you were proficient in using React, knew how to optimize SQL queries, and could build a microservice architecture, you were a hot spot in the talent market. But by 2026, these skills have become commonplace. More critically, AI tools are making these skills "civilian."

GitHub data shows that more than 1.3 million developers around the world will use GitHub Copilot in 2024, and this number will increase to 4 million in 2025. When AI can automatically complete code, automatically fix bugs, and automatically generate test cases, tasks that once took senior engineers hours to complete can now be done by junior engineers with AI tools.

What does this mean? This means that the supply of technical skills is increasing rapidly, while the growth rate of demand cannot keep up. When a skill becomes accessible to everyone, its value declines. This is not alarmist, it is a basic law of economics.

At the same time, the demand for domain expertise is rising. A report by the McKinsey Global Institute shows that by 2025, what companies lack most in digital transformation will not be technical talents, but "technical talents who understand business." The report surveyed 1500 large companies, and 78% of CEOs said their biggest headache was "we have technology, but we don't know what business problems to solve."

This gap between supply and demand is changing workplace rules. In the past, the promotion path for programmers was "junior engineer → senior engineer → technical director → CTO." Now, this path still exists, but another path is emerging-"domain experts → business-oriented technical experts → industry solution leaders."

In other words, technical skills are the ticket to admission, and field expertise is the key to making you stand out.

4. Industry data and trend support

Regarding the importance of domain expertise, there are several key data worthy of attention.

The first is the penetration rate of AI in various industries. According to the 2025 AI Index report released by the Stanford University Institute for Artificial Intelligence (HAI), the adoption rate of AI tools in knowledge-intensive industries is rising rapidly. In the financial sector, 62% of institutions are already using AI for risk assessment and investment decisions; in the medical sector, 48% of hospitals are using AI to assist diagnosis; in the legal field, the figure is 35%. What do these numbers mean? This means that AI is entering areas that were once considered to "require a high degree of expertise."

But there is an interesting paradox here: the popularity of AI has instead highlighted the value of experts in the human field. Because when AI can do basic work, people realize that the real problem is "how to make AI do things correctly." Take medical care as an example. AI can read CT films, but it does not know that behind this film is a 72-year-old rural man's last opportunity for examination; it can give diagnostic suggestions, but the one who ultimately signs it is the doctor who needs to take responsibility. The stronger the AI, the more precious human professional judgment will be.

The second is the change in developer skill needs. Stack Overflow publishes a developer survey every year, and its 2024 report shows that the ranking of skills most valued by employers has changed. In 2020, what employers value most is "programming language proficiency" and "framework experience." But by 2025, the first place will become "problem solving ability", the second place will be "domain knowledge understanding", and "programming language proficiency" has dropped to fifth place.

There is also data from LinkedIn's 2025 Talent Trends report. The report analyzed more than 10 million recruitment positions and found an interesting phenomenon: compound talents with both technical capabilities and industry experience have an average salary of 34% higher than purely technical talents, and the risk of unemployment is reduced by 40%. "Companies are moving from 'we will teach them the business' to 'they must bring their own business understanding,'"said LinkedIn's chief economist. "

Finally, look at the data in the field of entrepreneurship. Among the 200 startups in Y Combinator's Winter 2025 batch, 43% of the founders are "industry veterans who have turned into technology," compared with 22% five years ago. These founders may not be the top engineers, but their deep understanding of the industry allows them to make products that are closer to market needs. After Demo Day, the average amount raised by these companies was 27% higher than that of a purely technical team.

5. Actual implementation cases

Case 1: From actuary to founder of InsurTech

Wang Ming (pseudonym) has worked as an actuary for a large domestic insurance company for eight years. In 2023, he decided to start a business and build a commercial insurance price comparison platform for small and medium-sized enterprises. Technical background? Almost zero. He studied a little VBA in college, and the most complex code he had written in his life was Excel macros.

But he has one thing that most technology entrepreneurs don't have-a deep understanding of the insurance industry. He knows how insurance companies set prices, what is the biggest headache for insurance agents, and why small and medium-sized business owners always complain that insurance is "easy when buying but difficult when settling claims."

He found a technical partner, and the two spent more than half a year polishing the product. During this period, Wang Ming used his connections in the industry to pull up the API interfaces of three insurance companies-this is called "docking" in the industry. It sounds simple, but it requires understanding the insurance company's data format, business logic, and risk control rules. Wang Ming solved things that might take months for people with a purely technical background to figure out, in two weeks, because he knew the IT person in charge of the insurance company and knew how they handled data internally.

One year after the product was launched, the platform has more than 200,000 monthly active users, and the products of 12 insurance companies have been successfully connected. Wang Ming said: "If I had a purely technical background, I might make something that is very technical but no one uses. Because I understand insurance, I know what users really need. "

Case 2: The dilemma of domain experts in medical AI companies

Li Tao (pseudonym), CTO of a medical AI company in Shenzhen, has had a headache recently. The team he led has strong technical strength. The core members are from Tencent and Huawei. They have published many papers and have beautiful algorithm indicators. However, three years after the company was established, its products have never been able to make a success.

What is the problem? After reflection, Li Tao found that what they lacked was not technology, but domain knowledge. No one on the team has actually worked in the medical industry, and their understanding of the medical scene comes from literature and imagination. For example, the assisted diagnosis system they built has beautiful algorithm indicators, but in the actual clinical environment, doctors don't buy it at all-because the system does not understand the actual work flow of the hospital, does not know what kind of assistance doctors need in what scenarios, and knows that there may be reasonable medical explanations behind some "irregular" diagnosis and treatment records.

Later, the company adjusted its strategy and hired a retired director of the imaging department from a tertiary A hospital as a consultant, who was specifically responsible for "translating" clinical needs. What happened was a magical thing: the director wasn't here to change the code, he just told the engineers,"The first thought a doctor has when he sees a lung nodule is to judge whether it is benign or malignant, so your system should output this judgment first, not a pile of data." With just this sentence, the product logic has been changed, and doctors 'acceptance has suddenly improved.

Li Tao said: "It took us two years to understand that in the medical field, technology is the foundation, but it is definitely not a barrier. The real barrier is the depth of your understanding of the industry. "

6. Comparison between technical routes and deep cultivation in fields

There are several different paths for people who want to build a competitive advantage in the workplace or entrepreneurship. Each path has its advantages and disadvantages and is suitable for different scenarios and goals.

As can be seen from the table, there is no absolutely correct path, only the question of suitability or unsuitability. But there are several trends worth noting:

The risk of a purely technical route is that AI is rapidly compressing the value of technical work. It is not that technology is not important, but that technology alone is becoming increasingly difficult to constitute a competitive advantage. If you choose this path, you have to be at the top-not the top 10%, but the top 1%.

The advantage of the deep cultivation route in the field lies in its strong accumulation effect. Industry experience will increase over time, rather than diminish. But the disadvantage is that switching costs are high. If you choose the wrong industry, you may be very passive.

The T-shaped composite route looks the most ideal, but it is the most difficult to implement. Most people have difficulty reaching high levels in both technical and business directions. This road is more suitable for those who have clear industry goals and are willing to spend 5-10 years cultivating deeply.

7. Technical challenges and practical limitations

Having talked about the benefits of expertise in so many fields, we must honestly say that this path is not perfect.

The first challenge: The accumulation of domain expertise takes time

No one can become an industry expert overnight. Brethorst himself admitted that it took more than a decade to accumulate his knowledge in many fields. For newcomers to the workforce, this means that you can't expect rapid success based on "domain specialization." A more realistic approach may be to first have a solid technical foundation, and then gradually accumulate industry experience at work.

Second challenge: Domain expertise is difficult to verify

Programming ability can be verified through interview questions and code works, but domain expertise is difficult to quantify. You say you are an expert in the medical industry. How do I know how much you know? This leads to a problem: Employers often value "verifiable" technical capabilities rather than "hard-to-verify" industry knowledge when hiring. This is unfair to domain experts.

Third challenge: The industry itself may disappear

Choosing to delve deeply into a field means that you have to bear the risks of the rise and fall of that industry. Twenty years ago, paper media was a golden industry, and newspaper editing was an enviable profession. What now? Similarly, if you are deeply involved in a traditional industry, you should always pay attention to the changing trends of the industry. Domain expertise may be more difficult to transfer than technical skills.

Fourth challenge: AI is learning domain knowledge

Today's AI does lack real industry experience, but this gap is narrowing. GPT-4 can already pass the medical exam, and Claude can give reasonable advice in legal consulting scenarios. With the improvement of AI capabilities, relying solely on "I know this industry" may not be enough. What may be needed in "

Fifth challenge: Domain experts often have insufficient expression skills

This is the stereotype that many people with technical background have of domain experts, but it is real to some extent. A veteran who has been in the industry for 20 years may know a lot, but it is difficult to clearly express his knowledge and teach it to others. If you choose to deepen your field, communication and expression skills are equally important.

8. Who should pay attention to this trend

Inspiration from independent developers and entrepreneurs

If you are starting a SaaS product or tool business, Brethorst's views are worth considering carefully. The mistake many developers make is to have technical solutions first and then find problems. But the correct order should be: first find a real pain point, understand the business logic behind this pain point, and then use technical means to solve it. Technology is a means, not an end.

Specifically, before deciding what product to make, it is recommended to spend time doing industry research. Instead of reading research like industry reports, we really chat with people in the industry to understand how they work every day, what they complain about the most, and what things give them headaches but they get used to it. These insights often guide the direction of a product more than any technical ability.

Inspiration from corporate CTO and technical managers

For technology managers, the inspiration of this article is: Don't just look at technical indicators when recruiting. Interview a back-end engineer and ask him what he understands of your industry; interview a data engineer and ask him what problems he thinks he will encounter if you let your data scientists use this data for analysis. Those who can answer are often people who have really thought about business.

In addition, team composition is also important. Instead of pursuing that all employees be top technical talents, it is better to allocate some "business translation" roles in the team-people who understand both technology and can talk to business departments. They may not be the fastest people to write code, but they are often the key to reducing communication costs and improving project success rates.

Enlightenment from career planning for technical practitioners

If you are an engineer with 3-5 years of experience, now is a good time to think about your career direction. Ask yourself: Where do my technical capabilities stand in the market? If AI can replace 50% of my current job, what is the remaining 50%? What kind of capabilities are needed for those parts that "AI cannot replace"?

One possible direction is: Choose a vertical industry that you are interested in and start consciously accumulating industry knowledge. You don't need to quit your job to get an MBA, but you can make more friends with people in that industry and understand their job content and pain points. After a few years, you may become a scarce talent who "understands both technology and industry."

Inspiration from investors and headhunters

For investors, when evaluating an entrepreneurial team, the founder's industry experience may be a more important indicator than technical strength. Brethorst's views also resonate in investment circles: many investors now prefer to invest in a combination of "industry veterans + ordinary technical team" rather than a combination of "top technical talents + industry newcomers."

For headhunters, understanding this trend means redefining "quality candidates." Candidates with only technical keywords on their resumes may no longer be the most sought-after, while candidates with dual "technology + industry" training are becoming scarce resources in the market.

9. Prediction of future trends

I believe that the importance of domain expertise will continue to rise, but the form may change.

Short-term (1-3 years): AI-assisted domain knowledge equality

AI tools are lowering the threshold for domain expertise. A person without a medical background may also be able to understand basic medical concepts with the assistance of AI. But this does not mean that domain experts will be devalued. On the contrary-when everyone can obtain a "basic version" of domain knowledge,"in-depth industry insights" become more precious.

Medium term (3-5 years): A large number of composite positions in the "AI+ field" appear

Now we have seen positions such as "AI Product Manager" and "AI Solution Architect." In the future, more field expert roles specifically targeted at AI applications may emerge, such as "medical AI implementation experts" and "financial AI compliance consultants." The core competitiveness of these positions is not the technology itself, but the "ability to apply AI in specific fields."

Long-term (5-10 years): The form of domain expertise may change

Future domain experts may not be those who "know the most", but those who "know what AI doesn't know." As AI capabilities become stronger and stronger, the value of human experts may become more and more reflected in: knowing where the limitations of AI are, knowing when the output of AI cannot be trusted, and knowing how to take the lead when AI fails. These capabilities also require deep accumulation in fields, but the focus of accumulation may shift from "knowledge itself" to "understanding of knowledge boundaries."

X. Summary and action recommendations

Technical skills are the ticket, and field expertise is the key to standing out from the crowd. This truth is simple to say, but not many people really understand it and put it into action.

If you have time now, stop and think: How much time do you spend learning new technologies and how much time you spend understanding the industry you serve? The ratio of the two could be 8:2, it could be 5:5, or even 2:8. Different ratios are suitable for different career stages, but if you have a few years of technical accumulation, it may be time to invest more in your domain knowledge.

Specific action suggestions: Choose an industry that you are interested in or working in, find the most troublesome problem in that industry, and then understand why the problem exists and why it has not been solved. The answer often lies not in the technology itself, but in a deep understanding of the industry.