Original · Unique Research · 2026-04-08
Editor's note: This is a historical roundtable reported by Unique Research on April 8, 2026. Descriptions of company operations, customer cases, prices, adoption, awards and future AI capabilities are retained as claims or opinions from the source and its speakers, not independently verified findings. Nengcheng Culture, Dijuti and the personal account name AI Lipu She / AI Absurdity Society are provisional English renderings; M-verse and its AI+WISE explanation appear in the source. Monetary amounts without a stated currency remain unspecified. The comparison of “ten million Tokens” with “ten million in performance and ROI” does not specify consistent units or an ROI calculation. The source attributes the wheelchair-and-brain quotation to Hawking, but provides no primary provenance; that attribution has not been verified. Dates and relative time references belong to the historical source.
Extraordinary Awards
As AI Reshapes Brand Content, What Is Truly Expensive Is Only Just Coming into View
When Anyone Can Generate a Blockbuster at the Click of a Button, the Scarce Resource Is No Longer Knowing How to Make It, but Knowing What to Make
"
If generating an advertising video with AI takes only three days, why is your quote still so expensive?
Before discussing how AI is reshaping brand content, let us confront a real question that runs beneath every client–agency relationship but is rarely spoken about openly:
If generating an advertising video with AI takes only three days, why is your quote still so expensive?
At the “Content and Brands” roundtable during Extraordinary Awards' Hangzhou AI WEEK, four veteran advertising professionals and content creators offered sharp, even counterintuitive answers.
They did not deny that AI had driven production costs down to rock bottom. Instead, they revealed a harsher truth: when anyone can generate a blockbuster with AI at the click of a button, what is truly expensive is only just coming into view.
Do Not Stop at Prompts. Build a Context Warehouse.
While many companies are still studying how to write a few dozen characters of a “spell”—a Prompt—Focus Media has already moved on to the next race.
Sang Zhuohao, known as A-Sang, who leads key-account sales and AI projects at Focus Media, shared a powerful internal tool: the “Jiang Nanchun Agent.”
At first, they fed AI the advertising-slogan methodology of Focus founder Jiang Nanchun, along with nearly 500 client cases he had revised. Employees brought it a Brief, and it would critique it as sharply as Jiang Nanchun and produce a slogan with an internet-native feel.
But they soon discovered that prompt engineering was nowhere near enough for major campaign decisions involving tens of millions.
“If you tell the agent about a brand using only a few fragments, it often fails,” A-Sang said. So they moved beyond a simple Chatbot, using platforms such as cursor to build a “Context Warehouse of client information.”
What goes into that warehouse? Consumer research, recordings of interviews with business owners, executive strategy meetings, competitor-analysis reports… Dozens of client meetings all become context that can be retrieved as needed.
Feed it data from every angle to find the intersection of an opening in the market, the client's strengths, and consumer needs. That is the most valuable “computing power” AI offers marketing.
Nor is that all. AI is being deployed at Focus at remarkable speed. A-Sang said approximately 10% to 15% of the advertisements now shown on Focus elevator screens are already generated directly by AI. For content at Focus's level, an In-house team can produce a spot in just three or four days.
“Put the finished ad right in front of the client and they immediately see the result. The decision-making process becomes much shorter.”
How Can Agencies Present a Credible Quote in the AI Era?
Production time is being compressed dramatically. But that does not mean agencies should meekly accept bargain-basement prices.
M-verse founder Huang Chujie and Nengcheng Culture AI artist Wang Shenshuai are award-winning veterans who have lived through major advertising-industry cycles. Faced with clients pushing down prices, they offered textbook explanations.
“I cannot simply tell you how much one minute of AI production costs.” Huang Chujie's strategy is to change the weighting within the quote.
In the past, location shoots, lighting, and camera crews accounted for most of a TVC budget. Now he reduces pure AI production costs to a minimum, while substantially increasing the weight of upfront creative planning, writing and directing, and post-production editing.
“The value of the core people in the project does not change,” Huang explained. This not only keeps agencies out of destructive price wars, but also makes clear to clients exactly what they are paying for.
Wang Shenshuai went straight to the point. He had experienced traditional advertising's heyday, when the creative fee alone for a single video could run into the hundreds of thousands. With widespread AI use, production prices—for posters and other routine materials, for example—have fallen somewhat. But in the most important areas of strategy and creativity:
“We have not cut a single cent. AI cannot replace us in this part yet. Clients and agencies that care about quality agree on the value of human creativity.”
Lin Gengxu, founder of Dijuti, which handles many commercial CGI and AIGC projects, confirmed the same point. Brands' desire to reduce costs is a fact of life. “But you have to make it clear to clients that they are ultimately paying for your content and creativity. AI simply helps us realize grand, ambitious scenes that were previously very difficult to execute.”
AI Has Not Devalued Creativity. It Has Washed Away Pseudo-Creativity.
So what exactly is this “creativity” that these industry veterans defend so fiercely and refuse to discount by a cent?
Wang Shenshuai set a demanding definition: a previously undiscovered combination of existing elements. If a child spontaneously pours chocolate sauce on ice cream and it is the first time anyone has done it, that is creativity. If the child is copying McDonald's, it is common knowledge.
“Large language models learn from existing human achievements and recombine them. AI can give us 1,000 ideas in a second. But which are good ideas for this situation? Would an idea that works for McDonald's work for KFC? At the decisive finishing touch, AI still falls a little short.”
Far from devaluing real creativity, AI acts like a mirror exposing pretenders, stripping away the pseudo-creativity that relied on piling on labor and recycling formulas.
A-Sang added Focus's internal “1+N+X” model:
— 1 is the brand's soul, DNA, and universal values.
— N is the growth strategy for different groups.
— X is the vast range of expressions and creative materials.
AI can generate endless variations at the X level: “Produce 100 personalized short videos at the click of a button.” But “if you do not look for the ‘1’ at the brand's genetic core, all that creativity is a sham.”
That led to A-Sang's provocative take at this roundtable:
“When AI-generated content floods the world like seawater and becomes extremely personalized, centralized things—brand DNA and core memories—will become scarcer instead.”
The Future: From Agency to Trainer of the Client's AI
When generative models become so powerful that clients can produce videos themselves In-house with a few mouse movements, where do agencies go?
Huang Chujie predicted a complete restructuring of the client–agency relationship. Future agencies will become “consultants to the client's AI” and “managers of agent clusters.”
In the first year, the agency defines the annual Big Idea—the “1”—for the client.
In the second year, it trains 100 dedicated Agents to tirelessly develop 100 smaller Ideas for different audiences, 24 hours a day.
In this marathon of human–machine collaboration, the starting line has been leveled.
As Wang Shenshuai put it in his final, decisive remark:
“Once AI is available to everyone, it will be difficult to earn money merely from knowing how to use AI tools—just as no one charges more today simply because they can use Word or Photoshop.”
“When everyone is sitting in an identical wheelchair, the real competitive strength still lies in the brain sitting in that chair.” (Hawking, as attributed in the source.)
More Details from the Conversation
Extraordinary Awards · Hangzhou AI WEEK Trends Roundtable Panel: “Content and Brands: Personalized Expression and Brand Reshaping in the AI Era”
Guests: Wang Shenshuai, AI Artist at Shanghai Nengcheng Culture | Huang Chujie, Founder of M-verse | Lin Gengxu, Founder of Dijuti | Sang Zhuohao, AI Project Lead at Focus Media
Moderator: Jiang Zhiqiang (Alvin), Founder and CEO of Vitally AI
Jiang Zhiqiang: Hello, everyone. I am Jiang Zhiqiang. I have a small AI company of my own. We started with image applications, and over the past two years we have moved into serving enterprise needs. Because we do a lot of image generation and AI content, we have recently also entered the field of premium AI-produced dramas. I am moderating today, and our focus will be personalized expression and brand reshaping in the AI era.
Before introducing our guests, I wonder whether you have noticed how much faster AI has been developing since the start of this spring, compared with the past two years. There are two paradigm-level trends. One is the emergence of capabilities such as Sora, the multimodal video-generation model. On the surface, what people see is an extremely powerful ability to generate video multimodally, but most of us in the AI community see it as a paradigm-level breakthrough. The second, also one of the themes of today's conference, signals the arrival of an era of multi-agent (Agent) assistants powered by large language models. My own feeling is that progress over these past few months has been extraordinarily fast. Next, I will introduce our guests from right to left. Please introduce yourselves and offer a view related to our topic.
Let us start with Wang Shenshuai. He is usually very modest and would not tell you himself that he has won an award at the Cannes Lions International Festival of Creativity.
Wang Shenshuai: Hello, everyone. My name is Wang Shenshuai. I have worked in advertising for more than 20 years. Having lived through these 20 years of ups and downs in Chinese advertising, I feel that I have gone from being an experienced creative professional to an upgraded, highly hybrid advertising professional with an AI identity. The question I would like to discuss today is: can AI really help us do creative work?
Jiang Zhiqiang: Shenshuai, that was so brief. Next, Mr. Huang, please introduce yourself and bring us a point of view.
Huang Chujie: Hello, everyone. I am Chujie. My background is quite similar to Shenshuai's: we both came from advertising. In 2023, we founded a new brand called M-verse, focused on AI graphics and imagery. We began with experimental projects, moved into real commercial deployment at the start of ’24, and have continued through to the arrival of today's multimodal models. The question I bring today is: what will advertising look like in the future?
Jiang Zhiqiang: You have left us with a little suspense. A-Xu, please introduce yourself.
Lin Gengxu: Hello, everyone. I am Lin Gengxu, the founder of Dijuti. Dijuti is a digital-content creative advertising company founded in 2023. In 2023, we made many ads incorporating CGI that reached well beyond our usual audience. At the same time, we were exploring commercially viable AIGC advertising internally. Over the whole of last year, we helped more than 30 brands launch related AIGC co-created content, and we have more than 150,000 followers across online platforms. What I would like to discuss today is how we can use AI to collaborate better with brands in the future.
Jiang Zhiqiang: A-Sang, you can tell us a little more, because your experience may overlap with the other guests' in some ways and complement it in others, especially when it comes to clients and agencies.
Sang Zhuohao: Hello, everyone. My name is Sang Zhuohao; you can call me A-Sang. I currently handle key-account (KA) sales at Focus Media, so we work with many of the leading clients in the market. I also lead Focus's AI projects, and we are co-creating the company's overall AI system. When Mr. Jiang discussed this topic with me, I happened to have a provocative view that may run counter to its premise. In the AI era, everyone may say, “I can generate unlimited content and achieve extreme personalization.” But what I want to discuss is whether centralized media, centralized memories, and brand DNA still matter in this era. We know AI can empower us to do a great deal at the extreme-personalization end. Can it also empower us at the other end, where we find a brand's genes and its centralized memory?
Jiang Zhiqiang: A-Sang, please expand on that. From prompt engineering to context warehouses to agents, how are these being applied in practice, in light of the point you just made?
Sang Zhuohao: I can share a little about that. Focus is indeed a large company, and we work on projects for large clients. Often, we are not merely the final place where a client runs an ad; we also get involved further upstream in overall strategy development, and even in brand positioning. We were thinking about how AI could empower us.
The first stage was prompt (Prompt) engineering. We distilled many of Focus's advertising-slogan methodologies, including Mr. Jiang Nanchun's, into prompts. Internally, we now have what we call the “Jiang Nanchun Agent.” Whenever you bring it a Brief, it delivers a very sharp critique and tells you how it thinks the slogan should be written. But I have found that telling an agent about a brand in only a few fragments is often insufficient. A client's ad spend may involve a major campaign decision running into tens of millions or even hundreds of millions. We need to gather as much market information as possible.
So we gradually began moving from a Chatbot format toward IDE formats such as Coze, Claude, or Dify. We want to build a complete “context warehouse of client information.” It includes consumer research, interviews with the client's business owner, executive interviews, and competitor analysis. By analyzing information from multiple dimensions, we ultimately work out which point this Campaign should target to have the best chance of winning: the intersection of three circles—an opening in the market, the client's strengths, and consumer needs.
It is difficult to meet that need with just one Prompt in a context window. When we work on complex brand projects now, we retain the records of dozens of client meetings as context to retrieve and use as needed. I suggest that people working on KA accounts and complex consulting projects try this, starting with a knowledge base. Now that we have Agents, we are also experimenting with them, but I think we are still exploring. Context warehouses are relatively more mature.
Jiang Zhiqiang: I would like to ask a little more. Focus is integrating more and more services for clients, from research and planning to sometimes even having its own production team. In the age of AI-generated video, could you share more about what Focus itself is doing?
Sang Zhuohao: In fact, we started exploring whether Focus advertisements could be generated directly with AI in the second half of ’24. Today, approximately 10% to 15% of the advertisements you see on Focus elevator screens are already AI-generated. We have an independent In-house team producing videos for clients. For creative work at Focus's level, we can generally produce a spot in three or four days. This greatly improves clients' decision-making efficiency. In the past, clients would hesitate, unsure whether to run an ad. When we put the finished ad right in front of them, they see what the result looks like, and the decision-making process becomes much shorter.
Jiang Zhiqiang: Moving along to A-Xu: in our earlier conversation, you said your company had gone through several stages of development. Could you elaborate?
Lin Gengxu: In 2023, we produced many advertisements combining 3D and CG. Even then, we were making extensive use of AI to generate early-stage creative ideas and artwork during production. AI-generated storyboards let the design team and the client see the visual expression of an idea directly, improving communication and decision-making efficiency. We already had a small team exploring AI-based visual commercial advertising.
Later, as AI technology continued to develop, it could help us generate highly polished content involving people, cinematic storytelling, and large-scale shots. At the front end of the creative process, we can more boldly tell clients that we can use AI to realize ideas that used to be difficult to execute. There is more we can do commercially, and what we deliver is both more demanding and more imaginative.
But this brings me back to my point: brands and production teams are not really on the same plane when it comes to understanding AI's capabilities and limits. In collaboration, how do we align everyone's expectations—what AI can do, what it cannot do, which stages require traditional methods, and where we can use AI's strengths? This is a real test of a team's understanding of AI workflows and how it communicates with clients. Only by reaching agreement early can we avoid repeatedly scrapping the work and starting again during execution.
Jiang Zhiqiang: Could you give an example you are able to disclose? How do a brand's expectations of AI affect client–agency collaboration? Also, might clients think that because you are using AI, their expected costs should be lower?
Lin Gengxu: For example, when working with some brand clients, everyone may initially ask whether we can use AI to create content that feels like a film blockbuster. But the client's real need is still to communicate the brand and the product. Cinematic treatment needs to establish atmosphere and tone, yet the visuals must ultimately return to the product narrative. If expectations are not aligned upfront, you can spend a lot of time producing grand, spectacular visuals, only to discover that there is nowhere to put the product. Then you spend a great deal more time fixing and adjusting things. So we explain clearly to clients from the start which stages involve working with AI and why it is particularly suited to this creative idea.
As for expenditure, brands certainly hope that incorporating AI will reduce costs to some extent. That is an objective reality. But we must make something clear to clients: whether we use AI, CG, or 2D, it ultimately comes back to the content itself. People are paying for your content and creativity; technology is only the means of execution. That is why our team cares so much about creativity. AI simply helps us accomplish things that used to be more difficult.
Jiang Zhiqiang: Thank you, A-Xu. Mr. Huang, you and Shenshuai are both advertising veterans who have lived through several cycles. Do you have anything to add about client collaboration and costs?
Huang Chujie: That is a very good question, and the answer is reflected in the structure of the quote. I cannot simply tell you what one minute of AI production costs. I always break it down into creative development, writing, directing and planning, art and editing, and only then AI production.
AI production has the lowest weighting in the overall quote. I increase the weighting of upfront planning and post-production editing. That addresses clients' questions about what to expect from AI very well. Location shoots for a TVC used to be very expensive. Now I have minimized pure production costs, but the value of the core people in the project remains unchanged. This both keeps agencies out of a price war and provides a reasoned, defensible explanation.
Jiang Zhiqiang: As generative models become more capable, some clients are considering developing their own In-house capabilities because they think that would save costs. What is your view?
Huang Chujie: I have been thinking about this a lot recently. Now you can generate a 15-second piece of content using the simplest of methods, so clients can handle everyday work very efficiently themselves. But I believe the client–agency relationship and the services provided will be restructured. As large models improve, automation will replace routine steps, and the value people provide should move to a higher level.
Future agencies may become consultants to their clients' AI. For example, producing a TVC used to cost so much that a client might communicate only once a quarter. With AI, brands can communicate much more frequently, even delivering customized ads tailored to each individual. But when we return to the brand itself, its unchanging core is its spirit. Understanding that spirit requires deep familiarity on the part of both the brand and the agency. One future model may be for an agency that understands a client's brand philosophy well enough to help manage its AI groups—its agents.
Jiang Zhiqiang: Shenshuai, as a creative professional who has won a Cannes award, do you think creativity can be redefined in the AI era? Can AI really come up with good creative ideas?
Wang Shenshuai: I lived through a time when creativity was extremely valuable. Television-commercial production could cost tens of millions, and the creative fee alone could be four or five hundred thousand. Has AI made creativity less valuable or more valuable? That depends on what we mean by creativity.
Creativity means “creative ideas.” A child pouring chocolate onto ice cream is an idea, but if the same thing is already sold on the market, it is not creative. For advertising professionals, creativity is a previously undiscovered combination of existing elements. Large language models work by learning from existing human achievements and recombining them. In my view, their originality still falls a little short at present; that final, soulful touch is still missing.
AI can offer 1,000 ideas in a second. But experienced human creatives are needed to judge which are good and to add a distinctive finishing touch to the ones they select. An idea that works for McDonald's may not work for KFC. That is what creativity means. AI cannot yet create in the true sense, but it is tremendously helpful in brainstorming and providing starting points that spark better ideas.
As for quotes, what do we do if a client says prices should come down because we are using AI? For execution-heavy production work, such as making posters, we have reduced prices appropriately because we genuinely save time. But in the most essential areas of strategy and creativity, I have not cut a single cent, because AI cannot replace me there. At this stage, the value of the human brain remains irreplaceable. I hope both clients and agencies will hold that line and recognize the value of human creativity.
Huang Chujie: To add something about the Big Idea: the service model may change in the future. After developing the annual Big Idea in the first year, we could then help the client train an agent cluster, letting AI take on repetitive work and extend the concept into 100 smaller Ideas for 100 types of audiences.
Lin Gengxu: I think different people get completely different results from AI tools. It is like creative directors at advertising agencies in the past: much of their daily work involved listening to ideas and determining which were best for the brand. AI reduces the time spent retrieving information and offers a large number of ideas. But identifying the right ones, connecting them to a brand's pain points, and presenting them to the client is an irreplaceable human contribution. AI can make creative production more efficient, but it cannot fully replace creativity.
Sang Zhuohao: On the Big Idea, Focus has a model called “1+N+X.” The 1 is the brand's soul, positioning, and DNA—the universal values that can move consumers across cycles. N is the growth strategy: all creative work must serve growth. X is the specific creative expression, such as combinations across different online and offline platforms. Without that 1, all the creativity is a sham.
Also, regarding Tokens: now that we are entering the agent era, Token consumption is growing exponentially. Do not get stuck on whether a client should go In-house. What matters is the output—if you spend ten million Tokens, can you generate ten million in performance and ROI? If an outside company can handle this more efficiently than you can, that is the essence of the division of labor in business.
Jiang Zhiqiang: We are running short of time. To close, could each of you offer a bold prediction, provocative view, or piece of advice about AI-powered brand marketing, and give your own media account a plug?
Wang Shenshuai: As AI becomes widely available and the barriers fall, it will be difficult to earn money simply from knowing how to use AI tools. Just as no one charges more merely because they use Word or Photoshop. We should calmly accept the trend toward lower production prices. When everyone has AI tools and is sitting in a wheelchair, the real competitive strength is still the brain. You can follow our company's “Shanghai Nengcheng Culture” WeChat Official Account and Channels account, where we have work for some leading brands.
Huang Chujie: When brain–computer interfaces become viable in the future, we may not even need to type: an idea in our heads could simply be realized. Two types of work will not disappear. One is top-level architecture and design—you need to lead AI. The second is being interesting enough, because AI does not tell jokes well, and human work that provides emotional value is difficult to replace. Our company is called M-verse, meaning AI+WISE—intelligent AI. My personal account is called “AI Lipu She,” or “AI Absurdity Society.” Initially, I wanted to use AI to create visual content that runs counter to reality.
Sang Zhuohao: My provocative view is this: when AI-generated content becomes as abundant as seawater, centralized things will instead grow scarcer. However technology develops, people are still the ones who ultimately pay. So we still need to make content that touches people and changes how they think. Anyone interested in building top-level architectures is welcome to talk with us.