A conversation with Manus AI's cofounder and CPO Tao Zhang

I'd love to take us back to the first time that I learned about Manus, and probably the whole world. So it's March 2025, we've had AI chatbots for 2–3 years, everyone's been interacting with chat, but that's really it. And then a video starts circulating the internet. No flashy stage, no Silicon Valley keynote, just a raw screen recording of your product, Manus. Tell us what this video shows the world. I think it's just 7 days before the launch day, and when I'm looking at our website, I think it's too simple. So, I was thinking maybe we should make a video for our launch. So I called some video agency, and they all told me that it will take them maybe 2 weeks or 3 weeks to build a launch video. I told them that we only have 6 days, and they said, "Oh, this is impossible. " So, we decided to shoot the video by ourselves. I'm using my own camera, I shoot the video, and Peak, our chief scientist, is the actor there, talking about the product. I shoot it, I edit it. Because we're not a very professional launch studio team, we only put Peak in the camera for only 1 minute talking about our product. Rest of the time is Peak demoing our product for three use cases. That can be the main reason why we went super viral in the first place, is because we just show the work. What an AI product could be other than Chatbot. We actually have that video right here, so let's take a look. Hi, I'm Peak from Manus AI. For the past year, we've been quietly building what we believe is the next evolution in AI, and today, we're launching an early preview of Manus, the first general AI agent. This isn't just another chatbot or workflow. It's a truly autonomous agent that bridges the gap between conception and execution. While other AI stops at generating ideas, Manus delivers results. We see it as the next paradigm of human-machine collaboration, and potentially a glimpse into AGI. Now, let me show you Manus in action across three completely different tasks. Let's start with an easy one. In this example, we'll ask Manus to help screen resumes. I've just sent Manus a zip file containing 10 resume documents. Since each Manus session has its own computer, it can work like a human. First, unzipping the file, then browsing through each resume page by page, and recording important information in documents. Manus works asynchronously in the cloud, which means you can close your laptop anytime, and Manus will notify you when everything is complete. The video shows basically, that with Manus, you cannot just chat, you can actually execute on tasks, and it writes Python code, and it goes and opens browsers. Yeah, it does everything. And it goes entirely viral. I think you had 2 million signups for the private preview after that. Yeah, we had 2 million in the waitlist in the first week, and we got 3. 5 million in the first month. A lot of people on the waitlist. I read somewhere that people were reselling the invite codes for up to $14,000 online. Yeah. At that time, if an influencer, got an invitation code to Manus, he will attract much more traffic comparing to the price he paid for the invitation code. But, unluckily, we didn't get any money from that. No, of course not. That's a whole new… We can do a whole different podcast on the thought of the Internet now with AI, but we'll leave that for a different time.

What do you think was it in the video about the product that struck such a nerve with people? What was the aha moment for people in that time? Yeah, because before Manus, people treat AI just as an answer machine. I give you a question, and you give me an answer. It's still telling you how to do things. But in the end, you have to do things by yourself. But in our launch video, we always focused on the outcome, on the deliveries. Maybe it delivers you sites, it delivers you a website, or it delivers the actual code that works. I think people just get amazed. Another thing is that in the whole process, when the AI is doing things, it's totally without human intervention. You're just telling Manus what you want, and Manus will figure out what's the plan and how to execute the plan step-by-step just by itself. It's all automatic. I think that's really something new and amazing for the whole world to see. I remember thinking… It was the first time where I felt, Oh, this can actually save me time and help me do my job. Versus just a little bit of a thought partner. It was really, really magical. Did you know it was going to be so successful? The night before launch, were you were like, yes, we've hit on something really special? Yeah. We were a very small staff at that time. We only had 40 people there. We thought it's a very good product, because when we developed the product, we were getting amazed by our product every day. Every day in our office, it's a feel of, Oh, it's awesome. Oh, damn. How can Manus solve this? Every day is somehow like this. So we think it's going to be successful, but I think nobody were expecting that we can be at this level of success, where we are. -It's a high bar to clear. -Yeah. This is a good time to talk about the product, I think. From the very beginning of Manus

I feel like you really reinvented what agentic AI products should look like. I wanted to maybe go over some of these product areas with you, and talk through some of the thinking that you had behind them, and why you developed them. Virtual sandbox, tell us what is that? The virtual sandbox, internally, we call it virtual machine. It's super important. As it's going to be somehow as the fundamental part of the whole Manus system. Because for us, we think in the past, why most AI wouldn't work is because they only use the brain. Maybe here is a time I can just share why we choose the name Manus for this product. We take this word Manus from MIT's motto, which is Mens et Manus. It's an old Latin word, which means mind and hand. And mind, we think they are LLMs. They are the models, they are super smart brains. But only with brain, us humans, we can make real impact to the physical world. That's why we think we have to be the one who build hands for this very smart brain, then they can operate tools, and they can make impact into the physical world. When I say what wouldn't work for other AI tools is because they only have the brain. What they do is they keep thinking, keep thinking inside the head. But what we do is that we provide a computer to the AI. Anything the AI thinks, "Okay, maybe I should do something, or maybe I should run some code to test my thesis. " It can use its own computer to browse the internet, to run the code, to store a file on the file system, or maybe 30 minutes later, to retrieve the document again from the file system. The virtual machine really changes everything because it provides maybe the most powerful tool, the computer, through the AI. Now you're building tools for the AI to do its job. That's really, really incredible. One of the interesting product decisions that I thought you guys made was for the user to be able to see what was happening, like opening the browser, writing the code, etc. What was behind that decision? Actually, that's the first important decision we made after we kicked off the project, because we saw that agent is so new and people are not very familiar with this product. So we think it's very important at this stage, maybe it will last for 1, 2, or 3 years, to show every step, the whole process to the users, to make users understand what is happening in the background, and then they can trust the product. This is very important. We build transparency as a feature in our product. It's an incredible feature. Tell me a little bit about the Async workflow, because you create that trust, and now the user can step away and not really worry. Where did that come from? In our founding team, we're all engineers. I run the product team, but I write code for more than 30 years. We all know coding, we're all Cursor fans. We use Cursor a lot of times. But there's a problem that when you are using Cursor, you have to keep your laptop open because it runs locally on your computer. You have to keep it open and watch it. Because in the first maybe 4 or 5 months, it needs you, us humans, to click the accept button. Sometimes it will ask you, accept, accept. When we use the product, we say, "Oh, why does the Cursor need me to click the accept button? " It's because it runs locally on my computer. It needs my permission to run this command, or it has a chance to break the whole computer. But in Manus, the computer is in the cloud. It's not in your computer. It's super safe to perform each step just by the agent itself. We don't ask the users to keep pressing the accept button. Instead of that, we let the agent run everything in the cloud. Then it gives us another advantage, which is the Async advantage, because everything is running in the cloud, which means after you assign your task to Manus, you can just leave it. You can just close your laptop, you can put your phone back to your pocket and do your own things. After maybe 10 or 20 minutes, when the work is done, we will give you a notification, and you will see the outcome. What do you use Manus for? Nearly everything. For me, it's like I'm doing prototyping every day with Manus. It's like whenever I have a new idea, I will not just ask my designers to put the ideas alive to me, I will use… Instead of that, I will just tell Manus everything in my head, and the Manus will build the first prototype to me, and this is not just something we're prototyping in Figma, this is a fully functional prototype, which means it has the front-end, back-end, and it has the AI capability inside that. In this prototype, actually, you can just try and play with it and tweak it. After the prototype is done, I will give the prototype to my designer, and the engineers, and they will instantly understand what I want because this is the real thing. It's putting the fun back in software development, I think. Who is your ideal Manus user? Right now, we describe them as long tactical users. Because in the first month, when we kick off the project, at that time, we already have Cursor, Windsurf, Devin, all these fancy tools for engineers. We decided that we will not build another fancy tool for engineers because they have enough. But for normal people, for average users, they are underserved. They don't have these fancy tools. They don't have cloud code, they don't have Codex. The way we built Manus is build Manus for everyone. The ideal user is someone who wants to achieve much more with AI, but with less technical background. I love that. There's so many more of them. Yeah, them, exactly. A bigger market. -You've been coding for 30 years? -Yeah. Do you think it's different building for AI products, or do you think really fundamentally, software is software? I think building AI software is totally different. For me, I'm already in this industry for more than 15 years. Before, the way we built product is somehow like, I have an idea, and I just build the prototype on Figma, and then I talk to my designer, talk to my engineer, and they will just do the design, do the engineering design, and then write code, we test it, we launch it, and we hear the feedback, and then we just iterate. That's the way we did things before. But now it has changed. Because previously, product manager and designers, what their job's outcome was somehow like a PRD or a design. It's about interface. But In AI world, the interface doesn't exist anymore. For most AI product, it's just about conversations. The conversation is the new interface, which means the prompt itself is the interface. I know there are many companies. Today, their prompts are still written by engineers or researchers. But I think a better way to do that, is the product manager, write the prompt themselves. Because the way you write your prompt, actually, it's just like the way you design your interface like 10 years ago. -We do since before. -It's the new PRD, essentially. Yeah, it's a new PRD, it's a new interface. The way we do product here is that whenever we have a new idea, we will just use Manus itself to prototype with our idea. After we feel the function works really well, and this is the feel we want. Then we will go to our engineer, and ask them about how to put this prototype in our production environment. Once the engineer told us, "This thing can be delivered in our production environment," the last step is to go to the designer. The whole process is really in the opposite. -Wow, amazing. -We have the final experience first. Then we have to make it engineering ready. The last step is to ask the designer, "Everything is ready, just do the design. " You've completely changed the sequencing -and the process. -Yeah, completely change it. Amazing.

At Stripe, we like to paraphrase this quote from Pablo Picasso, which is basically, "When art critics come together, they like to talk about form and function and meaning. When artists come together, they like to talk about where to buy the cheapest turpentine. " Yeah, definitely. Actually, I put that sentence in my personal profile. -Really? -Yeah, I really love that sentence. That's amazing. Well, we love it here, too. One of the reasons I love doing these types of interviews is that the builders and the co-founders of these companies, you guys are the artists. I want to talk real tactics now about how you've built Manus and really built the company and gone globally. -You're headquartered in Singapore. -Yeah. You now have teams in Tokyo and in San Francisco. My first question for you, when do you sleep? For me, it's a harder question because I'm the one doing all the public speeches around the world, so I really travel. So I will sleep whenever I can, on the plane, on a car, and whatever. But I think for most of my team, it's like they will sleep very late in the midnight. Usually, I think our core engineering team, they will sleep after maybe 1:00 AM, 2:00 AM. -That's very long. -Wow. My question was meant in jest, but I'm glad you answered. -You have a very hardworking team. -That's very normal. But definitely, they will wake up much later. So usually, our office starts after 11:00 AM. Because most of us will come to office after 11:00 AM because they sleep very late in the midnight. You were saying you spent 400 hours on a plane last year. Oh, yeah. I spent 400 hours in the air. I'd love to see those travel rewards for you. You say there are about 100 people working at Manus today. Yeah, 100 people. Most companies don't build global teams until they're much, much bigger. Why did you decide to build out different offices in different locations so early on? Because after we launched Manus, we went super viral in the whole world, and we have a lot of Manus fans in different countries, and we believe it's super important to be there. We don't want to just be like someone in the cloud. We just want to do face-to-face. So even after maybe just 15 days after we launched Manus, I came to US, and I traveled from West Coast to East Coast, and we do more than 15 user meetups in six different cities. We just want to do that because Manus is a general AI agent, which means we can do many things. But in the end, actually, even for us, we don't know what the users are using us for. So we think it's super important to get to know them, to do face-to-face communication. That's the initial intention to open international offices. But later we found out it's really important to open sites in different countries because in different countries, there are different cultures, different how people do business there. So opening an office there will make us much easier to talk to local markets. You were telling me something interesting right before this conversation that when you guys opened these offices, it was very important to you that you kept the product and engineering and research all together. Tell us a little bit more about that. In our company, it's that we think it's very important to put the product guy, the engineering, and also the researchers sitting together. I'll just give you an example. Back in our previous office, I and Peak, our Chief Scientist, and Bin, which is our full stack guy. He's a real full stack guy because he's not a full stack engineer. He does everything. -Our logo is also designed by him. -Amazing. Yeah. So he's a full stack guy. We sit together, and every day, we will have many small talks. He's like, "Okay, I saw something very interesting. Can we try that? " When we sit together, it's very easy to try out some new ideas. I know there are some teams, they will just separate different teams. This is a product part, this is the engineering part. We insist that we should sit together. Right now, it's still me, Bin, and Peak, we sit together, and every day, we have many small talks. We believe that those small talks leads to all the innovations we did in the past year. Especially probably for a product like Manus, which is so far-reaching and so cross-functional. You said something interesting about how you structure your engineering team. Most engineering teams are structured by product areas or even the customer journey. How do you guys decide on who works on what in the engineering team? Our CTO, Pan Pan, he has his own philosophy about managing an engineering team. Right now, we think we have maybe 40 or 50 engineers, and they are in a very big team. We only have one team, the engineering team. -So whenever we have a- -50% team. Yeah, when we have a new project, Pan will just assign someone to working on this project. But after that, they will come back to the big pool. No one is specific, just appointed to a feature or a project. We think it's very important because it will make that much flexible. So a project comes up, and you're like, "Who here is available? " People raise their hand, you're like, "Great, you two go work on this particular project. " They're done with the project, they come back into the general pool, -and then it's the next project. -Yeah, that's it. Who maintains the code and the product after they're done with the project? That's a good question. I think the way we run our engineering team is also somehow benefits from the AI capabilities. Because I think before, why we should let the same engineer keep working on the same project is because we need a context. It's really hard to read others' code, right? But right now, with the help of AI, it's really easy. If we give an engineer, we think, "Okay, you should work on this," and there are some existing code, he can just use the AI to read all the code database and explain how the whole project works. So it only maybe took him 10 minutes to understand a very complex project. But before that, it would take days. It's an incredible unlock because now you have everyone, and also people just by definition of having different projects and coming back, they will have more context anyway, because they're working on so many different pieces. I think that's amazing. I hope that's the future of product development and building products.

I run global teams. One of the things that I found the hardest about running global teams, in addition to maintaining context, which I think maybe may end up getting solved, is how you create a consistent culture. How do you think about that? How do you make sure that the Tokyo office and the Singapore office and the San Francisco office, they feel the same when you show up? What are your tips and tricks for that? Yeah, because for Tokyo and San Francisco, mostly the marketing and the go-to-market team, and it's under my watch. So I might be the right person to answer the question. I think it's always about how you lead because to do management is not just to manage. It's always about making decisions. So you will let every team members know how the company says things and how the company treat different trade-offs is through decisions. So when I make every decision, I will tell the related person about why I made that decision, what's my thought behind the decision. It's very important to telling employees about that. The second thing is about we should have a consistent the way we do marketing and the way we tell the users about our message. So whatever we do in different countries, we will review just in our weekly meeting about, "Okay, this is the way we do offline events. This is the message we should tell our users. " That's how we align. We align on the message level. I love that because, like you said, so much of the culture is just top down from the leadership. What are they showing, and how are they behaving and acting? -That really shows a lot. -Because everyone will learn. Yeah, exactly. Everyone will learn from that. So you went from zero to 100 million in ARR in just 8 months?

-Yeah. -How? Okay. I know a lot of people are very curious about that. We are also curious about that, too. Because before last November, which is in the first 8 months, we literally put zero marketing budget into the whole product, which means most of the traffic are organic traffic from everywhere. I can just give you some examples. I think it's back in last June, we were at the top place in the app store of Egypt, and we don't know why. We just checked. We did some research with Manus, and we found out there is some random guy on Facebook. He writes a post with Arabic language talking about Manus, which made us somehow viral in Egypt, and it makes us at the top place. We didn't know that before the post. Also in Brazil, we have a huge user base there. The initial wave is two influencers on YouTube. They made two videos about us, and we went super viral. We didn't know these two influencers until then. So for us, I think a lot of viral wave just came from word of mouth because as you just mentioned, the first time you saw our video, you think, "Oh, this is something new," right? And they got amazed. They were like, maybe write something or do a video about us and introduce us to their friends. I think in the whole first 8 months, it's all about the product sales, it's about word of mouth. So your growth tactic is just build an insanely good product? -Yeah. -That's your feedback. Because I know in many teams, they will treat growth as a separate sector in their company. But for us, we think the best growth is the product sales. Yeah, it's super important. But definitely, after last year, the management team, we just had some review about the past year, and we saw there are some important decisions in our product that made us more viral. The first thing is about just maybe 20 days before the launch, we decided to build a feature, we call it the Session Replay. You remember that? It's like every task, whenever it's done, when you share the task to your friends, your friends will see the whole session in replay mode. It's like a… -We think that's very important because- -Very powerful. -when you share a session to your friends, if your friends only see the final step, they will not get amazed. Just another AI product, right? But when they see the session, you're in replay mode, it's like, "Oh, what is going on? " The AI is controlling the browser. The AI is writing a code. I think that feature is really important for our growth. You also have the build first, sign up later feature. It's using Stripe Sandboxes for that. So we love that. Has that helped at all with your growth? Oh, yeah, I think that part helped the growth of our Manus 1. 5 feature. Because in the version of Manus 1. 5, we introduced a new way to build websites. Because before that, Manus can only build static web pages. It's just static HTMLs. But in Manus 1. 5, we can build a fully functional website with the front and back-end databases and native AI capabilities. For that, many users are asking us about, can you integrate Stripe into the system? Then they can just run their real business on Manus website. But as you know, Stripe has the best developer experiences. But even as a leader, because our ideal customer is a non-technical person, right? Oh, yeah. It's work to integrate it, right? It's really hard for them to understand what's a token, blah, blah, a lot of things. We think the Stripe Sandbox really helped us to make the whole process much simpler. Even for non-technical users, they can build a functional business like an e-commerce website without any knowledge about the technical details. Once he is done with the website, and then he can just translate the Stripe Sandbox to the real environment. It's an incredible feature. I'm so excited to see all the businesses that are going to just happen because tools like Manus. It's really amazing.

One of the really cool things about Stripe, actually, is that we get some insight into all the businesses that build on top of Stripe and how fast they grow. One of the insights from us is that these AI cohorts just have fundamentally different growth curves. In this new world, how do you measure your success? Because you don't have that many benchmarks to go off of. I think for us, it's like we care about two metrics. The first is about AI. It's a traditional metric, but we think it's much more important in this AI era because right now AI is still very expensive. You can't just make a free product. I work in this industry for more than 15 years. In my early years in this industry, what we were always talking about, don't care about the revenue. Just bring the users in. After you have millions of daily active users, you can find a way to make money. But today, I think things are changing. You must just ask the users to pay for the product in the first version. Because the AI is expensive right now. It's really important to earn money, not just for revenue, but to prove your PMF. Someone will pay, only they feel the value is real. We think it's super important to focus on the AI because that means a lot, not just for revenue, but also for your PMF. The second metric we are tracking is about the user's rating -because after every task is done- -User ratings? Yeah, we will ask the user to rate the task 1-5 stars. That tells us a lot about our quality, our performance, because you can always see these ratings through different segments, by different features, by different countries, by different pricing tier. -Super useful. -That tells us a lot. Right now, we are following these two metrics, and we are analyzing every week. Speaking of ARR, how did you think about pricing? Pricing is actually pretty important, but we don't know what's the best approach. We learn from each other. Actually, if you are looking at today's cutting-edge AI applications, them from a yearly scope, and you will see everyone changes. Sometimes we will learn from others, and 6 months later, we found that they learn from us, too. Everyone has changed their pricing plans. Everyone's looking at the other. Yeah, so we don't know the best practice here, but we think it's very important that you keep something in your mind, which is, I think in today's AI world, there are some very heavy user, very powerful user. They will feel like they will get much more value than other users, so they're willing to pay more. So always leave a space there to make the users can be charged by usage, like usage-based billing. It's very important. Even you have a monthly subscription plan, but always leave a space there for users to pay more. It seems like it's unbounded in some ways. There are users who are willing to pay a whole lot more, which is- Also, I think this is not just about the revenue. It's also about signals. Because for us, our top plan when we launched the pricing plan is $200. We think it's a lot. After we launched the plan, and we leave an option for our users, which is that you can recharge your account after you hit the limit. Luckily, we found out there are some users who would pay us $5,000 a month. That's a super positive signal to catch because someone pay $5,000 a month, which means the output of the Manus may value 10 times more. We will do user interview, and we will find out, "Okay, what are you using us for? Why do you think it's so valuable? " That provides us many insights for the future product roadmap. Any insights you can share from those conversations? Definitely. Why we made Manus 1. 5 is because we are looking at there are some users who will build web pages, even static web pages in Manus, and they spend a lot. We ask them about, why do you want to build these web pages there? They told us, in their country, it's very hard to find engineers, even outsourcing engineers, to build a business website for small-medium businesses for them. So for them, spend $1,000 on Manus is super cheap. That gave us some inspiration because we live in a very big country, and every day we talk about US is also full of tech people, right? So we don't feel that pain. But the pain exists in the real world. There are people we never met that they have this pain. After we understand that, we think, "Okay, maybe we should make our website builder much more powerful. " That's why in Manus 1. 5, we introduced the fully functional website building. Yeah, front and back-end databases, a lot of things. What a fantastic insight. You're just bringing access to parts of the world that just never had it, and folks can just start their own businesses entirely on top of Manus. Just even 5 years ago, the bottleneck to being able to execute

an idea like this or do a task or be more efficient, it was coding. That's no longer the case anymore. What do you think is the bottleneck now? Is it people's imagination? Is it being able to prompt? Yeah, I think right now, the bottleneck is still the capability of defining a problem. Because solving a program is not a problem anymore. But to define a problem is somehow- I think that most people still lack this capability. Because in today's world, the AI still needs you to ask the question, to assign the task to them. So you have to be the one to watch around like, what's the pain? What's my customer's pain? What's our organization need right now? You can just generalize questions from this real user's pain or what the real needs inside your organization, and give the task to the AI, ask it to solve it for you. But if you can't see these questions, problems in your life, you don't have any task for the AI. So I think that's somehow still being the bottleneck. Do you think eventually AI will also help us ask the questions themselves and define the problems? I say maybe, yeah, maybe. But it is still very important for us to learn. I think actually it's fundamentally changed the way we learn, and we execute on our daily jobs. I think before, it's like only a few people can be the manager. Because definitely, not everyone can be the manager because that won't make the company run. -You need people to do the work. -Yeah, but right now, I think everyone can be the manager, right? Because if you know how to manage, you can manage 10 or 100 agents working for you. I think this is a new era for everyone to learn how to be a good manager. This is maybe a great moment for my final question. So fast-forward 20 years in the future, AI is doing a lot of the tasks and the execution. What do you hope the world will look like? I know a lot of people are afraid of that future because AI is taking my job. But if you're looking maybe 100 years ago, when we just invented cars, a lot of people were thinking that, too. When we invented machines to lift the heavy things, we were like, "Oh, I don't have my job. " Because before that, maybe we need 20 people to lift some very heavy things. Right now with a machine, one man can do that. But let's just say in history, after we have cars, we have machines, after we somehow release this labor from this tedious work. There are so many things we found out to do. Like arts, many new things, new business models. There are so many things. I think AI is somehow, we like to see. I think 20 years later, there are so many tedious works we don't have to pay our attention on. But definitely there will be more and more important things for us to do. I think it's a beautiful thought, maybe it'll get a little closer to what it means to be really human. I'm so excited for you and Manus to help us get to that future vision. Thank you so much, Tao, this was fantastic. Really enjoyed chatting. Thank you, Eeke, me too.