From Sourcing Engine to Agentic Commerce with Alibaba's Accio Agent

Much of the AI conversation is concentrated on frontier models and applications. But what happens in between—the work of making fragmented compute, model libraries, and enterprise systems function together—is just as important to the economics of adoption. In this episode, I speak with Maryann Tseng, Chief Strategy Officer of Phancy 06682.HK (formerly known as 4Paradigm), about the infrastructure middle layer that helps enterprises deploy AI more reliably and at lower cost. We unpack Phancy’s HAMi and RISE vGPU technologies, ModelHub, and its approach to improving GPU utilization and adapting models across heterogeneous hardware.Maryann explains how Phancy works with, rather than simply competes against, major cloud providers; why it sees a different growth trajectory for public-cloud Model as a Service (MaaS) than for its more established enterprise business; and how it thinks about SMEs, international expansion, and distribution partnerships. Underlying our discussion is a broader question: will the next stage of AI adoption be driven primarily by more compute and better models, or by making much better use of the hardware and technology already available?*apologies for the audioThe AI Proem Podcast is part of the AI Proem newsletter, which has ~13k followers globally. To learn more about China AI, the business of AI, and how AI is impacting society, please check out the ⁠newsletter here ⁠and more ⁠insightful conversations here.Chapters* 0:00 – Intro: Maryann Tseng’s role and what Phancy does* 4:32 – Who Phancy’s enterprise customers are* 6:09 – Why AI infrastructure matters, in plain terms* 8:36 – Enterprise bottlenecks: hardware limits, cost and reliable scaling* 10:59 – Phancy vs. traditional and new cloud providers* 13:35 – A crowded middle layer: competition and Hami vGPU’s market share* 15:59 – Flexibility, long enterprise relationships and trust* 19:16 – “Token factory” vs. Models-as-a-Service (MaaS)* 21:47 – Which industries are adopting AI fastest (entertainment, SMEs)* 24:01 – Competing with Alibaba, Huawei and Baidu as an independent platform* 26:16 – 12–18 month outlook: private vs. public cloud growth* 34:43 – Industry challenges: energy, chips and efficiency* 38:06 – A non-consensus view: efficiency over more spendingAI-generated transcript, for reference onlyGrace Shao (00:01)Maryann, thank you so much for joining us today. I’m really excited to be speaking with you. To start, tell us about your role and help us understand what Phancy does today. For many of our listeners, the company was previously known as Fourth Paradigm, and it has gone through a number of iterations over the years. Help us understand your main businesses, your core revenue drivers, and how the company has evolved.Maryann Tseng (00:30)I’m currently the Chief Strategy Officer at Phancy. I help our co-founders realign our longer-term strategy, identify long-term business partnerships, and translate our technology into more sustainable growth. Expanding into international markets and building awareness of Phancy internationally is also one of our key priorities this year.Phancy was founded in 2014. Today, we provide AI infrastructure middleware, connecting heterogeneous compute environments and resources with different models to support a variety of enterprise solutions and workflows. We have two important products: HAMi vGPU and RISE vGPU. They improve GPU usability and utilization.Another key product is ModelHub, a community with a library of more than 300,000 models and capabilities for model adaptation and optimization across heterogeneous hardware environments. We commercialize these technologies through our AI Platform business, supporting enterprise deployments in public clouds, private deployments, and MaaS, or Model as a Service, which delivers inference services through APIs.Grace Shao (02:15)Maryann, as we mentioned earlier, the company has gone through several iterations. You now have a few different businesses. What has really changed from the pre-IPO period to being listed on the Hong Kong stock exchange?Maryann Tseng (02:32)What we’ve always done, from day one, is serve enterprise customers and understand the needs of companies across different industries. When we founded the company, we realized that enterprises wanted to use more AI products internally, but finding suitable AI models and integrating them into their existing workflows and systems was a challenge. That is why we built our AI Platform: to help enterprises deploy AI within their existing workflows and improve internal efficiency.Today, there are millions of models, but compute and hardware remain severely constrained. How can we help enterprises make use of heterogeneous chips so that these models run efficiently and effectively? With agentic AI, we’re seeing more workflows broken down into smaller, more fragmented tasks.Our vGPU technology virtualizes GPU resources to make compute more usable and improve utilization. Together with ModelHub and its library of more than 300,000 models, we can help enterprises deploy models effectively across different heterogeneous chip environments and integrate them into their systems with higher efficiency.Grace Shao (04:32)Who are your main customers, then?Maryann Tseng (04:37)We focus on enterprise customers, including major telecom operators, financial institutions, electricity providers, and power producers. These are some of the key industries we’ve served over the past decade.Grace Shao (05:58)Maryann, let’s talk about what’s top of mind in the industry right now. Much of the AI conversation focuses on models and applications. But what you do could be considered the less glamorous, yet much-needed, work in between. Help us understand why solid AI infrastructure matters and what it actually includes. Break down the technicalities for us in layman’s terms. What exactly are you doing, why is it so important, and why are so many customers paying for this service?Maryann Tseng (06:45)If we take a step back, AI infrastructure determines whether models can be delivered as reliable, affordable services at scale. It includes GPU virtualization, heterogeneous resource management, scheduling, model adaptation, and inference optimization. Both usability and utilization matter. How can we use the available hardware to run the required workloads, and how can we make sure that hardware is being used productively?With agents, we’re seeing workflows divided into much smaller tasks. Allocating an entire GPU to a task that only needs a fraction of its capacity creates a lot of waste. Our vGPU technology enables fine-grained slicing and concurrent execution, while ModelHub helps select and adapt models to run effectively on the underlying chips.This middle layer is what we do. It connects models and hardware and, more importantly, helps enterprises deploy AI more efficiently. Sometimes we see utilization improve from around 30% to 65–85%. That helps companies bring down deployment costs. Even though it may sound less glamorous because it’s a middle layer, we believe it’s a critical component for mass AI deployment.Grace Shao (08:36)That makes a lot of sense. What I’m hearing from enterprises is that adoption speed is one challenge. There’s a lot of bureaucracy in older companies. The second is the high cost of adopting frontier models. Beyond cost, what are the biggest challenges you’re seeing among your clients in turning AI into something they can reliably scale? What other bottlenecks do they face?Maryann Tseng (09:06)Hardware constraints are a major challenge. How do we maximize the existing supply chain and make the best use of the resources we already have? This is something we’re actively helping enterprises solve.Going back to ModelHub and vGPU, because hardware is constrained, we need to intelligently combine different types of chips in heterogeneous environments and break workflows into much smaller tasks, sometimes into a hundred pieces. At the same time, we need to combine th… Get full access to AI Proem at aiproem.substack.com/subscribe

This episode was a special collaboration project with Tencent’s Behind the Penguin podcast. AI Proem remains editorially independent but is open to collaborating with companies and individuals if that means they can bring better resources/ guests/ insights to the AI Proem audience. I speak to Serrino Liu, the AI Product Expert at Miora, Tencent. He is responsible for product innovation across film, 3D, and the interactive formats built on top of them, bringing AI agents into real content-creation and workflows.He tells me that Miora is not another image or video generator. Rather, it is a creative agent studio that runs entire workflows for filmmaking, e-commerce, game and 3D professionals.What makes the conversation unusual is Miora’s stance on models: it deliberately routes across mainstream models from different providers: DeepSeek for documentation, Seedance for video, Hunyuan Image for graphics, and models from direct competitors like ByteDance and Kling. Serrino is refreshingly candid about why: Miora is not a foundation-model company, and every improvement in the underlying models flows back into the product. He also unpacks Miora’s relationship with Tencent’s office agent WorkBuddy (”two pieces on the same board” where one manages the workspace, the other the creative work), and how the newly launched Hy Image3.5 preview is “professional-grade” in the sense that matters: usable, consistent, and able to survive revision, not just prettier.The heart of the episode is a nuanced take on taste and judgment. Serrino splits creative work into what can be verified, such as frame format, dialogue, and character consistency — and what only a human can judge: whether the emotion lands, whether the camera movement serves the story. Features like the new StoryWeave and side-by-side version comparison are built to keep the human in the loop, “a partner in making, not replacing.” The same philosophy extends to memory: Miora’s visible, editable memory system exists to kill tedious repetition — remembering your color palette, your story decisions, your style — so the next revision respects everything you’ve already decided, instead of forcing you to re-explain yourself.The episode closes with two big-picture exchanges. On geographies of preference in AI adoption, Serrino argues the difference between Hollywood and Asia is less about values than stakes: Hollywood’s AI debate arrived as a contract question (consent, credit, compensation), while Asia’s arrived as a capacity question — enormous demand meeting thin mid-tier production, with both anxieties eventually converging. And in his contrarian take on a fully automatic creative AGI, he lands the episode’s most quotable line: in creative work, the most expensive thing was never execution — it’s knowing who to pass the ball to. Craft will always change, but the author and the user never do.The AI Proem Podcast is part of the AI Proem newsletter, which has ~13k followers globally. To learn more about China AI, the business of AI, and how AI is impacting society, please check out the newsletter here and more insightful conversations here.Chapters00:00 Welcome from Macau: Meet Serrino Liu03:35 What Is Miora? Agents, Memory & the Professional Harness04:53 One Product, Many Models: Routing, Rivals & Token Economics11:14 Building for Real Creators: The StoryWeave Approach18:12 Taste, Judgment & Keeping the Human in the Loop28:09 Memory as a Creative Asset & the Hy Image3.5 preview Launch37:20 Asia vs Hollywood: Two Different AI Stories44:23 Misconceptions, Copyright & a Contrarian View on AGIAI generated transcript, for reference onlyGrace Shao: hello everyone welcome back to AI Proem Podcast this is your host Grace Shao. Today’s episode is very special because we’re partning up with Behind the Penguin a newly launched Tencent backed podcast. We’re in Macau today because we are at the 2026 International Film Camp and joining us today is Serrino Liu, Tencent’s Miora AI product expert. He had just revealed and demonstrated Miora’s capabilities in film making to industry really excited to hear from him today about how the creative industry is reacting to AI, adopting AI and how he sees the future ai I’m really excited to hear from him not only just about the product itself but how the industry is embracing AI or maybe not.Serrino Liu: Okay, thank you so much for introducing me and it’s great opportunity to meet with all kinds of the different directors today and I think I actually have already got some questions I wanted. I think this conversation about making film with AI is actually welcoming a different a gate for the young director to try AI with some film production and I should say i’m only speaking for the room and not industry. But in the room nobody was impressed with the film itself, though like the output anymore. It’s actually how they can steer it.Grace Shao: It’s very cool. But before we get more into Miora and the Tencent’s ecosystem and just how AI is being adopted. I just found your personal background very interesting actually before we start recording we just had a brief chat and you’re actually a French/Russian/ Chinese descent, but now you just returned from the US to China to work for Tencent. You have a very interesting professional journey as well. You’re an artist by training, having studied design at Parsons, studied CS at NYU, studied at Harvard. Tell us a bit about yourself and what brought you to Miora and the vision that you have for Miora.Serrino Liu: okay thank you. Though “artist” is a hat I‘m not sure I’ve earned — I‘m really a generalist. From computer science, design and business and machine learning and AI... I have touched all of it. But it sounds pretty scattered but I think there is a clear line behind it. It’s like I know the tech I know the product and the whole time i’ve been obsessed with is just one thing: How much idea in someone’s head can get real and that is actually the reason I joined Miora. I want to see how creative agent can provide some opportunities for customers for users to get some inspiration from their head to the real life and to see their inspiration come true.Grace Shao: If you had to use one or two sentences to describe Miora or what it is today. How would you describe it?Serrino Liu: I think it’s just agents and memory and the professional harness to serve as the assistant directing different area.Grace Shao: so it’s not really a generic AI image or video generator in any sense.Serrino Liu: No, it’s not. It’s more like a creative agent studio or like a system for you in personal so it can help you to do some professional work and jobs etc.Grace Shao: Very cool. So what industries do you think Miora can be used in or is being used in right now?Serrino Liu: I think we have some targeting area like the e-commerce, filmmaking, game and even 3D... Those different areas have some different professional harnesses with the agent to do the specific workflow to serve this kind of area.Grace Shao: but I imagine for each of those use cases just mentioned or each of those industries just mentioned, their need for how to even use a creative tools actually very different right, so in that sense how are you guys optimizing which models you’re using because I understand that actually miora routes through quite a few different kinds of models from different providers and it’s not like solely committed to just offering Hunyuan.Serrino Liu: Yeah, so Miora isn’t just tied to a single model we’re actually routing different mainstream models like the Seedance and the different agent model as well. So user can pick what fits their needs best and in particular Miora is actually one of the way (for) the hunyuan model to reach the overseas uses. We have the newly launched Hy Image3.5 preview releasing today and that is actually the agents own choices to inside the Miora and users can actually choose it.Grace Shao: but so say in my workflow, I need to script some data I need to … Get full access to AI Proem at aiproem.substack.com/subscribe

In this episode, I speak with Kristy Loke, an AI governance researcher focused on US-China relations, about what the latest Trump-Xi meeting could mean for AI governance, safety, and cooperation between the two leading AI powers. We discuss the possibility of an incident reporting mechanism for malicious AI-enabled cyber activity, the growing risks around frontier models, and whether the US and China may have more shared interests than the current competition narrative suggests.One part of our conversation I found particularly interesting was Kristy’s framework for understanding China’s AI strategy. Rather than simply racing toward the frontier, she describes a four-pronged approach focused on staying close to the technological frontier, translating AI into productivity gains, securing core technology supply chains, and protecting people’s welfare. This helps explain some of China’s decisions around AI diffusion, chips, and regulation of emerging technologies such as AI companions.Kristy argues that the American approach is more closely tied to technological leadership and power, while China’s approach is more pragmatic and focused on longer-term economic and social goals. That distinction also raises questions about how AI safety is framed, particularly when scientific concerns about frontier models become intertwined with geopolitical and commercial interests.We ended by looking at whether meaningful US-China coordination is actually possible. Kristy sees a potential path from continued dialogue to sharing safety practices and, eventually, common standards, even as competition remains intense. We also talked about what middle powers such as Singapore, the UK, and countries in Europe can do as AI governance increasingly becomes a conversation between the world’s two leading AI powers.Relevant articles by Kristy:* https://www.theinformation.com/articles/americas-ai-dream-failing-launch?rc=sqpcn6* https://www.wired.com/story/china-isnt-buying-silicon-valley-call-for-ai-slowdown/ The AI Proem Podcast is part of the AI Proem newsletter, which has ~13k followers globally. To learn more about China AI, the business of AI, and how AI is impacting society, please check out the newsletter here and more insightful conversations here.Chapters00:00 AI Governance and International Relations02:48 The Stakes of AI Safety05:42 China’s Pragmatic Approach to AI08:05 The Dynamics of US-China AI Competition10:56 The Future of AI Cooperation16:05 The Four Prongs of China’s AI Strategy24:36 Contrasting AI Visions: China vs. America27:43 Personal Journey into AI Governance36:09 The Intersection of Governance and Development38:59 Non-Consensus Views on AI Leadership40:24 Middle Powers in AI GovernanceAI- Generated Transcript (for reference only)Grace Shao (00:00)Hi Kristy, thank you so much for joining us today.Kristy Loke (00:02)Hi Grace, great to be here.Grace Shao (00:03)Yeah, so you’re studying in Toronto, but you are watching the Trump-Xi meeting very closely. I I believe you were just on like all kinds of media throughout the day commenting on what happened. Help us understand the high level takeaways. Xi and Trump just met in DC, I believe it was yesterday, and AI governance, AI safety was very high on the agenda.Kristy Loke (00:24)Yeah, I think a good way to think about it is good vibes actually matter in this kind of summit where the actors play a major role. Xi and Trump clearly have a rapport. And I thought it was really good to see a continuation, if not a formalization, of the AI dialogue that started in May. so it’s good to see that. And then there were some pre-summit discussions by Bessent and He Lifeng, the Chinese and US representatives. about potentially setting up an incident reporting mechanism which would meet some of the growing needs for them to either de-escalate or warn each other of like let’s say malicious actor activity around models.Grace Shao (01:03)So what is the de escalation plan looking like? Do you know anything about that?Kristy Loke (01:08)Yeah, I think we’re very thin on details currently, but it’s promising that they’re talking about it because they might urgently need it in some near point in the future. Although they are meeting each other soon, which is also great. Yeah, another thing of note is that it’s very much like a continuation of the Busan truce that started last year, like last October, where because China tried the Trump card of Rare Earth and to counter the US’ escalate in terms of trade, but also in terms of chip supply. and so we’re basically at this point where there’s almost like a ceiling, although tenuous between them when it comes to tension over chips and tension over other technological points of tension.Grace Shao (01:50)Yeah, you can say that like since twenty eighteen, there’s been a bit of a back and forth, tit for tat for a while. And at this point, it almost feels like there’s a window for AI cooperation, tension, de-escalation, because you know, both sides have realized, you know, if there’s no international standard in terms of governance, AI models themselves could actually be of risk for society of their own nation, like each nation, right? How do you see that? Like what’s the kind of c realistic collaboration we can see or something that could meaningfully, you know, be beneficial for two nations or even the world?Kristy Loke (02:25)Mm. I yeah, absolutely. I think the stakes got really real for them around April this year when Mythos was released or is somewhat withheld from release by First Anthropic on safety grounds and then also by the Trump administration on security grounds. And so that kind of woke up a lot of countries around the world and from China’s perspective, it’s also a little bit scary because are we treating AI like a weapon? Is it going to be used against China? So it prompted a lot of discussions. But part of the surprise is: whoa, this is really real. The cyber capabilities are getting kind of really good. and another point, the point of more joint stake and joint like common ground between them is that they both do not want malicious cyber actors to use it against their critical infrastructure. So the setting up of the incident like protocol or communication can also address this problem so they can deal with it quicker together. But of course, part of cyber is heroism, part of cyber is very much a common ground place. And so there are two facets to that as well. But yeah, so essentially it got really real for them in April. And then the subsequent months have also been crazy as those of us who follow AI closely, there were the hugging phase, OpenAI, cyber incidents that then, you know, we then find out that it also happened to Anthropic. and so that episode is not over, right? People are still like Trend Micro is an organization that recently found that it was kind of much deeper, you know, it’s kind of spreading in some ways. And so it’s an evolving situation which prompted the two leading AI powers to kind of get together and figure out what they can do. I don’t think the climate is right for pacing. And as Xi’s speech kind of pointed out, or kind of like pointing us towards I I really I I thought his framing was really interesting. So the way that he talked about AI was in the vein of the US and China are the leading AI powers. And therefore we have the ability to and we have the responsibility to to make sure AI goes well and is managed well. So it’s almost like a little bit of a subtle response to people who say, we can’t do anything, you know, we can only stop it, or you know, go full steam ahead and China’s more like. we kinda gotta control it, but for now we’ll keep it going. But we’re worried. Yeah.Grace Shao (04:40)How do you actually view that?Kristy Loke (04:42)Mm.Grace Shao (04:43)How do you view the whole narrative around that we must pace our efforts in pushing towards frontier suddenly, right? Like from the US side and also from the China side.Kristy Loke (04:52)Yeah, so it’… Get full access to AI Proem at aiproem.substack.com/subscribe

In this episode, I speak with Aoni Wang, product designer at Harvey, and Yiyang Hibner (Silicon & Spice with Yiyang), product manager at Airbnb, about how AI is changing the way they work day-to-day. This conversation is a bit more light-hearted than the usual conversations on AI Proem, but I think it is equally insightful because these two professionals work in the heart of Silicon Valley, and they give us a glimpse into how people are really adopting AI.Both are already using AI as part of their regular workflows, from thinking through ideas and understanding complex systems to research, writing and building interactive prototypes. We explore the tension between making execution faster and deciding what is actually worth building. As AI makes it easier to produce prototypes and generate work, Aoni and Yiyang discuss the risk of confusing speed of output with speed of progress. More work can be produced, but that also creates a greater need for filtering, judgment, and accountability. They also discuss why AI can be too agreeable, and why users still need to challenge its outputs, verify information and bring their own expertise to the process.The conversation then looks at how AI could change the boundaries between product, design and engineering. Then they discuss the growing importance of skills such as systems thinking, prioritization, communication, taste and the ability to define what a good product or experience should look like. For Aoni, this includes thinking about visual language and coherent systems rather than simply producing individual screens or prototypes.Finally, we get into the more practical questions around working with AI, including how teams maintain shared context when everyone has their own AI tools, how people decide which products are actually useful, and where human judgment remains difficult to replace. The conversation offers a view of AI adoption from inside the workflow rather than from the perspective of the technology itself, raising a broader question about the future of work: as more of the execution becomes easier, how will people decide where to focus their time, attention and expertise?The AI Proem Podcast is part of the AI Proem newsletter, which has ~13k followers globally. To learn more about China AI, the business of AI, and how AI is impacting society, please check out the newsletter here and more insightful conversations here.Chapters* 00:00 The Evolution of Design Work with AI* 05:14 AI’s Impact on Product Management* 07:59 Efficiency vs. Overwhelm in AI Utilization* 11:26 Navigating AI Slop and Quality Control* 14:19 The Future of Roles in Design and Product Management* 17:59 The Negative Impacts of AI on Work* 21:22 The Changing Landscape of Collaboration with AI* 28:03 Shifting Workflows: AI vs. Traditional Software* 30:17 Creative Tools: Evaluating Effectiveness and Usability* 33:38 The Importance of Human Skills in an AI-Driven World* 36:49 AI’s Role in Decision Making and Workflow Management* 40:57 Personal Use Cases: AI in Everyday Life* 46:28 Differentiated Views: Perspectives on Work CultureTranscript (AI-generated, for reference only)Grace Shao (00:00)Hi guys, thank you so much for joining today. Really really excited to have you on.Yiyang Hibner (00:03)Thanks for having us.Aoni Wang (00:04)Thanks for having me, Grace.Grace Shao (00:04)Yeah, so before getting started, I know you guys both have some disclaimers from the company you have to say. So go ahead, do your thing.Yiyang Hibner (00:12)All right, I’ll go first. I’m Yiyang, I’m based in Bay Area and currently I work at Airbnb. And opinions are my own. So you know, Airbnb is one of the companies I’ve been working at in the past few years. There are a lot of other tech companies, so like some experiences are not unique to the current company I’m at. Yeah.Aoni Wang (00:30)Hi, I’m Aoni. Similar to Yiyang, I’m based in the SF Bay. I now work at Harvey as a product designer. But today all views are my own and also coming from my own experiences.Grace Shao (00:41)Yeah, perfect. So once we get that out of the way, I would love to hear about the your own experiences a bit more. Could you guys just give us a little bit of color on what is it that you do? You know, what brought you here today? You know, compare the work that you are doing today to maybe even two years ago prior to the proliferation of like AI tools and, you know, AI agents.Yiyang Hibner (01:03)I’ll let Aoni go first.Aoni Wang (01:05)Yeah, yeah, sure, sure. Yeah, I could go first, yes. So yeah, I’m a product designer, and usually to people that means I’m designing software interfaces. So whether that is desktop software or mobile apps, and I would be designing the interface that people see, the interaction, collaborating very closely with product managers and engineers most of the time. For me, I have about 10 years of experience in this field working across different tech companies. Mostly working on mobile consumer type of products. And to us designers or product designers, Figma is our best friend. So if I were to think about two years ago, I’d be spending a lot of time in Figma. A lot of my design work actually happened in Figma. For example, the ideation, sketching out ideas before it actually becomes finalized or pixel perfect, sketching up mock-ups, interactions, flows and So from like rough sketches all the way to that polished mock-up that I can hand off to my team to build, all of that would happen in Figma. And now I would say where I spend my time and how I design have changed the most. So I guess it’s probably true for everyone. I’m chatting with Claude or ChatGPT a lot every day. And even before getting to Figma, I would do a lot of design thinking, sketching with AI agents. And Figma is only a small part of my design journey now. Or depending on the work, like it’s not all I use to design anymore, especially with coding agents. Now it is much easier and much faster to prototype, meaning that I can build something that looks functional, feasible, and realistic, and that my team can play with it, I can use it to communicate my ideas. In the past, that is usually a much longer process.Grace Shao (03:01)Yeah, I definitely want to double click on the Figma moat later because I think that’s something we talked about offline as well. It’s quite fascinating how back then Figma’s moat, like, its value is also the shareability of it. But now, you know, I’m surprised that you’re moving away from Figma in some of your workflow. But let’s talk to that a bit later. Yiyang, tell us about your work. Like what is it that you do? How are you using AI already in your day to day work and how has that really impacted your workflow compared to maybe two years ago?Yiyang Hibner (03:30)Yeah. So I because I’m a product manager, so I interact with a lot of different cross-functional partners. Good engineers, designers, sometimes legal partners, or like product operations. So I think the biggest change is to have I think AI does make me more organized. Because as a PM you live and breathe in documents, in presentations, pitch decks, one-pager, strategy vision docs. So a lot times I you have to wonder, okay, what’s the best way to convey my I ideas in a very efficient way, tailored to my audiences. I think whether Aoni mentioned Claude Code or other tools, like Gemini, you know, company has different tools. I think these help me to find my voice, especially during kind of presentation and influence across, that really helps. Another big change is that Unlike designers, like you know, people really use Figma a lot. I think product managers and people have different like specialties. Sometimes some people prefer one way of working, other people prefer different ways of working. And for me, having AI to kind of be a thought partner, whether it’s through user research or competitor analysis or industry kind of news, is sometimes you know, you can only read so many newsletters and what’s the best essence of all or what’s happening in the payments world… Get full access to AI Proem at aiproem.substack.com/subscribe

In this episode, I speak with Ang Li, co-founder and CEO of Simular AI, about the rise of computer use agents and his vision for computers that can increasingly do work on behalf of humans. We discuss why computer use matters beyond the current wave of API-based agents, particularly for the vast amount of enterprise work still carried out through legacy desktop software that was never designed to be accessed programmatically.We explore where these agents could have the most immediate impact, from processing invoices and extracting information from unstructured documents to navigating financial, healthcare and other enterprise systems. Ang argues that the goal is not necessarily to remove humans from the workflow, but to shift the balance between execution and judgment, with agents handling repetitive tasks while people remain responsible for decisions and sign-off.The conversation also gets into the economics and technical challenges of computer use agents. Ang explains why relying on a frontier model for every click can be expensive, slow and difficult to control, and how Simular’s neurosymbolic approach turns repeated workflows into executable playbooks. We discuss the role of smaller and open-weight models, the changing economics of AI agents, and why Ang sees computer use as a way to make automation more accessible beyond the highest-value technical work.Finally, we look at what this shift could mean for SaaS and the future of work. Ang distinguishes between systems of record and software that primarily serves as a portal or interface, and argues that these categories may face very different levels of disruption from agents. We end with a broader question about human agency: if AI increasingly handles execution, will the more valuable skill become knowing how to identify the right problems to solve in the first place?The AI Proem Podcast is part of the AI Proem newsletter, which has ~13k followers globally. To learn more about China AI, the business of AI, and how AI is impacting society, please check out the newsletter here and more insightful conversations here.Chapters00:00 Introduction to Simular and Its Vision05:14 The Future of Work and Human-AI Collaboration10:21 Technological Landscape and Market Demand15:31 Real-World Applications and Use Cases20:40 Challenges and Competitive Landscape29:55 The Dual Nature of Work: Content Creation vs. Execution33:50 The Competitive Landscape: Frontier Labs vs. Open Weight Models37:17 The Rise of Computer Use Agents43:30 Empowering the Average Person: High Agency Through Technology48:46 Understanding SaaS: Infrastructure vs. Portals52:32 The Future of AI: High-End Jobs vs. Repetitive TasksAI-generated Transcript (for reference only)Grace Shao (00:00)Ang, thank you so much for joining us today. Really excited to have you. To start with, could you tell us a bit more about yourself and why you built Simular? What is your long-term vision for the company? And what really brought you here along your academic journey?Ang Li (00:13)Yeah, thank you so much, Grace, and thank you for having me here. My name is Ang Li. I’m the co-founder and CEO of Simular. We’ve been working on this company for almost three years.In short, we call ourselves the autonomous computer company, meaning we are building autonomous computers. The goal of Simular is basically this: everyone has computers right now, but we have to work on them manually by moving the mouse, typing on the keyboard, looking at a screen, and understanding what’s going on ourselves.We envision a future where computers will do the work on behalf of humans, on their own. That’s really the technology that we’re building towards. Nowadays people also call them computer-use agents. It’s basically a general-purpose agent that can use the computer just like a human.A bit about my background: I’ve spent almost 20 years researching AI. My personal research direction has basically been trying to figure out what people sometimes call AGI, artificial general intelligence. The goal is to have a general-purpose system that can learn like humans and perform actions just like humans.I see this autonomous computer problem as the first possible realization of AGI technology that could have a huge impact on society.Grace Shao (01:29)It’s quite interesting. You mentioned computer-use agents. It seems like there’s been quite an influx of capital going into this space right now in Silicon Valley. Has there been a genuine technological shift here? Why is everyone suddenly so hyped up about CUAs?Ang Li (01:44)Yeah, so that’s the interesting part. We started three years ago, and when I told people we were building agents, people didn’t understand it. I was telling people, “Okay, we’re building agents that use computers.” And people would ask, “Why? Why are you building agents that use computers? Why not just use APIs?”Agents aren’t actually a new concept. It’s a word that has been around for tens of years in the research community. We already talked about agents within DeepMind when we were doing research towards AGI. It just wasn’t very familiar to the broader public.Three years ago, we had the first version of ChatGPT, and we knew the scaling laws for foundation models were working. Foundation models were becoming very powerful.We looked at the trajectory of the technological shift and realized that, for a computer to work like a human, you need APIs. If you want a computer to go into your Gmail, look at an email, and send an email on your behalf to someone else, there’s already a Gmail API. So if you want to do those kinds of tasks, you just let the agent call the Gmail API.Three years ago, that was basically the case for tool calling: basic APIs.Then we asked: suppose we have all the APIs readily available to agents, what’s remaining?The answer became very natural. What’s remaining is all the software that has no APIs.For example, lots of companies have legacy software on Windows computers, like old ERP systems that nobody is maintaining anymore. That software has been running for many years, and people don’t really want to change because they’re so familiar with it.The problem is that because the software is outdated, people still have to manually work on it. There are no APIs.If our goal is to liberate human labor, we don’t want people spending so much time sitting in front of computers doing repetitive, tedious tasks. Nobody likes that. Everyone thinks, “Why can’t I do something more interesting with my life? Why should I spend eight hours every day doing repetitive stuff that doesn’t require much cognitive load, just looking at a spreadsheet and filling out the same form over and over again?”Those problems cannot be solved by agents that only have APIs.So then we realized this is actually a harder problem. It requires technology that can look at a screen, decide, “Should I click this button here? Should I type something?”, move the mouse to the coordinates of the button, click on it, and move to the next page.It feels a bit like a self-driving car in the digital world. A self-driving car looks at the streets and decides, “Should I turn left or right? Should I press the gas?” In this case, we’re looking at a screen and deciding where to click.When you combine the two, API agents and computer-use agents, you cover the full spectrum of computers. There’s nothing else remaining.Once you have API agents and computer-use agents and combine the two, computers can become autonomous. That’s the AGI that everyone is striving for.Grace Shao (05:25)It’s pretty crazy. Your vision of the future of work is essentially that computers run themselves.I get that vision. But wouldn’t work itself, by nature, just change? So much of our work right now, like you said, filling out PowerPoints or spreadsheets, you can almost call it performative because it’s ultimately for humans to view.But if it’s agents or computers viewing the output, do we still need to fill out those PowerPoints and forms?Ang L… Get full access to AI Proem at aiproem.substack.com/subscribe

Hello everyone, I've been on the road, so I'll be delaying this week’s written post and next week’s podcast episode. Apologies in advance.In this episode, I’m joined by Asia Bureau Chief Jing Yang and Senior Reporter Juro Osawa from The Information to unpack DeepSeek’s surprising shift toward external capital, its unique investor structure, and what its rapid revenue growth reveals about the economics of AI. We also explore why China’s AI labs remain fiercely competitive despite a far smaller capital market than the US.The conversation moves from DeepSeek and Huawei to accusations of distillation, ByteDance, Alibaba, and Tencent, and the growing race to turn AI models into real businesses. We also look at China’s emerging advantage in robotics, from its dense hardware supply chain to the vast amounts of real-world data being generated through deployment.Finally, we discuss the next frontier: world models. As the race moves beyond language models, can China’s hardware and data advantages translate into an edge in embodied AI? And in a provocative final take, Jing argues that Chinese frontier models may never fully overtake their US counterparts.The AI Proem Podcast is part of the AI Proem newsletter, which has ~13k followers globally. To learn more about China AI, the business of AI, and how AI is impacting businesses, please check out the newsletter here and more insightful conversations here.Chapters00:52 The Rise of Chinese AI Models04:06 DeepSeek’s Unique Position in the Market12:31 Capital Structure and Investor Dynamics16:12 The Landscape of Chinese AI Labs19:17 DeepSeek and Huawei: A Strategic Partnership25:43 The Competitive Landscape of AI Labs33:51 Challenges in the Chinese Capital Market36:44 ByteDance’s Unique Approach to Model Training41:32 The Future of BAT Companies in AI48:22 China’s Robotics Landscape and Advantages53:51 Impact of US Regulations on Robotics54:43 Unitree’s IPO and Market DynamicsGrace Shao (00:00)Hey Jing, hey Juro thank you so much for joining AI Proem today.Juro (00:03)HeyJing Yang (00:03)Yes.Grace Shao (00:04)to start, you know, let’s start with Ox alpha. That was such a teaser. It turns out to be ZAI again, but I think it was very much expected by people who do watch space closely. It was no surprise in that sense. But I think what’s shocked a lot of people is that, you know, the inference was running on Chinese domestic ships and potentially GLM five point three flash.The price, it’s priced at about like one fortieth of Opus 4.8. That’s quite significant of a gap, right? Tell us about what you think about that, the implication on these continued Chinese open weight models that are coming out strong but cheap, and how that’s impacting US frontier models.Juro (00:43)Yeah, maybe I can start on this one, but I think those flash models are coming out, like Deep Seek had the V4 flash, which is a smaller size one, and that became very popular. And then this one comes out. Alibaba just had a three point eight flash as well. So those are smaller size, you know, lower cost models that can still handle a lot of like AI Asian type of tasks really well. Soknow, this is like a sort of sweet spot for demand, right? So a lot of Chinese companies with open source models are coming out. So that’s one, you know, one thing about this, you know, that’s part of that broader trend. But I think as you pointed out, the chip part, you know, it kind of shows how far the Chinese chips have come in terms of inference, right? AndThey have become more capable, especially Huawei chips, being used, you know, more and more for inference. And you know, Deep Seek came out with V4 and you know it was also optimized for you know to run on Huawei chips as well. And we’re gonna see more and more of this for sure. But I mean to what to keep in mind that when it comes to training, it’s not the same story yet, not quite yet. So, you know, a lot of Chinese companies are using, still using the most advancedNVIDIA chips and trying to gain access to that and which we also wrote about recently.Jing Yang (02:03)Yeah, I mean the only thing I I’ll add to that is I think the bigger picture is that there is a shortage of inference trips in both the US and China, but in China the shortage is much more runs much deeper and you knowA part of that is because of government policies, right? We still have not seen H two hundred actually officially being allowed to export into China. So the right course for all the major leading Chinese AI labs is they have to adapt the inference to domestic trips and to you know this concept of homogeneous compute, which has been discussed in Silicon Valley, but as well as in China.And then the hemogeneous part in the Chinese context is very different. It’s about being able to run your inference on a combination of NVIDIA and and and say five different other Chinese GPUs. Right. So we saw that Kimi K3 had to suspend our new customer subscription shortly after it was released, and that was directly a result of not having enough inference chips.to to to you know to to meet the surging demand.Grace Shao (03:14)I think Kimi wasn’t the only one that faced this kind of issue. And Deep Seek and ZAI throughout last year, I think, had to like, you know, manage their demand as well. But I wanna bring the conversation to Deep Seek. You know, you guys broke the story on well, actually Juro just broke the story on Deep Seek’s revenue. And I believe Jing Yu wrote the story about Deep Seek’s fundraising. It’s quite fascinating because DeepSeek, quite secretive, you know, their investor letter or investor conversation was leaked.Apparently the founder was not happy that it was leaked, But, you know, they are fundraising now. And that’s shocked a lot of people because for the longest time they didn’t want to take any external capital. Jing, do you want to take that first and just talk about, you know, their capital structure, their fundraising, how unique it was, and then maybe we’ll pass it to Juro to talk about, you know, his recent findings about the money they’re making.Jing Yang (04:03)Sure. Yeah, so both Juro and I and our colleague Qianer as well, we’ve been reporting Deep Seek very closely since the beginning of last year. So I think one thing I like to sort of take the credit for is that I think most media allies sort of moved on after the initial buzzy period around Deep Seek early last year, but we stayed. I think we immediately recognize that this company is gonna be a very uniquelike sort of existence in China AI landscape for the time being. and we reported, I mean I was just as shocked as anyone when we got a tip when we had our very first fundraising story. Because knowing the company the way we knew it for the last eighteen months, we were I don’t know if you were about you, but I was quite shocked. I think I edited it, I wrote that story in like huge disbelief. Like it’s a funny moment for ourjournalism because a lot of times you kinda expect things will happen in a certain way or the direction of travel, at least you have a sentence on the post, but this one I think it completely surprised me. And then so in the two months that we’ve been chasing every step of the way of the first ever fundraising, one question that always lingered in my head is whatPrompted this change of this dramatic change of attitude toward external money. So eventually we were able to do a sort of the deeper story in which we revealed that it was Anthropic’s mythos preview that contributed a lot toCEO Liang Wenfeng thinking, because I think there was a period of time, if you remember, like in the second half of last year, people were in AI research community, people were doubting or casting skepticism on whether the scaling law still exists, if if it’s still going to yield kind of you know progress. and then I think mythos showed that scaling law still exists. And then that’s when Liang realized that okay,even if we have done innovation when it comes to you know… Get full access to AI Proem at aiproem.substack.com/subscribe

In this episode, I’m joined by Robin Zhu, one of the sharpest observers of China’s technology and AI landscape.We talk about some of the biggest names in Chinese AI, from Z.ai, Moonshot and DeepSeek to Alibaba, Tencent and ByteDance. But rather than just looking at who has the biggest or most talked-about models, we get into what each company is actually good at, how they’re approaching the frontier, and what Robin looks for when trying to separate real progress from the hype.We also touch on how Chinese AI companies are operating under very different constraints than their US counterparts, yet they’ve continued to make impressive progress through techniques such as model compression and reinforcement learning. This raises a bigger question around where the value in AI ultimately sits: if models become increasingly capable, cheaper and more commoditized, who actually captures the economics?From there, we get into the business of AI — how open-weight labs can make money, what AI monetization might look like, and whether the biggest opportunities will sit with the models themselves or with the applications, infrastructure and orchestration layers built around them.Finally, we zoom out to the bigger picture: what China’s progress in AI could mean for geopolitics, model sovereignty and international adoption, and how investors should think about valuing these companies when the technology is moving faster than traditional financial metrics can keep up.The AI Proem Podcast is under the AI Proem newsletter, which has over 12k followers globally. To learn more about China AI, the business of AI, and how AI is impacting businesses, please check out the newsletter here and more insightful conversations here.Chapters00:00 Introduction01:06 China’s AI Race and the Rise of New AI Labs06:39 The Compute Bottleneck — And How China Is Closing the Gap11:37 AI Monetization: Who Captures the Value?13:06 Why China Has So Many AI Labs — And Who Will Survive16:10 When Does an AI Model Become “Good Enough”?19:10 Token Rationalization, AI Harnesses and the Future of Work25:21 Models vs. Applications: Where Will AI Value Accrue?30:31 How Open-Weight AI Labs Can Make Money33:07 Can Chinese AI Capture 30–35% of Global AI Revenue?45:02 How Should Investors Value AI Companies?49:56 Robin’s final thoughts on AI’s future in China and globallyTranscript(AI-generated, for reference only)Grace Shao (00:01)Hey Robin, good to have you.Robin (00:03)Thanks for having me. Good to be here.Grace Shao (00:05)Yeah, yeah. Tell us about your coverage and your recent initiation on Z.ai and MiniMax. I think that was quite exciting. It was a huge report — 50 pages or 80 pages, was it? What made you decide that now was a good time? And what is your main takeaway there?Robin (00:23)Sure. Yeah, look, you know, I’ve been covering internet at Bernstein for a long time now. I’ve been covering gaming for the last number of years in Japan. Year to date, I think something like 80% of our research has been about some form of AI or other. I’ve been using Z.ai and MiniMax as the examples to effectively fill my exhibits and illustrate different points. The stocks kinda ran away from me as we were doing that. We initially thought, okay, we were gonna you know, work out what AI does or what these businesses do and then they all went vertical. there came a point in the summer I was just like, All right, you know, the stocks can do whatever they want given such small free floats and we’ll wait a little bit and the lockup expiries were coming up at that point and yeah, we picked a week Shortly after. I was in the US for a month to kinda network and do different things. and I think we got lucky on the timing, to some degree. But yeah, you know, now there’s more price discovery. There’s, you know, it seems to be a new model launching every other week. So yeah, fun times.Grace Shao (01:34)Yeah, sorry, we were just talking about how there’s such AI fatigue. Like, there were literally eight models over the summer and there was no summer for any of us covering AI, right?Robin (01:44)Are you not excited about Ox Alpha?Grace Shao (01:47)Everyone’s excited about Ox Alpha, but we all have different conspiracy theories, right? Well, because like I don’t wanna like you know go into these conspiracy theory holes today. Let’s focus on some of the big pictures. I do want to askRobin (01:58)Okay.Grace Shao (01:59)You are one of the rare people who gets access to these labs and their executives. When I last spoke to you, you said you were hanging out in Beijing, meeting with some of the executives at Z.ai, MiniMax and whatnot.Robin (02:09)Mm-hmm. Mm-hmm.Grace Shao (02:12)Obviously not sharing anything sensitive, but what’s the vibe? What are the cultural differences?Robin (02:16)Yeah.Grace Shao (02:17)You know, do you think any of their personalities or cultural makeup actually, you know, differentiates them from how they go to market, how they build their products or technology, or maybe even their philosophy on AI?Robin (02:31)Yeah, sure. I mean, I think it’s kind of interesting, you know, I deal with investors day to day a lot. you know, the debate there is, are these labs raising prices? Are we gonna get competition? Do models get commoditized and pricing goes to you know, gets hammered and so on. you talk to the guys at these labs and it’s a very kind of singular focus on, you know, everybody thinks they’re changing the world. AGI is very much top of mind for everybody and Iterating the model is much more of a focus, obviously, compared to investors. But the vibe is very much: yeah, just keep going, keep cranking and see where we can go. Culturally, there are some quite big differences. You know, Z.ai came out of Tsinghua University. Dr. Tang is still kind of on both sides of the fence in some ways. You know, somebody else described it as being monastic. I’m not sure I’d go that far, but it is a much more kind of academic and nerdy organization. MiniMax and all the dealings I’ve had with them seem to be more commercial. You know, they’ve had a couple of pivots in terms of what the main focus has been. certainly more international than Z.ai. but yeah, Kimi’s kind of I guess in some ways halfway in between. You know, they are More international than Z.ai but yeah, you know, you’ve got kind of the more si how do I how’d you describe it? More kind of science based aspect of you know what they’re doing. So yeah, you know, these are they show through in how these companies behave, and the results that you’re seeing in terms of model progress. And yeah.Grace Shao (04:24)How do you think they’re defining AGI? Is it different from what SF is saying?Robin (04:31)What’s SF saying? It seems to be different every few weeks.Grace Shao (04:34)Huh.Robin (04:35)I don’t know if there is a single kind of monolithic, you know what — like we’re going to do AGI and it’s this thing. you know, I think the common analogy is summoning the machine god, which... But I think it’s a little bit narrower than that. I think it’s, you know, how do we get AI to iterate our models for us? How do we get into kind of, you know, I guess some definition of loose RSI or narrow RSI? I don’t tend to get into discussions about broad RSI with people, you know, where it does actually just become a little bit more religious. But yeah, I think, you know, everyone is just focused on kind of iterating the next generation of models.Grace Shao (05:17)And they’re all kind of facing the same issue, right? End of the day it’s compute. But potentially there’s domestic compute becoming more abundant. do you think that’s really gonna change the game here? Or is that even something that’s happening in the near term?Robin (05:33)Yeah, I think it is a bottleneck. Everybody is short on compute. Everybody makes comments on, you know, if you compare the FLOPs per engineer here versus in SF, there is a big difference, and it does hold back some of the progress. But despite that, you’ve s… Get full access to AI Proem at aiproem.substack.com/subscribe

In this episode, I speak with Bei Shen, CFO of Manycore Tech, the Hangzhou-based company behind Kujiale in China and Coohom overseas. Manycore started as a cloud-based 3D design software company serving designers, furniture brands, retailers and property developers. Today, the company is expanding into spatial intelligence, using its 3D data, simulation capabilities and software to explore applications beyond design.We talk about Manycore’s evolution from startup to public company, including its IPO earlier this year and how the company’s story has changed since going public. While its core SaaS business still accounts for the majority of revenue, Manycore is increasingly positioning its proprietary 3D data and technology as a foundation for spatial intelligence and new applications.We also dive into SpatialVerse, AholoWorld and Manycore’s work in robotics and embodied AI. Bei explains how the company thinks about spatial intelligence—not simply as a data business, but in terms of the systems and simulation environments needed to help AI understand physical space. We discuss potential applications in robotics, game design and filmmaking, as well as the question of how much intelligence different types of robots actually need.Finally, we discuss Manycore’s global expansion, partnerships and long-term strategy. We explore whether spatial intelligence will become a market dominated by a few global platforms or remain fragmented across industries and geographies, and what Manycore sees as its role as more robotics companies begin building their own physical AI systems.The AI Proem Podcast is under the AI Proem newsletter, which has over 12k followers globally. To learn more about China AI, the business of AI, and how AI is impacting businesses, please check out the newsletter here and more insightful conversations here.Chapters01:18 Leaving Investment Banking for a Startup in Hangzhou05:29 From Silicon Valley to Going Public in Hong Kong06:19 First of the Six Tigers to IPO07:28 3D Shift Toward Spatial Intelligence14:20 Data, Simulation or System: What Is the Product?18:14 Learning More of SpatialVerse and AholoWorld31:59 The Role of Spatial Intelligence in Robotics34:04 How Manycore Fits Into the Robotics Stack38:11 Global Expansion and Strategic Partnerships42:31 Will Spatial Intelligence Consolidate or Fragment?46:20 Manycore’s Focus and Strategic PillarsAI-generated transcript (for reference only)Grace Shao: Hi, Bei. Thank you so much for joining us today.Bei: Hi, Grace. Good to be here.Grace Shao: Yeah. So, you guys are one of the hottest AI companies that listed in Hong Kong this year, and a lot of people have a lot of questions. But to start with, tell us about yourself. When I was learning about your background, I thought it was quite fascinating. You’re almost like a Joe Tsai story. You had a very successful finance background and a successful career in Hong Kong, and then you decided to jump over to Hangzhou to join what was still a relatively unknown startup. Tell us what made you want to make that jump and leave your cushy banking role. What was the spark about this company for you? And tell us a little bit about where the company is right now.Bei: Sure. My name is Bei Shen. I’m CFO of ManyCore. I joined the company in 2019. Before that, I was an investment banker for 14 years. I worked for Citigroup in New York, then moved to JPMorgan in Hong Kong. The last nine years of my banking career were at Goldman Sachs.Around 2018 or 2019, I started thinking about what I wanted to do with the rest of my career. Traditionally, I focused a lot on clients in more traditional industries. I covered companies in the power, mining and energy spaces. It’s an interesting job, but the sectors are relatively traditional.So I started asking myself how I could get more exposure to technology and internet companies. For me, it was very difficult to switch industries within the bank, so I started looking around for opportunities. Luckily, ManyCore was looking for a CFO. After talking with the founders and the team, I found the company very exciting, so I joined in 2019. I can’t believe it, but it’s been almost seven years now.Grace Shao: Yeah. And I know you recently took the company public, but before we get into all that, tell us about your three co-founders, because they have quite interesting backgrounds. They’re quite young. They came back from Silicon Valley, bright-eyed and wanting to start something in China. Tell us about the vision they had in the early days and where it has led you now, roughly 15 years later.Bei: Yes. The company was founded around 2012. The three founders were classmates at UIUC, which has a very strong computer science program in the U.S. Our chairman, Victor, and our CEO, Chen, actually went to the same undergraduate university, Zhejiang University, which is also where our company is based. We still recruit a lot of people from Zhejiang University, which is a great school. And our CTO went to Tsinghua.All three of them studied at UIUC in fields related to computer vision and high-performance parallel computing. After graduation, they all went off to cut their teeth in Silicon Valley. Victor worked for NVIDIA for a couple of years, Chen worked for Microsoft, and our CTO worked for Amazon.They were all working in Silicon Valley, but they wanted to come back to China and participate in this exciting market. Back then, Victor had this idea of putting GPUs on the cloud to serve more clients. He was working at NVIDIA on the CUDA team, so he was involved in the early days of figuring out how to put compute on the cloud and serve more customers. Obviously, this was before AI became what it is today.They built a demo and came back to China. Luckily, the Hangzhou government was welcoming overseas graduates and helped them start the company.The original idea was very simple: they wanted to put GPUs and compute in the cloud and make that compute available to more people. In the beginning, it was very difficult because AI hadn’t taken off yet, autonomous driving wasn’t in full swing, and crypto wasn’t either.Luckily, they found a very interesting application in interior decoration. In the old days, if you used on-premise software, it could take a very long time to render a photorealistic picture. With their technology, you could put that computation on the cloud and use multiple GPUs to accelerate the rendering process. That enabled users to create photorealistic renderings in minutes. Now it’s seconds.That really changed the industry in a big way. So that’s how they started. It was fundamentally a technology company trying to find applications for its technology.Grace Shao: How would you describe the company today? How would you position ManyCore in two or three sentences? Clearly, it’s no longer just about Kujiale and 3D interior design.Bei: Obviously. The company has had 14 or 15 years of history. Before 2023, we were basically the largest 3D design software provider for interior design. But since 2023, the company has increasingly focused on spatial intelligence.To put it very simply, we’re trying to help AI perceive, create and eventually act in a three-dimensional world, so that AI can eventually move from the digital world into the physical world. That’s where the company is focusing right now.Grace Shao: Perfect. I think you were definitely one of the hot IPOs earlier this year. You went public in April and were one of the first of the Hangzhou “Six Little Dragons” to list. It felt like a point of pride for Hangzhou and for this new wave of Chinese AI companies. What did going public mean for you and for the company?Bei: Obviously, it’s a big milestone for the company. We raised fresh capital to fund our future growth, especially in spatial intelligence. We need more compute and we need to hire more talent.But from a business perspective, it also put us on the international radar. We already have many international clients, but it can s… Get full access to AI Proem at aiproem.substack.com/subscribe

In this episode, I speak with Robert Wu, the founder and CEO of Baiguan . Our conversation focuses on two questions that increasingly overlap: how AI is reshaping the business of information, and how China’s distinctive mix of pragmatism, markets and state involvement shapes the way new technologies get adopted and financed.We start with the professional data industry. As AI agents become a new orchestration layer above terminals, APIs, and research products, Robert argues that the biggest disruption may come not to the production of proprietary data itself, but to its distribution. For niche data providers like BigOne Lab, the opportunity is to make differentiated real-time data accessible at inference time. The unresolved problem is economics: licensing, access control and who ultimately captures the value when an AI agent becomes the interface.From there, we widen the conversation to culture and political economy. Robert explains why debates about AI in China tend to focus less on existential or metaphysical questions and more on what the technology can actually do. We discuss whether that pragmatism comes from China’s history of technological catch-up, whether similar attitudes extend across East Asia, and the potential trade-off between being exceptionally good at applying technology and creating the conditions for more fundamental scientific discovery.We then turn to the role of the state. Robert rejects the simple idea that China’s technology industries are created through top-down planning. Instead, he describes a hybrid system in which entrepreneurs often discover the opportunity first, while the state later supplies policy support, capital and the resources needed to scale. We use EVs and DeepSeek to explore that model, before moving into state subsidies, local-government incentives, private capital, Beijing’s evolving approach to public markets and why so many young AI and technology companies are choosing Hong Kong for their IPOs.We close with two of Robert’s more differentiated views: that China could be entering a multi-decade equity bull market, and that outsiders often misunderstand China by assuming it has the same impulse to export its own political or cultural model. And for a slightly lighter ending, Robert explains one of the Chinese stock market’s most vivid metaphors: why generations of retail investors are compared with chives that get cut, grow back, and get cut again.btw sorry for the weird glitch in the video around 12-13 min of the recording.The AI Proem Podcast is under the AI Proem newsletter which has over 12k followers globally. To learn more about China AI, the business of AI, and how AI is impacting businesses, please check out the newsletter here and more insightful conversations here.Chapters00:00 Robert Wu, BigOne Lab and Baiguan03:09 From alternative data to the AI era06:02 Why AI disrupts data distribution12:11 Inference-time data, licensing and economics15:17 Why China feels more pragmatic about AI21:47 East Asia, belief systems and scientific discovery30:32 China’s hybrid model of state and private innovation45:24 Funding AI: state capital, private capital and IPOs50:30 Beijing’s market-stabilization playbook and policy risk01:07:29 Robert’s non-consensus views and the meaning of “cutting chives”Transcript (AI-generated, for reference only)Grace Shao (00:00)Hey Robert. Good morning. So good to have you join us today.Robert (00:05)Good morning, Grace. Hello, everyone.Grace Shao (00:08)Robert doesn’t need much of an introduction. If you spend as much time in the Substack world as I do, you’ll know he’s a prolific writer covering everything from capital markets and property to technology and culture. My favorite niche is when he calls out Noah Smith’s articles for being wrong. Those are pure entertainment for me.For today, though, Robert is a student of history, politics and business, and I think it will be interesting to have him walk us through some of the bigger questions people have about China. I’m also curious, because he runs a data company, about how he sees the future of data providers as AI changes that relationship.So I’m handing the mic over to you, Robert. Tell us about yourself, your journey with BigOne Lab and Baiguan, and how you’re seeing the business evolve.Robert (01:12)Yeah, hi. So this is Robert. As Grace mentioned, we run a newsletter. But that newsletter is really our kind of side business. The actual BigOne Lab is a team of over forty people, which exclusively most of us work on data products and research products for institutional investors and corporates. Both in China and outside of China. But we have we’ve been very China focused. All of our data and research are about China, Chinese companies, Chinese industries, businesses. The Baiguan to me was partly accidental, but partly also kind of fateful. Actually in the very beginning during college I actually wanted to be a journalist. But I didn’t find a way. So I kind of dabbled in capital markets in investing, corporate finance for a few years. But eventually it kind of hit me that, there’in this new world there’actually other ways to do journalism. Data tracking, data analysis is actually could be a new form of that. And even with data you can do more powerful storytelling and that was the genesis of our newsletters as well. Right. So here we are. We are backed by S&P Global as well, which is I would say one of the most ris backed respectable, respected companies in our industry. And yeah, so it’a brief intro about ourselves.Grace Shao (03:09)Yeah, so tell us like what is unique about your data then in that sense.Robert (03:14)Right. So we started as a so-called alternative data company. Alternative is alternative to the traditional financial data, macro data, market trading data. It’no longer alternative now. All alternative data is mainstream data now. But it was first happening, it was because the explosion of data and information in the internet and especially the mobile internet age. There are just so many data being tracked. There’payment data, there’online com commerce and social media data, so vast number of data and multiplying exponentially every year. And some investment firms they realize that by harnessing all these data and aggregate them together and put them in the right context, you could actually generate a lot of alpha that is previously not available. Right? So that’how we started the business. It was a very investment firm hedge fund driven business. So that you know kick us to look at a lot of the industry verticals, a lot of the different kind of industries where there’data and we try to find the most granular, the most high frequency data we can find. Perhaps the you know we can have massive amount of data about mobile transactions in China, for example, every day, even every minute, all the transactions that we can have access to and analyze on. So that’different from many of these you know mainstream data providers, which we try to be very granular. We try to be very frequent. Yes.Grace Shao (05:18)Yeah, so that’really interesting. I think p one thing that really stood out to me and relating it back to AI is that when we were having our catch-up conversation, we were saying, okay, data plays obviously a huge role in AI. But what you distinctly said, there is the people that are involved in the pre-training data bit, there’like the Mercores of the world. There is the people who are more pivoting towards kind of the post-training data provider, which is what you guys are doing. Just tell us about that relationship and how you think the whole data vendor ecosystem is adopting to AI and or evolving with AI, especially and how like AI is now affecting, say, Bloomberg, Factiva, those mega data platforms that we traditionally know of.Robert (06:02)Yeah. So the term data company is really problematic for us. It’really a kind of a name that covers very different kind of businesses serving different needs, entirely different kind of … Get full access to AI Proem at aiproem.substack.com/subscribe

Show NotesHi all,Christian Hu from Alibaba’s Qoder joins the podcast again to talk about the fierce domestic workplace agent competition. For context, Anthropic and OpenAI have decided not to allow access to their models and products in Greater China; thus, the competition for models and agents is largely driven by domestic players. For workplace agents — coworker-style products — the most popular ones at the moment are Tencent’s WorkBuddy, Alibaba’s Qoder, and ByteDance’s Trae. The other agents we see coming from the labs are mostly coding agents. With that, I’ll hand over the floor to Christian, who graciously found ~40 minutes while on a business trip to talk to us about the landscape and Qoder’s own strategic shifts.This episode was recorded on the day Qwen3.8 Max launched.For more, check out the podcast lineup here and explore the episodes!Chapters* 00:00 — Southeast Asia Expansion and Market Dynamics* 02:43 — The Shift Towards Business Logic in Coding Agents* 05:29 — Industry-Specific Applications and Vertical Agents* 11:22 — Global Strategy: Lessons from Market Differences* 14:00 — Partnerships as a Key Element in Market Strategy* 16:41 — The Future of Coding Agents and Market Trends* 19:22 — Pricing Models and Inference Economics* 22:08 — Open Source Trends in AI and Market Competition* 27:34 — Building vs. Buying AI Models* 29:38 — Future Directions for Coding Agents* 32:14 — Final Thoughts on Market OpportunitiesTranscript(AI-generated, for reference only)Grace Shao (00:00)Hi Christian, friend of the pod, you’re back on. Really excited to have you here. Where are you these days?Christian (00:07)Yes. I just landed in Singapore last night.Grace Shao (00:12)Okay, perfect. Cause we’re gonna talk about your Southeast Asia expansion. But since we last spoke, competition among coding agents has intensified quite a lot in China, particularly — the market has changed a little bit. Where are you seeing the whole market going, and how do you think Qoder fits into all of it?Christian (00:32)Yes, it is a war, you know, between every major tech company in China. Because it’s a war, nobody wants to lose. I think the logic behind this coding agent war is that most companies believe that a coding agent shall be the fundamental path to AGI — artificial general intelligence. So nobody wants to be behind in this race.But the most interesting thing behind the war, or this race, is that something is changing. When we look back to the last twelve months, everyone is talking about the models, everyone talking about what kind of model will be the most competitive advantage for your coding agent. But now, for most leading agents, they are trying to be more focused on the business logic and workflows. Just like what Qoder is doing — we want to be more into the business logic of our customers. And even for some individual users, they are trying to do something for business. They have one-person companies or one-person workflows. So they need the coding agent to do more about the business workflow, not just code generation.So that’s the shift behind the war. For at least most of the companies, they are trying to raise not just for a coding assistant, but want to be dominant as a desktop assistant for employees or maybe some individual users. And maybe in the future, they want to be the digital employees for the industry. I think that’s maybe the ultimate race for the coding agent.Grace Shao (02:28)And you’ve got players like WorkBuddy from Tencent that’s been doing really well, right? Exactly to your point—Christian (02:33)Yes.Grace Shao (02:33)I think they’ve been plugging in, kind of operating as an assistant on desktop, very intuitive and user-friendly. You’ve got ByteDance with Trae pushing in a similar direction. So how do you feel about that? Where is Qoder’s differentiating point or offering here?Christian (02:51)I have a few things to share with you. I think for the past year, Qoder has accomplished very remarkable business performance. In terms of ARR or revenue, we are the leading one — we are number one. And maybe we are more than the combination of number two to number four. That’s a huge achievement for Qoder in business growth.And as I just said, Qoder from day one is more focused on building an agentic platform for developers and AI builders. It’s not just a coding tool or coding assistant — we want to be a platform for the next generation of agentic power. So the skill marketplace, the plugins, the connectors, the MCP connectors — all of these features are now shining. All of these features are bringing advantage to our business. That’s the truth of what’s happening in China and even in overseas markets.Grace Shao (04:04)That’s really interesting. And so when we caught up in Hangzhou recently, you were saying that Qoder appears to be broadening from just coding into industry-specific agents. That was something quite fascinating, because from the start of the conversation, you say coding capability is the fundamental foundation for all of these agents, but people are moving towards these vertical agent use cases. So tell us a bit more about that. What kind of industries are you guys targeting? What is your thinking behind this new strategy?Christian (04:34)Okay, I can share an example. We got a very important customer case with Xiaopeng. Xiaopeng is one of the leading electric vehicle producers in China, maybe one of the best. For Xiaopeng, for the company now, Qoder is not just a coding assistant or code generation tool. Qoder has become an agentic driver for the restructuring of their workflows and business logic. They’re trying to educate their developers and their engineers, maybe even the HR department, to use Qoder to renew their business logic and their workflows.For example, for the legal department, they’re trying to use Qoder to reduce the legal review cycle from six days to one day, or even less than one day. And it’s not just Xiaopeng — I think there are a lot of similar cases. So for Qoder, it’s not a shift to pivoting to vertical applications or specific industries. From day one, Qoder wanted to be a platform. The platform means we want to be a tool — we don’t want to be just a coding assistant competing with other coding tools. We want to be more embeddable and compatible with the customers’ business logic and workflows. So it should be more vertical. But it’s not for us to do the vertical things — it should be let the developers, the customers, and even some individual power users develop domain-specific agents on top of Qoder. That’s the philosophy for Qoder.Grace Shao (06:51)Okay. So the thinking is you guys are offering the tool, but really it’s still on the user to build out the tailor-made agent for themselves. Not that you’re specifically pushing—Christian (07:01)Yes, that’s the truth. Actually, we got some organic creation of skills by the power users. They are creating some skills, some creating some plugins on our marketplace. We are going to open the marketplace to all the business users so that the users can deploy different kinds of skills or plugins from the marketplace. That will be the market for Qoder. Maybe in the near future, we could be a marketplace for agents, a marketplace for skills.Grace Shao (07:49)Interesting. Okay. Well, this leads to something else you talked about. You said that you guys were building an ecosystem around Qoder. So tell us a bit more about what that means, and how should we understand or expect how Qoder might change as a product in the next few months.Christian (08:05)Okay. I think we can change the view of Qoder — from a coding assistant to an agentic platform. I can give an example: we are doing something together with Microsoft. That may be an example. We want to collaborate with Microsoft to make Office more usable and more accessible on an agentic system just like Qoder.You know, in the past, Microsoft Office is Microsoft Office, ChatGPT is ChatGPT — they’re quite different products. For the users, they hav… Get full access to AI Proem at aiproem.substack.com/subscribe