Original · Unique Research · 2026-04-11
Editor's note: This is a translation of the original article and roundtable, including the author's analysis and the speakers' own claims. Company figures, product capabilities, medical and elder-care statements, and security opinions have not been independently verified. The reported 50% improvement is a combined business metric, not a clinical efficacy result. The source's 9073 elder-care formulation is reproduced as stated, not established here as measured population data. Fortune-telling claims are descriptions of a product pitch, not validated guidance. The remarks about electricity's invention and open-source software reflect the speakers' analogies or opinions, not verified historical or security conclusions. English renderings of names and titles follow the source and may not be official.
Unique Awards
What AI Truly Rewrites in Traditional Industries Is Not Efficiency, but Who Does the Work
When AI is no longer merely smarter software, but begins to take on judgment, execution, service, and relationship maintenance, what changes is the division of labor itself.
"
When AI enters an industry, the deepest change is not “working faster.” It is “a different set of actors doing the work.”
Over the past two years, almost every traditional industry has been talking about AI.
Pharmaceuticals, education, elder care, livestreaming e-commerce—even a seemingly traditional business such as bead bracelets is now discussing AI.
Frankly, this is commonplace now. At almost any forum or summit, you hear similar phrases: reduce costs and improve efficiency, enhance quality and innovate, upgrade intelligently, transform digitally.
Those terms are not wrong.
But after hearing them often enough, you gradually notice a problem:
Many people talk about AI transformation while still thinking about “installing one more tool.”
In their understanding, AI is at most smarter software, a more useful assistant, or an add-on that speeds up a process.
That is not necessarily wrong.
But I have always felt it is too shallow.
If you look at how companies actually use AI on the front line, the question is no longer whether one step can become 20% more efficient. It is this:
AI is beginning to take over many roles previously performed by people.
That sounds uncomfortable, but it is the crux of the issue.
When AI enters an industry, the deepest change is not “working faster.” It is “a different set of actors doing the work.”
At Hangzhou AI WEEK, a roundtable was titled “AI and Industry: Paths and Practices for Intelligent Transformation in Traditional Sectors.” Looking only at the title, it would be easy to turn it into a bland industry article: a few cases, some trends, standard conclusions, and a closing line about a promising future.
But the discussion was interesting. Although the speakers came from different fields, their remarks ultimately pointed to the same thing:
AI is not patching up old industries. It is rewriting their division of labor.
I think that matters.
The First Work Being Rewritten Is Not Physical Labor, but Experience-Based Labor
Many people may assume that when AI enters traditional industries, it first replaces standardized, repetitive work.
Theoretically, that makes sense.
But following the discussion further, you find that in practice, the first areas to shift are often those that previously relied heavily on experience.
Take biopharmaceuticals.
Zhejiang Kangbaiyu Biotech said it began a “digital upgrade” last year and already uses AI for sequence targeting in early research and development, moving from multiple targets toward more precise development. The overall effect was described as roughly 50% in cost reduction, efficiency, and quality improvement.
If you look only at that figure, it may seem like a standard story about AI improving efficiency.
But to me, what deserves attention is not the 50% itself, but where that 50% comes from.
How did critical parts of biopharmaceutical work function in the past?
They relied on a dense structure of talent: expert experience, accumulated team knowledge, long validation cycles, and repeated trial and error. Many judgments were not things software could easily handle.
But once AI enters target screening, early decisions, and sequence optimization, the change is not simply “doing it a little faster.”
Algorithms begin taking over part of the work that depended heavily on expert experience.
That is not efficiency improvement in the ordinary sense.
It changes the role of experience itself.
Detong Biotech was also direct. It has long worked in women's health, molecular and immunological testing, and pathological diagnosis, and is now collaborating with AI teams to apply AI to pathology.
Pathological diagnosis is a classic example of experience-intensive work. It is not simple recognition, but a combination of detailed judgment, prolonged training, and accumulated industry knowledge. Once AI enters that process, the logic is the same: Gradually hand some work previously dependent on human judgment to machines.
Returning to pharmaceuticals, I think many people still underestimate this shift.
We used to assume that industrial upgrading meant “machines handle physical labor, people handle thinking.”
Today, that is no longer the whole picture.
It looks more like this:
Machines first take over some judgment,
then some execution,
and ultimately force you to redefine which jobs still require people.
That begins to change how many industries organize themselves.
Beyond Automation: Outsourcing Relationships to AI
If pharmaceuticals illustrate a restructuring of judgment, the part of this roundtable that struck me as especially interesting was livestreaming and virtual interaction.
Hei Yu of Dayou Space said they are making each person's OpenClaw tangible through 3D and holographic presentation, allowing a livestreamer's AI counterpart to keep interacting, chatting, completing tasks, and maintaining connections with fans after the stream ends.
At first, many people may find a project like this flashy.
I initially felt that way too.
But think about the practical business problem it addresses, rather than the visual novelty:
A livestreamer's most valuable asset has been not content, but the network of relationships.
Why do fans pay?
Not just because of the merchandise or a particular stream, but because sustained interaction creates familiarity, trust, and participation.
The problem is that maintaining relationships is expensive.
It requires the streamer to remain available, the team to provide repeated support, and substantial effort in outreach and operations. The number of people you can reach is limited, as is the depth of those relationships. Much fan engagement ultimately remains shallow: announcements, events, communities, and replies. It looks lively, but is still far from deep conversion.
Once an AI counterpart takes on some of that work, things change.
It is no longer merely a customer-service system.
Nor is it a chatbot that can say a few pleasant things.
It begins to become a new kind of actor:
A relationship agent.
When you are offline, it is online.
When you lack time to maintain relationships with important fans, it does so for you.
When you can offer only shallow coverage, it can pursue deeper interaction.
Hei Yu was clear: The first wave of gains came from tools; the next comes from channels. Many industries still use AI to improve livestreaming or video efficiency, but that provides only shallow coverage. The larger opportunity is moving from shallow coverage to deep interaction and then deep conversion.
I thought that was a valuable observation.
Many people still understand AI as “a tool that helps you work,” but it is beginning to become “an intermediary connecting you to users.”
If AI can take over the intermediary relationship itself, the nature of the business changes.
Trust previously maintained by people becomes a system that can be replicated, scaled, and operated semi-automatically.
Operational capabilities previously maintained by a team become an execution chain AI can keep running.
This is not merely an efficiency revolution.
It is more like reproducing commercial relationships in a new form.
The Industries That Struggle to Explain Their Value May Be the Ones That Take Off
Another case at the roundtable was interesting and representative.
Yuanji Digital Intelligence makes “smart metaphysical hardware”—in plain terms, smart bead bracelets. After activating an App, users can see a daily fortune, receive clothing suggestions, and obtain a kind of “advice pouch,” with ongoing feedback around emotions, relationships, and romantic luck.
Some readers may wonder: Does this count as an AI application?
I would say that precisely this sort of direction deserves attention.
In many traditional industries, customers have never bought only a function.
When buying a bead bracelet, they often buy something to place their hopes in, not merely jewelry.
When buying companionship services, they often buy a sense of being understood, comforted, and answered, not simply a service.
When buying from a livestreamer, they may be buying not only a product but a sense of continuing connection.
In other words, much of the real demand in these industries is not for practical necessities, but for emotional support, explanations, and companionship.
Why was that difficult to scale in the past?
Because it depended too heavily on people.
People had to understand, respond, and sustain the relationship, which was difficult to standardize or scale.
But as AI matures, these may be the industries it can most readily absorb.
That is why elder-care settings are beginning to feature companion robots, patrol robots, doll-like robots, vital-sign monitoring, and connected sensing solutions. Xialinghui has even converted a traditional nursing home into an intelligent elder-care facility, turning a real environment into a space for product testing and data feedback.
On the surface, that looks like robots entering elder care.
But to me, it seems more like something else:
Industrializing emotional value that previously could only be provided by people.
I think that idea will become increasingly important.
If emotional value, companionship, and explanations can be industrialized, they become more than an experience upgrade. They become a new market that can be measured, replicated, and expanded.
Why One-Person Companies Suddenly Begin to Make Sense
Another term kept recurring at the roundtable: OPC, One Person Company.
A few years ago, discussion of that term still felt somewhat ahead of reality.
Now it feels different.
Many capabilities really are beginning to move to AI.
AI can perform market analysis.
AI can produce content.
An Agent can handle customer service.
An AI agent can manage pre-sales communication.
Even product design, project management, and understanding requirements can often begin with AI doing an initial pass.
It is not that people no longer matter. Rather:
Problems that once required team size can increasingly be solved through system capabilities.
That is why many people suddenly feel one-person companies may genuinely be viable.
Li Qinfeng put it directly: For highly capable individuals, this is the best of times. Traditional companies also benefit, because software used to be paid for by the seat, whereas increasingly it will be paid for by results.
The logic is straightforward.
Whether a business could work used to depend on whether you could assemble a team.
Increasingly, it depends on whether you can combine AI, use cases, demand, and delivery in a new way.
So the viability of an OPC does not come from people suddenly becoming stronger.
It comes from systems taking over steps that previously required teams.
It sounds uncomfortable, but it needs saying:
A team may no longer be an inherent advantage.
Sometimes weight, slowness, and multiple management layers become liabilities.
Do Not Get Carried Away: AI's Pitfalls Are Already Visible
One of the roundtable's most valuable parts was not everyone praising AI, but someone openly offering a caution.
Xiao Yineng, director of the Intelligent Education Laboratory at Peking University's Institute of Advanced Information Technology, said he was cautious about traditional industries deploying OpenClaw blindly. His reasoning was practical: Open source makes code public, which may allow code injection, Skill injection, backdoors, exposed keys, and data leaks in different versions.
He repeated a joke that was popular at the time:
First you pay to install OpenClaw, then discover you must pay even more to uninstall it.
That is not necessarily alarmism.
When a company installed software in the past, its biggest problem might have been that it was hard to use, too expensive, or ultimately unused.
But an Agent is different.
Once an Agent has permissions, accounts, and access to processes and customer data, it is no longer a static tool. It is an actor that can take action.
It can work for you, but it can also expose your system.
It can improve efficiency, but it can also put your most valuable data into someone else's hands.
The point is not that AI is incapable. It is that:
The more AI resembles an employee, the less you can treat it as merely a tool.
I think that matters.
One of the easiest mistakes for companies today is to believe they have embraced the future while skipping basic governance, security, permissions, and process design.
That is dangerous.
What Is AI Actually Transforming?
Ultimately, I think the most important thing to retain from this roundtable is not an individual case, but a deeper judgment.
Many people still ask:
Can AI really be applied in traditional industries?
That question certainly matters.
But following the discussion further reveals a more important one:
When AI becomes not just an assistant, but a judge, executor, service provider, and relationship manager, which roles in your industry still need to be played by people?
That is the real question.
Once we seriously answer it, industrial upgrading becomes more than “digital transformation.” It becomes a genuine reorganization of roles.
Existing jobs change.
Existing teams change.
Existing customer relationships change.
Existing business chains change.
Ultimately, AI is not simply patching up traditional industries.
It is forcing every industry to answer three questions again:
Who does the work?
How is it done?
And why must you be the one to do it?
This shift has only just begun.
More from the Conversation
Unique Awards · Hangzhou AI WEEK Trend Roundtable Panel
“AI and Industry: Paths and Practices for Intelligent Transformation in Traditional Sectors”
Guests:
Yang Wenjun, Chair of Zhejiang Kangbaiyu Biotech
Hua Shaobing, Founder and President of Hangzhou Detong Biotech
Guo Yuchen, Chair of Zhejiang Xialinghui Intelligent Health and Elder Care
Hei Yu, Founder of Dayou Space
Li Qinfeng, Founder of Yuanji Digital Intelligence
Xiao Yineng, Director of the Intelligent Education Laboratory at Peking University's Institute of Advanced Information Technology
Moderator: Zhang Xuguang, Chair of Hangzhou Juexingdao Artificial Intelligence
Zhang Xuguang: As you can see, our topic is AI and industry. Today's biggest challenge is how traditional industries transform and upgrade. Let's discuss the paths and practices of intelligent transformation: how we actually do it. Our guests have just been introduced. Chair Yang is one of Hangzhou's “Five Little Phoenixes” and has done extremely well in her field. But everyone faces a challenge: How can AI help our companies transform, reduce costs, improve efficiency and quality, and innovate? I am also director of the expert committee of the Hangzhou Association for the Promotion of the Technology and Cultural Industries. I run a company, conduct research, and supervise graduate students. So rather than talk too much myself, let's hear from these six entrepreneurs. First, let's invite one of today's “Five Little Phoenixes,” Ms. Yang Wenjun, chair of Kangbaiyu, to discuss the company's business and her thinking or practices around AI.
Yang Wenjun: I'm Yang Wenjun, founder and CEO of Zhejiang Kangbaiyu Biotechnology. We have worked deeply in biopharmaceuticals for 11 years. Kangbaiyu's values are truth-seeking, integrity, and rigor. Our vision is to lead technological innovation and benefit human health by developing Class I innovative cell-based medicines ordinary people can afford. Our current product scope includes hematological tumors, solid tumors, autoimmune diseases, and broader health products. In this traditional field, our technology is advanced and innovative. With the arrival of AI, how can we use it to accelerate an innovative company? Last year, we joined Hangzhou's “digital upgrade” enterprises. Since then, we have expanded from individual applications to broader use, accelerating research and innovation. We use AI for sequence targeting early in research and development, moving from multiple targets toward greater precision and speeding development. We have also reduced costs. The combined improvement in cost, efficiency, and quality has reached around 50%, which I think is very good. We will continue working in innovation, combining technological innovation with AI to accelerate the industry's development together.
Hua Shaobing: I'm from Detong Biotech, which established itself in beautiful Hangzhou in 2010. We focus on molecular and immunological diagnostic testing for women's health. More than a thousand medical institutions nationwide currently use Detong products, and we have expanded overseas. Alongside traditional molecular and immunological testing products, we are developing AI-enabled testing applications for women's health and tumor detection. Working with relevant AI teams, we have already introduced an AI-enabled cytopathology diagnostic product. I think today's topic is excellent. AI will be applied to every aspect of life, and Detong will be no exception. Building on our molecular and immunological foundations, I believe AI can help us improve women's health and build better future technologies, markets, and a better world.
Guo Yuchen: Let me introduce the Xialinghui brand. We have worked deeply in health and elder care. One number to note is 9073: More than 90% of people in China receive care at home, 7% in the community, and only 3% enter institutions. The “hui” in Xialinghui refers to bringing together many services and products to provide home- and community-based elder-care activities. We have also received the national “Civilized Service for Respecting Older People” designation. Last year, we established Xialing Interactive Robotics. We have already produced affectionate companion robots, patrol robots, and doll-like companion robots in the “Shibao” and “Yuanbao” series. We will introduce many more home-safety and protection products, including integrated vital-sign monitoring and connected sensing solutions. To support the robotics company, we transformed one of our traditional nursing homes into the “Xialing Interactive Intelligent Elder-Care Home.” The first floor has a community canteen, the second a five-star day-activity center, and the third a dementia-care area. We can test all our products in that setting and collect data. We see this not just as a business and industry, but as work with social value and a sense of purpose.
Hei Yu: Dayou Space has a simple focus: helping celebrities and major livestreamers hold virtual concerts and large global fan gatherings. Recently, one project has become particularly popular because everyone is “raising lobsters.” Previously, those lobsters could only converse through chat, not speak with you by voice, much less appear visually in front of you. Dayou Space now aims to make everyone's OpenClaw tangible, appearing directly before them through 3D holographic presentation. If you have VR glasses, you can see your OpenClaw vividly in front of you, talking to you.
Zhang Xuguang: Does that enhance emotional value for customers?
Hei Yu: Yes. Previously, not every fan could chat directly with a livestreamer or add a celebrity as a friend. With this approach, many fans can.
Zhang Xuguang: How does that improve service for fans?
Hei Yu: In many ways. Celebrities or livestreamers may have resources and activities they want fans to join, but outreach was limited and usually text-only. In a virtual space, people can play games and complete tasks together. You can even give a Token directly as a gift to the streamer, or the streamer can give one to fans.
Zhang Xuguang: Who is the customer for the product you are developing?
Hei Yu: My first customer is a Korean esports star who will soon come to Hangzhou for a fan meeting.
Hei Yu: We see this as a future trend: how OpenClaw can represent its owner in better social communication and work collaboration. We believe a large virtual space can present that experience. That is one possible future, and why we are building this.
Li Qinfeng: I spent many years in very traditional industries and found transformation difficult. When ChatGPT appeared a couple of years ago, I discovered a new world. I could more easily break through and combine the core commercial secrets of other industries. I now make smart bead bracelets, or what you might call smart metaphysical hardware. We use AI throughout. Users activate the APP with their bracelet and can see today's fortune, clothing suggestions, and attractive visualizations. If someone says, “Why does everything seem to have gone badly for me recently?”, we can offer an advice pouch. If they say, “I feel people are always targeting me,” or “I want better luck in romance,” we can offer fortune-telling guidance. All of this is provided through AI. Potential partners are very pleased. I see AI as similar to electricity in its early days. Some may ask why a bead bracelet needs AI. You can make bracelets without it, but if you do not use it, someone else may replace you. We may displace traditional bracelets, because AI can extend their functions without limit.
Xiao Yineng: As Professor Zhang said, I personally offer a caution about “raising lobsters”—deploying OpenClaw—including in research and intelligent customer service. My core reasoning is that OpenClaw is an overseas Open Source model. Open source means its program source code is publicly visible to developers in the community. They can inject their own code into different versions or add a Skill package, which creates many potential backdoors. There is a joke: First you pay to install OpenClaw, the lobster; then you discover you must pay more to uninstall it. If you authorize accounts and passwords, private backend keys may be exposed through remotely controlled backdoor vulnerabilities. So I personally think non-specialists, particularly people in traditional industries, should wait and observe. It resembles the early Android era with its many installation packages. Android is also open source, and there were many malicious applications. Gradually, ecosystem constraints, industry self-regulation, standards, and antivirus companies such as Zhou Hongyi's 360 help make things compliant and standardized. My one-sentence answer to Professor Zhang is: Let things play out a little longer.
Yang Wenjun: I hope different industries can combine their strengths, support one another, and connect the upstream and downstream industry chain. We would welcome opportunities for deep collaboration with everyone here, and have been working on this recently. Moving from individual applications to broader deployment, we started with research and development, then quality control, production, and sales—the whole chain needs upgrading. We want to safeguard and realize the value of data, and work with more specialized, innovative AI partners to build a center of innovation together.
Hua Shaobing: AI has been very popular in recent years. I recall arriving at the University of California, San Francisco, for postdoctoral research in 1991. A professor's team there developed a molecular Docking algorithm, using Docking to identify drugs related to target-protein structures. Today, everyone knows AlphaFold and its protein-structure predictions, with compounds sought according to targets. Back then, a supercomputer had to run for a month or longer to identify a few molecules. Now it can be done in minutes. The substantial progression in computing from CPU to GPU to TPU has greatly supported innovative biopharmaceutical drug development. Beyond Detong, we established a biopharmaceutical innovation investment fund through an innovation-alliance structure. We have invested in a company developing innovative medicines with this kind of AI technology, and the results have been good. If you have innovative biopharmaceutical ideas, we can consider investing.
Of course, Detong focuses on in-vitro diagnostics for tumors and women's health. We also welcome collaboration on AI-enabled innovation in these areas. Our distribution channels reach more than a thousand medical institutions nationwide, with market positions in Southeast Asia, the Americas, and Africa as well. Strong market access is crucial to successful collaboration. Thank you.
Guo Yuchen: People may have a misconception about health and elder care and older consumers. Those with the most spending power, such as people born in the 1960s and 1970s, no longer think simply in terms of living in a traditional nursing home. They seek more emotional, spiritual, and social value. We established a long-term research project at the China Academy of Art, where our robots are designed. To give teachers more inspiration, we started a student club. Students offered many ideas, such as a cute pet robot with a little tail that could hook onto you and accompany you at any time, or matching based on the popular 16 personality types. We later decided that if we adopt a suggestion, we will share patents and sales profits with the students. That effectively makes each product series an OPC. This is an open era; we should connect with more people.
Hei Yu: Hangzhou is a center of innovation, with founders each using their own strengths. With the OpenClaw lobster wave, many friends around me are experimenting. I see a current trend: People use OpenClaw as a tool. The first wave uncovers value from tools, with opportunities for customized services across industries. But with any new technology, looking beyond the tool reveals channel value. Finding that channel value can create a major opportunity. Whether in livestreaming e-commerce or short videos, using OpenClaw to improve stream or video efficiency provides only shallow coverage: relationships remain shallow and conversion rates low. Deep interaction that produces high conversion was difficult before because streamers and teams had limited energy. Now OpenClaw knows your strategies and priorities and can help explore important customers and information more deeply. Why do we promote interaction in a shared online space? It allows your OpenClaw counterpart to connect with core fans through as much semi-real-time interaction as possible, producing deeper conversion. I strongly recommend looking at the next wave of gains—channel-based pricing. Moving from shallow coverage to deep interaction is our direction.
Li Qinfeng: I think this is the best era for both individual founders and companies. For highly capable individuals, demand is unlimited. Every industry will be changed by AI. Visit the customers you know best, and traditional industries will certainly have needs. For these individuals, it is like being an electrical designer 200 years ago when electricity was first invented—extremely valuable. Traditional companies also benefit. Software previously cost several thousand yuan per seat per year; now you pay for results. Many highly capable individuals and one-person companies are willing to serve you. Our own team is somewhat like an OPC. Previously, a project required many people for market analysis and development. Now it does not.
Xiao Yineng: My advice is that algorithms keep evolving and learning never ends. New terms keep appearing: ChatGPT, agents, embodied robots, OpenClaw, End-to-End technology, and “tokens,” as discussed by National Data Administration Director Liu Liehong. As long as founders embrace lifelong learning, they can keep discovering entrepreneurial opportunities in their era. Thank you.
Yang Wenjun: Hangzhou is a center for both AI innovation and life and health sciences. I look forward to working with you to bring AI into the upgrading of our traditional industries and empower innovative medicines.
Hua Shaobing: From accurate molecular-level testing to AI-enabled testing that is both accurate and far-reaching, Detong has always focused on one thing: women's health.
Guo Yuchen: We see this AI boom as a technological turning point. We hope everyone can help shape an era in which technology serves the good and humans coexist with robots.
Li Qinfeng: For individuals and companies that actively embrace AI, this is the best of times. Every kind of creativity can be expressed to a much greater degree.
Xiao Yineng: Algorithms have no endpoint, and neither does learning. Believe in the power of belief and the power of seeing. Let us embrace the future together.