Original · Unique Research · 2025-12-03
Historical edition: The observations, commercial results, forecasts and speaker opinions below are those reported at the 2025 event, not independently verified current performance. Here, IP means a recognizable personal or business brand, rather than merely legal intellectual-property rights. The source itself conjectures that “Gym I3” meant Gemini; that conjecture is retained as a source-editor note, not a verified identification. The source transcript dates Yu’e Bao to 2011; that date is preserved as the speaker’s statement rather than independently endorsed.
If you drew a heat map of the sales world this year, the hottest area would undoubtedly carry three words: people + AI.
It is not about AI replacing salespeople, but about who uses AI first to amplify people.
At a roundtable during the 2025 Unique Awards Beijing, several guests who genuinely work on the front lines of selling products, closing deals, and driving growth broke the issue down thoroughly: Wu Bin, CEO of Infimind, which helps e-commerce companies grow through content; Heiqiang, founder of Chaohui AI, who acts as an IP manager; Li Shouguo, CEO of Beta Data, which provides sales practice and Copilot tools; Wang Zhaoqi, co-founder of Xiaobangbang, which serves one million salespeople; and Chen Lingshan, Investment and Financing Director at Vidau Technology, which works on intelligent marketing and Vidau AI. The moderator was Lushan, or Virgil, AI Overseas Industry Business Director at Cyberklick, who has long worked on the front line of Chinese AI companies expanding globally.
The questions they discussed were highly practical: what exactly is changing in sales, marketing, and personal IP in the AI era?
Who is being eliminated, and who is becoming a money-printing machine?
I. From Hiring 50 More Salespeople to Cutting 50 Salespeople
Wu Bin once gave Infimind's internal strategy a definition: use AI to build IP, and use IP to sell AI.
The clearest figures are these: last year they had more than 70 salespeople; this year they retained only 20. The team shrank by more than half, yet revenue could reach 2 to 3 times last year's level.
Under a traditional playbook, the company might have kept hiring 200 or 300 salespeople, visiting hundreds of customers each day and wearing out shoe leather on the road. Instead, he now posts one video and hosts one livestream every day, allowing leads to flow in on their own—dozens, hundreds, or even thousands at a time.
What gives him that confidence?
Because they work in e-commerce and understand one principle better than anyone: content is king.
A consumer purchase never begins at the shelf; it begins with a piece of content that appears in someone's feed: an image, a video, or a livestream. That is how interest is planted.
What Infimind does is simple but difficult: it uses AI to help merchants generate large volumes of content on Taobao, JD.com, Douyin, Xiaohongshu, TikTok, and Ins, while using the same AI tools to build exposure for the company and its founder.
The founder himself consequently became a creator with more than 600,000 followers. The company shifted from hiring salespeople to cultivating IP, and the top of the sales funnel was no longer a cold-call list but content reach.
You will find that many companies claim to provide MarTech while using none themselves; those willing to treat themselves as guinea pigs often run the fastest.
II. Products Are Being Sold, and People Are Being Seen
But "content is king" is only half the story.
Heiqiang sees things very differently: in his view, the market is flooded with AI solutions centered on products—intelligent media buying, automatic product-image generation, and automatic product-page copywriting.
Well-funded large companies have already pushed the "selling products" dimension to its limit.
What remains unmet?
He found that many companies' real pain point is not their products but their people: there are countless solutions for products, yet the people inside the company—the owner, expert, or consultant—remain unseen.
They therefore chose an angle that sounds a little like quietly building strength: becoming the managers behind the stars.
Their aim is not to make your products look prettier, but to turn the capable people behind them into credible IP.
He makes an important conceptual distinction:
Content is the starting point of traffic;
IP is the lever for traffic.
Post the same breakfast photo and Lei Jun's becomes a signal watched by investors; ours will probably interest only the bathroom scale.
The difference is not the photo, but who posted it.
Chaohui AI therefore focuses on industries with long cycles, slow decisions, and high transaction values—for example, bespoke clothing, complex To B services, and products too expensive to purchase with one click.
It then uses AI + content + IP to do one thing: build trust.
These businesses share a common feature:
Customers will not pay hundreds of thousands after seeing a single advertisement;
They repeatedly consume the content, assess the person, and silently ask themselves: can this person be trusted?
Heiqiang's conclusion is that AI is not the protagonist in this chain; it is the amplifier.
It amplifies a person's face, opinions, and credibility. That is why his company's slogan is so direct: AI+IP = a money-printing machine for business.
While everyone else discusses how to optimize media-buying models, some people are quietly working on something more fundamental.
They are making the owner AI-enabled, IP-driven, and digital,
And moving the meeting room where business is discussed into short videos and livestreams.
III. AI Is Rewriting Sales Training and Delivery
If Infimind and Chaohui AI are talking about front-end customer acquisition, Beta Data and Xiaobangbang are addressing the other side of the same question: how can AI elevate a sales team's fighting strength and delivery capability to a new level?
Li Shouguo builds sales practice tools. He uses a vivid analogy: sales skills are like swimming—you can attend countless lessons beside the pool, but you will never learn if you do not get into the water.
In 2021 and 2022, they tried using the previous generation of models for virtual-customer practice.
The result was terrible: the machine responded coldly, merely detected keywords, and could not understand context at all.
Salespeople quickly became frustrated and refused to use it. The situation changed after the new generation of large language models arrived.
The machine could finally behave like a customer with some human character: it could respond, change direction, and raise unexpected questions.
In complex sales fields such as real estate, medical aesthetics, and insurance, the time needed for a newcomer to grow from a trainee into someone ready to meet customers fell from 3 to 6 months
To two or three weeks.
This is only the first step, called efficiency improvement:
It saves training travel and instructor expenses, improving capital efficiency by several percentage points.
What excites Li Shouguo more is the second step: Copilot.
In the future, every authorized customer-call recording should become more than an archive. It should become fuel for AI, enabling it to whisper real-time guidance into a salesperson's ear during the next call: this sentence could be phrased differently; this question could be followed up in another way; the hesitation may reflect the customer's family structure or financial concerns.
Crude customer segmentation will gradually be abandoned in favor of returning to individual people—their family background, financial situation, communication style, and personality preferences—and then providing one-to-one communication-strategy recommendations.
In his view, the first step's efficiency improvement is a quantitative change of 10 times or 20 times, while the second-step Copilot is a change at the level of the species.
What Wang Zhaoqi sees is a break in the delivery layer.
Xiaobangbang has spent ten years building CRM and now reaches more than one million salespeople.
His feeling about the past ten years is: we spent a decade selling digital health supplements.
Companies could survive without them, and using them did not seem to create any astonishing change.
That changed when he fully embraced AI this year and launched the AI Benchmark 100 program: every month, it helps 100 companies put AI into real use.
The most obvious change occurred in delivery. Previously, one customer-success employee had to watch one customer for an entire month, performing the work of a highly paid customer-service agent at extremely high cost. After AI was used to reconstruct the delivery process, one CS employee could serve 30 companies with expert-level capability.
The result: customer satisfaction and referral rates soared,
And the first cohort of the AI Benchmark 100 program achieved a natural customer-referral rate of 40%.
They began to imagine a goal that sounds exaggerated but is not beyond reach.
Even if customers fail to achieve the result themselves, they will still thank us.
That is because he no longer sells a tool, but the result created after delivery is added.
More interestingly, he also took the IP route—livestreaming for 2 hours every day, painful as it is, and persisting. Many viewers coming through the livestream are small and medium-sized businesses with no digital foundation at all.
Yet this is precisely the group that most needs the complete package of AI + tools + delivery, rather than a cold list of features.
Vidau takes the perspective of a marketing-services provider and connects the front and back ends. Vidau AI shortens a material-production cycle that originally took a week by 30%–50%, transforming editing that cost several hundred yuan per piece into batch generation on the platform. It then feeds all campaign-performance data back into the system, allowing AI to learn which material converts best on which platform and with which audience, forming a closed loop from materials to the media-buying engine.
Once this chain is connected, AI is no longer merely a miracle tool for cutting costs; it begins to become an intelligent hub for continuous optimization.
IV. Will AI Take Away Salespeople's Livelihoods?
No one at the event avoided this question.
Lushan's concern was highly practical: on one side are internally developed digital humans and a paid-acquisition Agent; on the other are creative and optimization teams numbering 200 or 300 people. If AI can one day deliver materials in one step, run a Campaign automatically, and adjust bids automatically, what will people do?
Wu Bin's test is simple:
If every iteration of a foundation model makes you more anxious rather than happier, the business is in danger.
Editors feel nervous when they see Sora, and front-end developers feel nervous when they see Gemini; that anxiety shows that the layer where they work is relatively close to automation.
What should they do?
Either go more vertical, cover the entire chain, or build new barriers through IP and engineering capabilities.
Find the direction where stronger models make you happier instead, because they help you complete and improve work you previously could not finish or do well.
Heiqiang's answer sounds more like a story from the streets: people have their fate, and AI has its fate.
Tools evolve from hand planes to hoes and then to excavators; what does not change is that the people who know how to use them reach the ore faster.
In his view, only two kinds of people truly capture the dividends of this era: those with money, who can afford mistakes and try several paths, and those with courage, who dare to move first when they have no money and therefore seize the first opportunity.
He especially wants to correct one misconception: building IP does not mean gyrating for attention as an internet celebrity. Internet celebrities cater to traffic; IP attracts customers.
He prefers to call himself a business owner rather than a creator, because he makes content to discuss business, not to wait for an awkward tip.
Li Shouguo views the issue as a timeline. Extrapolating from the current steep trajectory of foundation models, highly standardized roles such as customer service and telesales are likely to be replaced, while people should gradually retreat toward fields involving more aesthetics and emotional experience.
He cites internet finance as an example: from Yu'e Bao to various P2P platforms, loud voices once claimed they would overturn offline wealth management. More than a decade later, P2P has become a historical term, yet private-bank relationship managers and insurance brokers have not disappeared; their systems have instead matured.
The reason is simple: when faced with complex products and decisions, customers are not buying a product name, but an entire solution—the feeling that "I can hand this problem to you and rest easy."
In this process, language is only a small part. AI has great difficulty learning solely from corpora what lies behind a look, an action, or a silence. He therefore judges that in complex sales fields such as real estate, insurance, and medical aesthetics,
AI will find it very difficult to truly replace people over the next five to eight years.
Wang Zhaoqi uses an expensive lesson of his own to remind everyone not to worry excessively.
When Sora appeared, he concluded that image generation would be rebuilt, so he spent more than 300,000 yuan exploring an AI photo studio.
Now a search on Meituan shows that there are already more than 2,000 AI photo studios nationwide.
The form did not change; the logic of making money did.
He places greater weight on a statement from FanRuan's COO: in the short term, make money from enterprise human resources.
Find someone earning an annual salary of 500,000 who makes money through professional expertise, use AI to standardize 80% of that person's capabilities, and then sell it to more companies for 1,000 or 2,000 yuan.
Chen Lingshan offers another reminder: no matter how advanced technology becomes, people will still scramble to buy handcrafted Swiss watches.
Because people never pay only for functionality; they also pay for emotion, identity, and cultural symbols.
In a multilingual and multicultural market such as Southeast Asia, relying only on AI translation for marketing can easily produce the right words for the wrong people.
The effective approach is to let AI handle the automation and optimization it does best, while putting people who understand the local culture in front.
So rather than asking whether AI will replace salespeople, ask this: can you use AI to move yourself from the layer that can be replaced to one that is harder to replace?
V. A Few Sincere Words for Salespeople Who Want to Build IP
At the end of the discussion, Lushan raised a highly current question:
This seems like the best era for building personal IP. Should you get on board? How do you go from 0 to 1?
The guests gave very different answers, but all were useful.
Wu Bin offered a tool-oriented answer: imitate across fields by selecting an IP you respect and studying its playbook; persist positively, because seeing value and leads gives you the motivation to keep producing; make good use of AI rather than forcing yourself to handle everything from scriptwriting to editing; and collaborate whenever possible instead of fighting alone, with more co-creation and connection.
Heiqiang was more direct: IP is leverage, and the leverage created when the owner builds IP is far greater than when an individual salesperson does so alone.
He recommends putting department heads and company owners forward whenever possible, because customers trust the person who can make decisions.
Before creating content, there is an even more important choice: do you want to sell cheap goods everywhere, or expensive goods that stand tall? If the business itself is unclear, the harder you work on IP, the further you stray.
He repeatedly emphasizes one sentence: do business, not internet celebrity.
Every piece of content should be sales material, not an attempt to amuse the algorithm.
Li Shouguo's advice is simple and forceful: to make your way in the world, the most important thing is to come out first.
Do not be intimidated by words such as persona and positioning. Start filming, and get used to putting yourself in public. That is itself a long-term trend.
Wang Zhaoqi offers a dissenting view: he does not recommend that everyone begin by rushing to build IP,
But instead advises getting the business running smoothly before considering how IP can amplify it. He learned this from experience: in 2021, he livestreamed on DingTalk for one year and gained hundreds of thousands of followers, but generated little business.
He later returned to teaching offline courses without any personal-IP halo and earned real money instead.
Starting from persona, Chen Lingshan gives IP a very practical definition: IP cannot make everyone like you; instead, you must accept having sharp edges.
You must know clearly whom you want to serve and what position you represent, and remain black-and-white in your views. Only then can content penetrate and match products and customers more easily. In one sentence: do not treat IP as a traffic skin; treat it as a business role operated for the long term.
Conclusion: What Will the Next Generation of Sales Organizations Look Like?
Connecting the discussion reveals a faint outline: future sales organizations may no longer have such large field-sales armies or such densely packed telephone rooms;
Their core assets will become three things:
An IP—often the founder or a core executive—willing to step forward, express a view, and be continually amplified;
A customer-acquisition, training, and delivery system reconstructed with AI;
And a group of people willing to learn new tools and partner with AI.
AI will not automatically improve a business;
It merely propels further those who were already smart, willing to experiment, and bold enough to show their faces.
So from today's vantage point, whether you are a founder, sales leader, or frontline salesperson, you need to ask yourself three highly specific questions:
If half your leads came from content starting tomorrow, would you be ready?
If your team could use AI to cut its training cycle in half, how would you design the process?
If one person had to be placed front and center by AI + IP, would that person be you?
This may be the starting point for reflection that Marketing and Sales: AI-Empowered Sales and IP Building
Seeks to leave with every practitioner.
More Details from the Conversation
Part One: Guest Introductions and How They Began Combining Their Work with AI
Lushan: I am Lushan, the business lead for the AI sector at Cyberklick. Friends who have worked on global-expansion projects before should be familiar with Cyberklick. We were among the first integrated marketing-service providers in China to help Chinese companies expand overseas. We began serving AI clients in 2023 and can say with confidence and pride that we may be the integrated marketing-service provider in China that understands AI best. Look at the Top rankings for global-bound products: at least 70%-80% of them are companies we helped grow from 0 to 1 overseas.
In line with today's topic, we as a marketing-services provider also began using AI very early to solve problems arising in actual business. For example, because filming advertisements overseas was too expensive, we developed our own AI digital-human project, KreadoAI, as well as an Agent product for small and medium-sized enterprises to acquire traffic through marketing campaigns.
Today's guests are all highly successful examples of combining AI technology with their own businesses. I would like each guest to introduce what you currently do and explain what opportunity or inspiration led you to combine AI technology with marketing and IP building.
Wu Bin: Hello, everyone. I am Wu Bin from Infimind. We use AI to empower e-commerce, providing multiple tools for e-commerce that help merchants create images and videos. Because we are in the software industry and previously felt that sales and customer acquisition faced major barriers, we began using our own tools for marketing very early.
I have also built an IP, so in addition to being a founder, I am now a creator with more than 600,000 followers across the internet. I mainly promote the company by serving as an Influencer in China. We therefore both provide AI tools and use those AI tools ourselves to support our own marketing. That is roughly the combination of roles I have.
Heiqiang: Hello, everyone. I am Heiqiang, founder of Chaohui AI. My approach differs from Mr. Wu's: his solution mainly centers on products, while mine centers on people. We therefore found an angle that could be called quietly building strength. We observed that large companies and well-funded players were all targeting the product-selling dimension with the greatest number of market scenarios and had built a great many solutions around products.
But while serving customers, we found that another dimension remained unmet—besides a company's products, its people also need to be seen. Mr. Wu is the star; we are the managers behind the star. Our solution primarily helps IP creators with corresponding content services and AI content tools.
Li Shouguo: Hello, everyone. I am Li Shouguo from Beta Data. We are building a sales-skills practice tool. Sales communication is similar to swimming or fitness: how much the teacher explains does not matter nearly as much as getting into the water to practice. No matter how many techniques are taught beside the pool, you will certainly not learn if you never swim. Sales skills work the same way.
In 2021 and 2022, we tried using the previous generation of models to build a tool that would free people from this work by offering Role-play practice, but we failed. It was extremely mechanical and salespeople were unwilling to practice with it. The same sentence must be delivered differently to different people, but the machine could not recognize semantics; it could only match keywords through Keyword Spotting. That was the weakness of the previous generation of models.
After the new generation of large language models appeared, we saw a possible opportunity and tried rebuilding the product with them. In one sentence, we use AI to simulate virtual customers so salespeople can practice with them and thereby improve their communication skills. The product is now performing relatively well in finance, telecommunications, and some offline-education fields.
Wang Zhaoqi: Hello, everyone. I am Wang Zhaoqi from Xiaobangbang. The speaker on the previous Panel said that their company had operated for more than a decade; we have also been operating for ten years. Today, about 1 million salespeople work on Xiaobangbang every month.
Although we have worked for ten years, those ten years were actually quite frustrating. It felt like selling digital health supplements for a decade—companies could survive without taking them, and taking them did not seem to make much difference. But everything changed this year. Starting in January, I embraced AI technology and launched the AI Benchmark 100 program, helping 100 companies succeed with AI every month. I finally felt that I had found a specific remedy for the stubborn disease of enterprise growth. Every company participating in my program increases its sales conversion rate by more than 20% within one or two months. These cases are all highly concrete.
Chen Lingshan: Hello, everyone. I am Chen Lingshan from Vidau Technology. We differ from other companies because we are not purely an AI company; we have worked in intelligent marketing for more than ten years.
We entered AI after seeing that it could empower many parts of the industry, leading us to develop our product Vidau AI. It provides substantial support for creative-material delivery and intelligent quantitative campaign delivery. Comparing performance before and after the use of AI, the efficiency of our intelligent services improved by 20% to 30%. One part of our business consists of the intelligent marketing services we have provided for ten years, while the other is Vidau AI.
Part Two: Changes AI Is Bringing to Marketing and IP Models
Lushan: As just mentioned, sales models and channels have changed dramatically since the development of AI technology. Cyberklick represents 30 or 40 domestic and overseas media outlets. In budget terms, the structure has changed greatly compared with the earliest global-bound e-commerce and pan-entertainment products. Clients are now more inclined toward stable SEM traffic, for example, while also emphasizing brand building through channels such as Reddit and Twitter. Our AI products manage at least tens of millions of US dollars in annual transaction volume.
I would like to ask everyone: as AI technology develops rapidly, what enormous changes are taking place compared with traditional marketing and IP-brand-building models? What concrete cases demonstrate gains in efficiency or conversion? I hope you can share specific examples that will inspire the audience.
Wu Bin: Let me begin with a somewhat subjective internal definition. Our company's current playbook is: use AI to build IP, then use IP to sell AI.
We previously hired many field-sales and other sales employees. Last year we may have had more than 70; this year we reduced the team to 20, cutting 50 people. Yet this year's revenue could reach roughly 2 to 3 times last year's. That is because our exposure and influence have gradually increased. Previously, we might have needed 200 or 300 salespeople just to visit several hundred customers a day. Now, when I post one video and host one livestream every day, we receive dozens, hundreds, or even thousands of leads. Efficiency has improved.
In e-commerce we say content is king. Many consumers place orders because an image, video, or livestream has planted the idea. Now that AI can generate content, we also want to help merchants produce large volumes of content on Taobao, JD.com, Douyin, Xiaohongshu, TikTok, and Ins, using a matrix of accounts to amplify it and drive conversions. We apply the same playbook ourselves, creating content for clients and for our own products. I find it rather incongruous when a MarTech company does not use MarTech itself.
Heiqiang: The moderator asked what has changed since AI arrived. I might answer from another dimension: first, let us consider what has not changed in this business.
For me, AI is still first and foremost a tool. Although ours is an artificial-intelligence company, I still believe today that AI cannot completely solve customers' business problems. It should be people + intelligence. Some things cannot change, such as the logic of closing a sale: from knowing you, to recognizing your value, to buying from you. That logic cannot change. If someone does not even know you, that person is unlikely to buy from you directly.
The first constant is this logic. We need traffic, and without content there is no traffic, so I must produce a great deal of content. But why can Lei Jun post about eating breakfast on Weibo and trend, while Heiqiang cannot? Unless I eat breakfast naked and turn it into a public incident. It is because he has IP. We discovered that content is the starting point of traffic, while IP is the lever for traffic.
I strongly agree with Mr. Wu's point that AI must be used to sell products to customers. We selected one type of customer characterized by long cycles, slow decisions, and high transaction values. Today, for example, I visited a bespoke-clothing customer that sells one outfit for 50,000 yuan. The company has produced a great deal of content but cannot persuade customers to pay because they do not trust it.
Our work therefore centers on one phrase: use AI together with content-based IP to build trust. AI continually strengthens and expands the opportunity to establish trust within this process.
What is the second constant? It is the relationship between people, so show your face. The owner remains constant, while every employee across the company may change. Mr. Wu cut 50 salespeople; I cut ten people from the marketing department and eliminated the sales team too, yet the business multiplied several times. Why? Because the owner is the constant in the company. I therefore need to make the owner AI-enabled, IP-driven, and digital. That ultimately becomes our current face-centered IP+AI logic. We have a slogan: AI+IP = a business money-printing machine.
Li Shouguo: Let us return to our area of expertise—building sales teams.
The first step is tools and efficiency. When training newcomers in complex sales fields such as real estate, medical aesthetics, and insurance, for example, it originally took 3 to 6 months for them to mature. With virtual-digital-human training, they can qualify to meet customers in two or three weeks. In cost terms, we help clients save direct training expenses such as travel and instructor fees, improving capital efficiency by about 3%-5%.
The second step is intelligence. It returns to what AI does best—Copilot assistance. With authorized customer recordings, for example, AI should be able to provide real-time recommendations on sales strategy. Current customer-profile segmentation remains rather crude; the next step should return to each specific person—their family background, financial situation, communication characteristics, and personality—and provide one-to-one communication training and sales-strategy assistance.
I think the first step's efficiency gain may be only 10 times or 20 times. At the second-step Copilot stage, however, the effect is truly nuclear; it becomes an entirely different species.
Wang Zhaoqi: Unfortunately, the previous two guests either cut sales or cut marketing, while what I do is precisely to empower salespeople.
After AI arrived, the greatest change occurred in delivery. Previously, one customer-success colleague needed one month to deliver for one customer; the cost was extremely high, and the work was essentially customer service. After AI arrived, we used it to empower the entire delivery system. One CS colleague can serve customers like an expert and handle around 30 in one month, producing a huge efficiency improvement.
The result was a major increase in customer satisfaction and referral rates. In the first cohort of my AI Benchmark 100 program, the natural customer-referral rate reached 40%, which would have been unimaginable last year. Our review suggested that this metric could reach 100%—even if customers do not succeed, they will still thank me. What we sell is therefore no longer merely a tool, but the outcome after delivery is added.
In addition, I see an enormous market that remained dormant in the digital era. Penetration of CRM or general digital-efficiency tools in China may be below 1%. DingTalk has 16 million organizations, yet only about 300,000 paying companies. Another 98%–99% of customers remain to be developed.
I am now also a creator, livestreaming for 2 hours every day. It is very painful, but worthwhile. The AI Zhenben online course we created can enroll more than 50 students in offline classes every day, exceeding what a traditional training institution with 400 employees can achieve. The customers arriving through livestreams have no digital foundation at all. They need AI + tools + delivery even more urgently to help these real small and medium-sized enterprises, traditional manufacturers, and trading companies.
Chen Lingshan: One pain point we identified was material production. A traditional production process might take one week, with editing costing 300 to 500 yuan. By using the Vidau AI platform to generate materials in batches, we improved efficiency by 30%-50% and substantially reduced costs, helping small and medium-sized enterprises expand overseas more effectively.
We also created a closed loop for intelligent media buying. We monitor final campaign performance and feed it back into the platform. The system examines which platform performs best and matches materials with outcomes to form an intelligent media-buying engine. AI connects the entire path from front-end materials to the terminal engine, empowering not only ourselves but also making it available to third-party small and medium-sized clients as well (the source describes this as “open source,” without specifying licensing or code release).
Part Three: Will AI Replace Human Salespeople?
Lushan: The next topic is relatively pointed. We have AI digital humans and a paid-acquisition Agent, but we also have creative and optimization teams numbering 200 or 300 people. As AI becomes increasingly advanced, if it can deliver materials in one step, run a Campaign automatically, and adjust bids automatically, what should humans do? Several guests just discussed cutting teams, and as a service provider we are also developing AI to disrupt our own work. What should today's sales practitioners prepare for before the rain comes?
Wu Bin: Early this year, our company held a discussion: as foundation models iterate, should you become happier or more discouraged? That is a way to judge whether a business is worth pursuing.
Just as editors worry when Sora appears and front-end programmers become discouraged when Gemini appears—the source editor says the original wording was “Gym I3” and conjectures that it meant Gemini or a similar powerful coding model—if foundation-model iteration makes you worried, that may not be a good business. We must therefore either become more vertical, cover the full chain, or build new barriers.
Why are we determined to build IP in the To B sector? Previously, meeting clients required drinking and socializing around the table. After we built IP, owners come directly to ask, "Mr. Wu, do you have any good tools to recommend?" Naturally, I recommend ours.
We therefore focus on barriers beyond AI, such as IP and engineering barriers. We need to find the direction in which every foundation-model iteration makes us more delighted because it enables us to improve things we previously could not do well.
Heiqiang: Humans have their destiny, and AI has its destiny. Tools change—from hand planes to hoes to excavators—and people who can use new tools work more efficiently.
Only two kinds of people enjoy the fruits of this world:
People with money: they have many cards to play, can afford costly experimentation, and are not afraid to lose.
Courageous people: when no one has money, whoever is boldest takes the world first.
The same applies to AI and IP today. Looking back at the To B sector, Mr. Wu and I count as living fossils of the IP world—not because we are especially brilliant, but because no one else did it. The future will be the same: either compete with money or compete with courage.
Many business owners mistake building IP for gyrating as an internet celebrity. I do not want to introduce myself as a creator. Creators cater to traffic; I am an owner, and I am here to attract customers. Internet celebrities cater to traffic; IP attracts customers. I have simply moved my old meeting room to short-video platforms and online channels.
Will AI replace me? I believe AI will always be a tool. I watched I, Robot, and I do not believe AI can replace my joy, anger, sorrow, or happiness; nor can it replace my greed or lust. It should be my assistant. People may be alike yet have different destinies—the question is who takes the first step.
Li Shouguo: If we extrapolate from today's steep rate of foundation-model iteration, technical work such as customer service and telesales will indeed be replaced. People should retreat into areas such as aesthetics and emotional experience.
Here is an example: Yu'e Bao was born in 2011, internet finance surged in 2013, and VC investors asked whether I still wanted to empower offline businesses. 12 years have passed, P2P executives have gone to prison, yet private-bank relationship managers and insurance brokers remain firmly in place. The same is true overseas: LendingClub did not perform especially well, while complex broker systems became more developed.
Why? Because in complex service sectors, customers do not buy the product itself; they buy service, experience, solutions, and trust. AI can learn only language, but language is merely a small part of interpersonal communication. Eye contact, actions, and emotional expression cannot be solved by AI through learning language.
In complex sales and decision-making fields such as real estate, insurance, and medical aesthetics, AI will find it very difficult to replace people over the next 5-8 years. We therefore firmly believe that these sectors need salespeople, and we want to equip them with tools.
Wang Zhaoqi: My advice is not to indulge in groundless anxiety. Demand will always exist; only the methods and efficiency used to satisfy it will change.
Let me share a small story. When Sora appeared, I judged that image generation would be revolutionized, so I spent more than 300,000 yuan exploring an AI photo studio. Now a search on Meituan shows more than 2,000 AI photo studios nationwide. The form did not change; the logic of making money did.
To borrow a statement from FanRuan COO Mr. Shen: in the short term, make money from companies and from their human resources. Find someone near you who earns an annual salary of 500,000 yuan through service knowledge and expertise, use AI to imitate 80% of that person's capability, and then sell it to companies for a price in the thousands of yuan. This goes straight to the essence, and it is what I am doing too.
Chen Lingshan: I hope everyone will consider how AI and humans can coexist. No matter how advanced technology becomes, handmade Swiss watches will still have buyers. What humans deliver is more emotional value or deep, complex marketing.
AI cannot understand human emotions and cultural differences. In Southeast Asia's multilingual and multicultural environment, for example, marketing based purely on AI translation will not perform well. Local creators with local cultural knowledge can reach people more effectively. We should let AI do what it does best and let humans use their advantages in understanding emotions and culture to provide complex delivery.
Part Four: Advice on Building Personal IP
Lushan: One final topic. I have not yet built a personal IP myself, although I have advantages such as using AI to write scripts and a digital human to appear on camera, and this seems like an unprecedented opportunity. I would like each of you to offer one or two sentences of advice to salespeople who want to build personal IP: how can they go from 0 to 1, or amplify it further?
Wu Bin:
I offer everyone these 16 Chinese characters:
Cross-domain imitation: I imitate Lei Jun's consumer-product IP playbook.
Positive persistence: a CEO values results and can persist when leads appear.
Use tools well: use AI tools to edit video and write scripts.
Collaborate and co-create: everyone can work with me to build together.
Heiqiang: IP is leverage.
Have the owner do it: a salesperson building IP has less leverage than an owner showing their face. I recommend that the department head or company owner do it.
Think through the business first: do not work harder and harder in the wrong direction. An early-stage business must either blanket the market with cheap products or stand tall selling expensive ones, earning only the pocket money of wealthy people.
Do business, not internet celebrity: all IP content is sales material. Do not gyrate for traffic; build trust. The destination of traffic is private-domain relationships.
Li Shouguo: In complex sales, it is simple: to make your way in the world, the most important thing is to come out. Start filming first; this will certainly be a long-term trend.
Wang Zhaoqi: My advice is the opposite: do not build IP blindly. Build the business first, then build IP.
I learned from failure. In 2021, I livestreamed on DingTalk for one year and gained hundreds of thousands of followers, but generated no business. Later, I taught offline courses with no IP and only a business, and it worked extremely well instead.
Chen Lingshan: IP has one core truth: it cannot make everyone like you. IP has sharp edges; its positioning must be clear and its views explicit—black and white. Match the persona to the product and customer, and you can shape a strong IP.