$10M ARR in 60 days with context engineering

1. 0:00 – 1:19

   Intro

   1. AGAakash Gupta0:00

      Why do so many people's prompts fail and they have to keep tweaking them?

   2. XWXiankun Wu0:04

      Probably you need to buy a course to learn how to prompt, right? It actually makes no sense. I expect the AI can deliver exactly as people wish within such a short description is basically impossible. Context engineering is one of the most important skills in 2025.

   3. AGAakash Gupta0:22

      Context engineering is the secret that 99% of people get wrong. While others burn through VC money on generic AI wrappers, Xiankun, or X.K., has cracked the code on visual AI that actually works. This startup passed 10 million ARR in 60 days with zero fundraising and zero advertising. The hidden story nobody knows.

   4. XWXiankun Wu0:43

      At the very beginning, actually-

   5. AGAakash Gupta0:44

      That's a genius strategy that you have come up with

   6. XWXiankun Wu0:47

      ... in data. I think all productivity tools might die in the future.

   7. AGAakash Gupta0:51

      Wow. What is the advice you would give aspiring AI founders who want to reach 10 million ARR?

   8. XWXiankun Wu0:57

      My advice would be...

   9. AGAakash Gupta1:01

      Really quickly, I think a crazy stat is that more than 50% of you listening are not subscribed. If you can subscribe on YouTube, follow on Apple or Spotify podcasts, my commitment to you is that we'll continue to make this content better and better. And now on to today's episode.

2. 1:19 – 5:23

   Why Prompts Fail

   1. AGAakash Gupta1:19

      X.K., welcome to the podcast.

   2. XWXiankun Wu1:20

      Thank you so much for having me here.

   3. AGAakash Gupta1:22

      It's a pleasure to have you after looking at you online for so long, and my first question for you is, why do so many people's prompts fail and they have to keep tweaking them?

   4. XWXiankun Wu1:32

      So I think there is a fashion, like people will believe that one prompt will like create anything for you, but actually think about that. If you hire somebody, that person knows nothing about your company and your goals and your progress, and you want like just talk to that person like within five seconds and that person can deliver anything to you, it would definitely be impossible, right? So-

   5. AGAakash Gupta1:53

      [laughs]

   6. XWXiankun Wu1:53

      [chuckles] The current situation of like vibe coding is kind of like in th- that situation. So, uh, I think essentially now like AI is now powerful. Like to be very frank, like the, the latest model is very powerful, even though it, it is not that powerful as people expected. But people expect AI can deliver exactly as people wish within such a short description of the goal of the progress or the status of their work is basically impossible. That's the reason why people need to like tweak their prompts or modify their, you know, prompts again, again, again. Like it's, uh, also a capability of the AI model thing and also expectation management of people problem.

   7. AGAakash Gupta2:35

      So you guys solve this by enabling richer context, visual context, putting in PDFs. Why is that so important?

   8. XWXiankun Wu2:43

      So I, I wouldn't use the term that, uh, we solve kind of like a problem. It's more like we... First, a- as I mentioned before, it's kind of like a two-dimensional problems. One is the technical problem. One-- another problem is actually how you can manage people's expectation. It, it is nearly impossible. Like no matter how powerful the AI will be, it is almost impossible that if that person, if that AI don't know what you want your project's progress, it, it is basically impossible that use the technical techniques that to... any techniques to solve that problems. So our solution is kind of like, okay, we try to persuade you that one prompt is not the way that we should use today. You should like accumulate your information and the materials within a one-- all in one place, so that you can accumulate and AI will know you more and more as you like use them more and more. Rather than like other like a typical chat interface AI tools, it work kind of like a more like a one-off tool that you, if you like want to make AI slides or you want to have some like Q&A sessions, you ask questions about that and it give you a prompt, give you a, a prompt information or it give you an answer that you immediately want to know. But we encourage you to accumulate, encourage you to build more context for AI so that you need to give them a little bit patience. You give them a little bit time. You use them more so that the AI will know you more. In the end, they will deliver a much better results than the, uh, initial, uh, tryout or whatever. So our solution, uh, that's the first part that we want to try to persuade people to spend a little bit more time getting along with the AI-

   9. AGAakash Gupta4:29

      Mm

   10. XWXiankun Wu4:30

       ... uploading more materials and providing more context to the AI. The second part is like, uh, chatting with AI is a very important part of, you know, telling AI what you really want. But there are a bunch of other ways to, uh, express yourself, like express your intent. Just like, okay, if you have a spatial ways to describe the relations between different objects, including like images or documents or whatever, so that AI can understand you without talking too much. So we are kind of like providing different tools so that you can easily and effectively to express your intents, so that like the two dimensions and two solutions combined together, the AI probably can know you better without less effort and deliver better and better results along with the time.

   11. AGAakash Gupta5:21

       And the proof is really in the pudding

3. 5:23 – 7:23

   $10M ARR in 60 Days

   1. AGAakash Gupta5:23

      here. You guys hit 10 million ARR in 60 days the last time you shared your numbers. What is the latest? What is your latest ARR and user numbers?

   2. XWXiankun Wu5:31

      We're like growing very fast. Like, uh, the number is definitely higher than before. Then we like, uh, we launched at 1.0, like, uh, I guess like two months ago, nearly two months ago, I guess, and number is getting bigger. What I can say is, uh, like we just launched like two, Kuse 2.0, and the, the peak traffic is 6X, and demo requested 10X before the Kuse 1.0 launch. And, uh, yeah, the, the numbers really like, uh, blew up.

   3. AGAakash Gupta6:01

      Most founders, [chuckles] they take years to hit 10 million ARR, and that too with millions in VC funding. Can you take us inside the 60-day sprint? How did you guys hit 10 million ARR so fast? What were the key milestones? What are the key philosophies and methods that other people can learn from?

   4. XWXiankun Wu6:19

      I want to be very frank here. Like if people thought that we only have like 60 days like history, but actually we have been like silently building the product for quite a long time. I guess like we have been building this since like, uh, early 2024, I guess. You know, the media outlets would tend to make the, you know, timeframe shorter, but actually we have been building this product for quite a long time. I want to be very frank here. So it's not just overnight-Like magic, but actually we accumulate a lot. We build up a lot of attention, uh, build up a lot of contacts, build up a lot of connections with local, like, uh, communities, like Hong Kong and Taiwan's, like, teachers' communities along the way. And, uh, we just, like, don't do, like, a marketing campaign before so that people on Twitter or on the other social medias might now know us. But actually we have do... We have done a lot of work before.

   5. AGAakash Gupta7:12

      Mm. So take us back. What did you start with? What's the hidden story nobody knows? What did you start with in early 2024? What happened in between then and the time when the 60-day clock started?

4. 7:23 – 9:07

   Hidden Story: Design Agent Pivot

   1. XWXiankun Wu7:23

      Yeah. So, um, at the very beginning, actually we want to create a design agent in the very early version. That's the reason why we have, like, an infinite canvas here. I see myself as a, you know, half like graphic designer. Like we said that probably we should create something that we are in really interested in or we know about. So essentially, uh, i- initially, we want to create a design agent at the very beginning. But later on, like, we found out people just upload a lot of files and documents into the product. Since we want to build a design agent, we allow people to upload their requirements into the canvas so that the agent can understand what they really want to design and convert those, like, documents or requirements into posters of whatever. But we find out that the frequency of people using it as a knowledge base or analyzing files and documents so much more than actually designing something. Part of the reason might be like the, you know, image models was not that powerful back then, and, uh, final design results might not really meet the requirements of users' expectations. But we see the trend, so we double down, say, "Oh, if our users really love that, like, uploading files and ask our AI to analyze and read this AI files just like chat PDF, might... We might need to, like, emphasize and reinforce that part of features to cater to our users' need." So probably around, like, late 2024, we decide that, uh, we should now continue building specifically a design agent. We should go for a more horizontal, like, knowledge-based AI.

5. 9:07 – 11:28

   Threads Growth Strategy

   1. XWXiankun Wu9:07

      So we start to do, like, user acquisition across different platforms, especially on Threads. Threads has been-

   2. AGAakash Gupta9:15

      Right

   3. XWXiankun Wu9:15

      ... a platform that has been, uh, ignored for quite a long time, especially in the US, 'cause people, you know, in the US especially, like, pay a lot more attention on X or Twitter. But, uh, actually Threads as a product has been growing so fast, and it's actually still growing, and especially in Taiwan or Hong Kong, where, uh, our users and clients, uh, primarily come from. It's a very popular application. So probably we are the very early batch of, uh, companies who did heavily social promotions in Threads. And actually, we... I think that we did a great, really great job. We s- almost spent no, like, money, and we hire a bunch of intern armies, and they create a lot of c- contents without, like, uh... 'Cause, you know, like, Threads even don't, don't have, like, a promotion, like official promotion ads platform so far. You-

   4. AGAakash Gupta10:14

      Yeah

   5. XWXiankun Wu10:14

      ... you don't... Y- y- you, you, if you want to spend money, you don't have... You have nowhere to spend money on. And our intern army is very smart, and as I perceive that, uh, Threads is a very special platform that, uh, 'cause it's a Meta's application, so it's grows, you know, outgrows so almost any apps in the history. So it doesn't build up a very, you know, structured, you know, creator system as YouTube do, and YouTube and X do, right? So actually, Threads give you traffic in a very generous way, uh, and we create, like, uh, hundreds of accounts, uh, you know, creating, you know, uh, use cases every day. And, uh, there are a bunch of features really win the hearts of our users, and it's, uh, like, uh, for example, like, uh, generating exam papers or a feature we call Formatter. It's ac- actually a kind of like a formatting or layout AI agents, 'cause there are so many industries people still need to follow a format that their companies requires. Uh, our Formatter or AI layout-

   6. AGAakash Gupta11:26

      That's amazing

6. 11:28 – 12:20

   Ad Start

   1. XWXiankun Wu11:28

      ... features can-

   2. AGAakash Gupta11:28

      I hope you're enjoying this episode with XK. A quick word for our sponsor, Reforge, who just released Reforge Build. You know that feeling when you're trying to prototype something with AI, and it spits out something completely generic? That's because AI prototyping tools like Bolt, Lovable, and Replit, they are built for founders, for people who are designing from scratch. But product teams aren't building from zero. You have an existing product, real customers, design guidelines, a backlog full of ideas you need to explore and validate fast. Reforge Build generates prototypes that reflect your real pricing tiers, your real features, your real customer language, not generic placeholder. It's a dream come true for product managers who are embracing the new way of working with AI prototyping. So check out Reforge Build at reforge.com/aakash. That's reforge.com/aakash, and use the code BUILD for one month free of premium. And now back to today's episode with

7. 12:20 – 17:06

   Threads Accounts Demo

   1. AGAakash Gupta12:20

      XK. Could you maybe show us some of these Threads accounts and what kind of content is working on Threads?

   2. XWXiankun Wu12:26

      Sure.

   3. AGAakash Gupta12:27

      I've not heard anyone talk about this growth strategy. I think a lot of people are probably sleeping on this channel. This is a huge insight. And the intern army component is also very interesting.

   4. XWXiankun Wu12:42

      You see here, like-There is a lot of, uh, contents we cr- we are creating every day. It's kind of like a formatter. It's a use case, you just need to like input and mark it on, like without any formatting or layouts, and without, with our formatter features, you can turn the content into a, uh, well-polished layout, uh, documents here.

   5. AGAakash Gupta13:13

      Wow.

   6. XWXiankun Wu13:13

      So there are so many like accounts here. Uh, let me show you other ones. Basically, all of this are our accounts. [chuckles]

   7. AGAakash Gupta13:26

      Wow.

   8. XWXiankun Wu13:27

      All the, all the accounts here are our accounts.

   9. AGAakash Gupta13:32

      And you said they translate pretty well to traffic? Because I average like 3M impressions a month on Threads, and I get like hundreds of website visits.

   10. XWXiankun Wu13:41

       Yeah. So you can he- see here, like we make a lot of use cases here. Basically, like we pretty-- We're right now pretty focused on the Taiwan and Hong Kong markets, so they're basically in traditional Chinese way. Yeah, so you can see here-

   11. AGAakash Gupta13:57

       Are there many competitors for you in those markets, or is one of the insights here that you've targeted kind of a geography where maybe other AI applications aren't as focused?

   12. XWXiankun Wu14:08

       I think that, uh, AI applications in Hong Kong and Taiwan is not that competitive in China and in the US, and Threads is not a crowded place, as many people just ignore that channel. There's actually a lot of people-

   13. AGAakash Gupta14:23

       Yeah

   14. XWXiankun Wu14:23

       ... a lot of real people. Even though that I talk to some like U- people from US, and people will say, "There is no real people in Threads." I would say like, "It is not true." There are a lot of real people in Threads, but you just not one of them, and you are not in that circle. But actually-

   15. AGAakash Gupta14:43

       Yeah

   16. XWXiankun Wu14:43

       ... there is a lot of like real humans and real users, active users in Threads you can, uh... And basically, like I know you are very successful on Threads, but, uh, Threads is, uh, I, I know you are very successful on X, but X is very, very bad in terms of promoting, especially for new projects, 'cause the creator system, uh, creator hi-hierarchy in X is actually very structured. If you don't know people and don't know famous people and don't know like, uh, you know, people with a lot of followers, you basically cannot. It will be very hard for you to build up, build up a lot of, you know, accounts to gain or farm the traffic, uh, X. And, but Threads is a different way. It's a different situation. And also like, uh, typically, as many other AI applications, we start to do Instagram promotion as well, and Instagram, Instagram is very effective channel as well. I love using X, but I have to say X sucks in terms of doing promotion. If you want to raise money from VCs, X is a channel that you cannot avoid. But, uh, if you just want to organically acquire users, X is hell difficult for new projects. But doing campaign, it's okay. Like we also like release our new campaigns and videos on X. But, uh, I would strongly advise people to explore, uh, Threads and, uh, other channels.

   17. AGAakash Gupta16:24

       So I think there's a really important insight here for every AI product builder. Underserved market, Taiwan and Hai- Hong Kong. Underserved or under, uh, tapped social media network, Threads. And then unique way to conquer that social media network, army of interns. So that's a genius strategy that you have come up with. Now we understand how you hit $10M ARR in 60 days. We promised people a tactical masterclass on context engineering, so what I want to do is, I actually found you on X. So we were just talking about how X isn't that great, but you guys are still-

   18. XWXiankun Wu17:01

       [laughs]

   19. AGAakash Gupta17:01

       ... killing it on X. This video went really viral on

8. 17:06 – 20:10

   Visual Context Engineering

   1. AGAakash Gupta17:06

      X. While this video is playing, can you explain to us what is visual context engineering?

   2. XWXiankun Wu17:12

      Yes. Also, like I'm trying to be very frank, it's more like a marketing term than a technical term. I was discussing with my CTO, trying to elaborate our product and our technology. Uh, my CTO said that, uh, probably we need to invent some new terms, even though that I hate inventing new terms. But, uh, instead of like spending a lot of time, uh, spending a lot of words to explain what we are doing, we need to like a very short term of things like to explain what we are actually doing. So for visual context engineering, as I ex-explained before, it's actually a way to help you to express your intent in a different approaches. Regularly, people thought that, uh, only, you know, you need to only prompt in, you know, languages or words. But we try to give you different ways, just like you can doing some like squares or graphs or sketches so that AI will know the special relation, spatial relationship-ships between different objects and documents, so that you can give a precise control of what you really want, uh, within the, uh, canvas or whiteboard. AI will understand you in a more effective way or can save you time, 'cause not everybody are very good at describing what you really want. Some people will prefer to use some other ways to describe. And also that the visual, uh, context engineering also want to demonstrate that we are actually give you a two-dimensional space that you can store your information and documents and files into this two-dimensional ChatGPT, or people call it like a, a, a whiteboard GPT or whatever, so that you can utilize orUh, 'cause in ChatGPT will be very hard to... 'cause you upload something and you give some prompts, and the files upload into the chat box, you cannot see it. If you chat too much, if you want to find the documents and, uh, reuse that documents, it will be very hard to use. But within this two-dimensional chat space, you can easily reuse, you can easily select multiple files, you can easily integrate the results or generation into the, uh, kind of like a library easily so that you can make a loop that create, utilize, and put the creation back to the library and reuse them in a more convenient way. So for context engine-- uh, for visual context engineering, in a word is, one, give you more ways or approaches to express your intents. Second, give you a, a graphic system or graphic, kind of like similar to Graphic operating system to easily and conveniently to utilize your files and generations.

   3. AGAakash Gupta20:05

      Love this. So can we explain this to me like I was just five year old?

9. 20:10 – 22:12

   The Mom Analogy

   1. AGAakash Gupta20:10

      Why does context matter so much?

   2. XWXiankun Wu20:13

      Yeah. Just like I will treat you literally like you are five, 'cause I just visit a friend of mine and, uh, his child is literally five years old. And I was trying to tell him, like, what is, like, context engineering, what actually I am doing. So imagine that, uh, you, you are, you want your mom to cook something for you, and you definitely want something delicious and cater to your, uh, purpose. Like if you want join some sports team in your home or in your, in your school, you want to join a basketball team, you want to grow stronger, you want grow taller, and of course you, you have like your food preferences. So context engineering is like your mom knows you very much, knows your food preferences, knows your purpose of eating foods so that it, your mom knows that what kind of, uh, materials should she buy and what's the way of cooking will definitely make you feel happy of eating and eating more to make you s- become stronger to get you into the basketball team of your school. So that's kind of like a context engineering. Your mom knows you, give you better food, you grow stronger and better relationship with your mom, and your mom feel happy cooking for you and spending more time buying good stuff and materials and cooking food for you again, again, again. It's a positive loop.

   3. AGAakash Gupta21:40

      All right. You heard it from a $10 million founder, guys. Context is making AI your mom. [both laughing] Amazing. So now I want to get a little bit more technical here. I'm gonna pull up some of the, uh, work I've been doing because I do a lot of teaching, right? Teaching for AI PMs, PMs, and a lot of them, they need to get up on context engineering at a more technical level. So let's go ahead and show you guys what is the core of context engineering,

10. 22:12 – 26:26

    RAG vs Fine-Tuning vs Prompt Engineering

    1. AGAakash Gupta22:12

       right? Everything is context engineering. Really, if you think about it, context engineering involves prompt engineering, RAG, state, and memory, which are all a little bit overse- overlapping. And then there's this other element of structured outputs, which somewhat goes into context engineering. Would you say this is an accurate picture of what context engineering is as a whole?

    2. XWXiankun Wu22:34

       Yeah. The, the title here is very interesting, like everything is context engineering. Literally, yes.

    3. AGAakash Gupta22:39

       And RAG and fine-tuning and prompt engineering, people often get confused about these. The way I try to think about it is that RAG is to give your AI access to really up-to-date information. Like if you want to have a knowledge base or enterprise search or something, you're gonna be using RAG. Prompt engineering works for almost everything. It's about giving really structured examples and creating a really nice prompt that maybe tells it chain of thought or whatever it might be. And then fine-tuning is about actually fine-tuning the weights of the base LLM, where you give it hundreds or thousands of examples and it changes them. And for whatever AI product you use, you might need to use some of those tools or a combination, right?

    4. XWXiankun Wu23:27

       Yes.

    5. AGAakash Gupta23:27

       And what do you guys use at Kuse? How does it work on the back end?

    6. XWXiankun Wu23:30

       Uh, we have a very, like a give you a very simple example that, uh, in regular chat bot, you upload something that, uh, in real time that RAG system will analyze and break them into small pieces or whatever, and, uh, so that you can retrieve from the results of a RAG system. But for us, 'cause we prioritize the file management system so much, so you actually probably upload a lot of files within the folder. It's actually can be a async process. We process the file even though y- you are away from the system. If the next time you want use, reuse those information or documents in the folder, we have processed that before even you want use those documents. So one very simple example or u- cases here is we are kind of like prepare everything you have used on the table so that you can cook faster than a regular chat bot. 'Cause every time if you use a regular chat bot to try to order something from your, like, Whole Foods app or whatever, but if you use something in Kuse, we will pre-prepared everything on the table so that you can cook with all the materials prepared for you. So that's kind of like a very special part for us.

    7. AGAakash Gupta24:53

       Fascinating. So a lot of people, they're building AI products who listen to this podcast. And so the way I try to explain it to them is thatYour product, you're taking some user input, and then you're using a model, and you're having this product feature. To create a good user output, you need to shape that model's output. You can't just core use ChatGPT, otherwise you're not gonna hit $10M ARR. You need to have some secret sauce. So you're trying to address hallucinations and generic answers, you're trying to address lack of domain expertise, and you're trying to address inconsistent brand voice. These techniques, you need to choose and have a point of view on which techniques to use so that you can build the right product.

    8. XWXiankun Wu25:33

       Yeah. I think there is a, like, very interesting part here. We just, like, launched a program to improve our internal management of our product development. I think that in the AI era, one very im- different thing is, like, you sometimes need to, like, get the results from model's response before you try to productize the AI feature. 'Cause previously we try, okay, uh, we, we find out a feature, we find out a user need, and we write product requirement documents, and we design the product, and we give the PRD to our engineering team so that engineering team can make that feature happen. And we give the feature and we go live, and we promote the feature so that people can use it. But actually, sometimes it's not that, you know, effective, 'cause what really matters is actually the model's

11. 26:26 – 31:43

    MVO Before MVP

    1. XWXiankun Wu26:26

       response. So in our internal management, we call it, like, MVO. People, like, normally use, like, MVP for, like, a minimal viable product, right? But, uh, within our team, we sometimes use, like, MVO, which means, like, minimal viable outputs. It is not important at all if you don't have desired outputs. You don't really need to spend any time to productize the AI feature unless you get the correct or, uh, somehow, like, comparatively stabilize the model's response with, like, a bunch of, like, like, approaches mentioned here. So in the AI era, I think that, uh, before you have the minimal viable product, you should have a, you should have, like, a minimal viable output first before you go the productization part.

    2. AGAakash Gupta27:16

       This is a really important point, guys. So when you're building an AI feature, you're gonna do AI engineering on the output. You're gonna build a MVO before you move into the MVP.

    3. XWXiankun Wu27:26

       That's, yeah.

    4. AGAakash Gupta27:27

       Amazing insight. So when it comes to these techniques, RAG, fine-tuning, prompt engineering, you're in there every day actually building a product. What have you learned about these tools and techniques that's really helpful in creating good context engineering?

    5. XWXiankun Wu27:42

       For us, to be very frank, we don't, like, use much of, like, a fine-tuning. Uh, it's actually pretty heavy for us, and we use a lot of RAG system, as I mentioned before. Uh, since it is n- for us, it's, it is not heavily a technical problem. It's actually a very product, product problem. We prioritize the file processing a lot compared to other chatbots, and we emphasize the document processing, even some OCR technology much more than, like, a ChatGPT or other as a, a regular chatbots. I wouldn't say, like, fine-tuning is a very, very heavy part in our product, but RAG is definitely, definitely a very heavy part in, in, in Kuse. And for pro-prompt engineering, the goal of our product is, like, is, you know, prevent people from heavily learning how to prompt. It's actually a very heavy and very unpleasant experience for most of our users 'cause people always like, "Probably need, you need to buy a course to learn how to prompt," right? It is actually makes-

    6. AGAakash Gupta28:48

       Yeah

    7. XWXiankun Wu28:48

       ... no sense. The, the goal of our product is, like, you don't really need to prompt that much. We build up a vehicle, we build up every environment. The AI should know the goals of your project, the pro-progress of your projects, your status of your projects without, you know, it prompting too much about giving a context to your AI. It should, like, be a living work OS. The AI will automatically know we really want to build the progress of everything, and you-- Just like, as I mentioned very, at the very beginning of this chat, this conversation, that imagine that you have a, you hire the new colleague. You need to give them a lot of backgrounds and context. But if you have a colleague that has been working with you for, like, 10 years, probably you need to just talk to that person briefly on WhatsApp, and that person will know what you really want and will deliver for you. That's our goal. Yeah. We, we don't emphasize people. We don't emphasize that part, uh, you need to write really good prompt. Our product, the mission of our product is prevent you from writing complicated and complicated prompts.

    8. AGAakash Gupta29:55

       Amazing. Can we see it in action? Can you show us how a PM should be bringing context into Kuse the right way?

    9. XWXiankun Wu30:02

       Definitely. Uh, as you can see on the screen, there are three steps. It's actually very simple. We try to break every workflow into three steps. One, drop files onto the canvas. The canvas is actually a, a folder that you can store your information and documents, and select contents and ask anything. Let me... And in the end, you can get amazing results. I can give you a simply, simply give you a brief use case so that you can have a sense of how Kuse works. I'm here trying to upload four documents here, like, uh, three PDF and one graph. So it is all about, uh, Kuse's product PRD and, uh, everything about product design. So here I circle the product documents and give prompts like, uh, "I want, I want to, uh, create a PRD about, uh, the AI sheets feature."We talked before. Can you make one for me? And here's the other tools, like, uh, you can use the image studio, webpage generator, formatter, exam paper, source only. Source only means that, uh, the generation results only based on the information or documents you uploaded. So-

    10. AGAakash Gupta31:32

        Oh, nice

    11. XWXiankun Wu31:34

        ... you can, you can choose different models here like GPT-5, Claude, Gemini, and also my beloved DeepSeek R1.

    12. AGAakash Gupta31:42

        Mm-hmm.

12. 31:43 – 32:48

    Ad Start

    1. AGAakash Gupta31:43

       You know that feeling when you try to prototype something with AI and it spits out something completely generic? Then you spend hours tweaking colors, fonts, copy, and features just to make it feel like your actual product? Here's the problem: most AI app builders aren't built for product teams. They're built for those starting from scratch. But product teams aren't building from zero. You have an existing product, real customers, design guidelines, a backlog full of ideas you need to explore and validate fast. That's what Reforge Build does, AI prototyping that starts from your product. Add your customer feedback, strategy docs, and product features as context. Create reusable templates using your product design. Explore multiple variants side by side. Collaborate with your team in one place. Reforge Build generates prototypes that reflect your real pricing tiers, real features, real customer language, not generic placeholder. Stop fighting tools built for founders. Start prototyping like a product team. Reforge Build, AI prototyping built for product teams. Try it free at reforge.com/aakash. That's R-E-F-O-R-G-E.com/A-A-K-A-S-H, and use the code BUILD for one month free of premium.

13. 32:48 – 44:43

    Demo: Creating PRD in Kuse

    1. AGAakash Gupta32:48

       So as you were saying earlier, you don't need to do much prompt engineering. It's a very simple prompt.

    2. XWXiankun Wu32:53

       Yes, 'cause you have provided a lot of context to Kuse and, uh, add to canvas. Here is the final results. You can upload, download the markdown documents here in different format. And, uh, 'cause we just released the Ku- Kuse 2.0, it will have like a more advanced edit and, uh, AI features and whatever in Kuse 2.0. Remember, this is the Kuse 1.5, and if you wanna try more advanced and more complete features, you can start to apply your code and try Kuse 2.0.

    3. AGAakash Gupta33:37

       So the prompt isn't the most important thing. The context is the most important thing. If I'm a PM and I'm about to write a PRD and I want Kuse to write a PRD with this simple prompt, what context should I be bringing into Kuse to make sure it can write a good PRD?

    4. XWXiankun Wu33:52

       Exactly, very simple. Like we are trying to, as I mentioned before, like I use a knowledge of a kid talking to his mom. We want to make the experience of talking to the, you know, using Kuse as talking to your most capable colleague. Think about that you are talking to, if you are working in Google or Meta, that you are talking to the CPO of Google or Meta or whatever, like the most capable colleagues that you are working with. What you really want is like a really let that person know the backgrounds of the projects, the progress of the projects, and, uh, the sometimes the details of the projects, the problems you are not decided to, like the, the technology framework you are hesitate to decide, the questions you haven't really-- the problems and questions you haven't figured out in the documents and the, the discussion, even the meeting notes you had with the other colleagues. You can put them together to present this context to your most capable colleague if that person is the PM. So I would say the prompt is not the fixed one, 'cause if you think about that, talking to the most capable colleagues of your, in your team, Kuse is trying to make the experience of using Kuse that kind of like experience, just like working with the people that you feel so capable and trusted.

    5. AGAakash Gupta35:20

       Got it. So think about it as if you're talking to somebody super smart. What would be the context they would need to just ramp up on your specifics that they might not know about? But you don't need to teach them the basics of product management or the basics of PRDs. You can assume they already know that.

    6. XWXiankun Wu35:36

       Yes, exactly. So that's the reason why that if somebody ask me, "Okay, teach me how to write the prompts," I would say like, "I don't, I, I really don't know." 'Cause think about that talking to your super smart colleagues, and you are telling me that I teach you how to, you know, uh, write your emails, communicate with your colleagues. I, I mean that it is also very important to write email to communicate with your, you know, your colleagues. But the most important part is like you really need to know what you are actually doing, what kind of like values or problems you are trying to tackle with, and what's the progress you have made, and what's the, uh, question mark you left there and to be solved by that person. So you really need to know like what you are talking about and what you are actually doing, not really the wording part.

    7. AGAakash Gupta36:28

       So we just showed how you can create a PRD in Kuse. You can also create prototypes. Can you show us what that looks like?

    8. XWXiankun Wu36:36

       Let me choose webpage generator and, uh, making a small simple prototype of a AI sheets feature in a, a canvas based AI productAnd let us see what Kuse will build. It might take a little bit longer time.

    9. AGAakash Gupta37:20

       All right. So tell us what's going on behind the scenes here while it's working.

    10. XWXiankun Wu37:25

        It's actually pretty simple data. We label and, uh, break the PRD here and other information into kind of like a summary, and it will give a summary to Claude. Claude and Claude will do the rest of the work.

    11. AGAakash Gupta37:40

        Claude is really the best coding model for these types of prototypes.

    12. XWXiankun Wu37:43

        Yeah. Actually, you know, pro- uh, product companies sometimes feel so insecure that because people keep questioning them that you are a AI wrapper, that what kind of values you're providing in the space. So they are doing a lot of work to prove that they're contributing a lot more than just using Claude or GPT. But sometimes if it is the useful solution, don't pretend to be, you know, creating a very complicated or comprehensive solution here. We need straightforward and the users only care about if you are really solve their problems or not.

    13. AGAakash Gupta38:22

        So should people be prototyping in Kuse or when should they be using like Claude Code or a Bolt or a Lovable?

    14. XWXiankun Wu38:29

        Yes. So this is a really good question. We still have a lot of things need to be done so that people will feel like it is even more obvious why people should use Kuse rather than Claude Code or Lovable or other coding agents. So the, uh, advantages here is that we are building a complete context of your projects that, uh, the AI will know is much more than... Because like Lovable is trying to-- I love Lovable, but Lovable trying to persuade people that only one sentence of prompts that can make you a viable product and the product can make you money. But it is basic-basically impossible and it is also an iteration pro-process. So we actually build up a space you can, you know, retain and maintain all the, uh, information and PRDs and all the relevant information within one space, and AI will know context of a project more and compounding. It's a kind of like compounding thing. Use the product more, use the AI more, the AI will gets you more and without, you know, prompting too much. In the end, at the very beginning, probably you need to provide pretty much more context than what other coding a-agents probably claim, especially for those agents that, uh, whose target users are those people, uh, who cannot really code. At the very beginning, Kuse might require you to upload or creating more context for AI, but along with the, the process of creating your product, uh, Kuse will in the end know more and create better results as it gets. And the second part is like, uh, uh, I think that for Claude Code and Codex, we're just targeting different group of users. We're still trying to help those who cannot really code that much. I think professional coders or engineers will still choose to use Claude Code or Cursor because their workflow is quite different from non-engineering people or non-tech people.

    15. AGAakash Gupta40:37

        Yep. But this could totally replace a Lovable or a Bolt or a Magic Patterns because of the power of the canvas.

    16. XWXiankun Wu40:43

        I would say not the canvas, because in the new version of Kuse 2.0, we kind of like make the canvas backend. If you want use the magic pen, the canvas will show up and pop up. If you don't use the magic pen, the canvas will just be a hidden, uh, status. But the really powerful point is the compounding power of the context, your data, your information. And we want to really create a, a better environment so that you can maintain all your information and your data, so that your creating process will be easier and easier. Even in terms of like creating a image or creating a video in the future, because I know the AI knows your project so much so that you don't need to brainstorm that much. Like you don't need to talk to AI that much because the AI knows, so that the AI can create a better AI images or videos for you as well. Not only for, you know, web, web, web pages or documents.

    17. AGAakash Gupta41:46

        Got it.

    18. XWXiankun Wu41:46

        I know that, uh, Lovable, Replit, and Bolt, they are building a more complete of context system. But I g- I guess that I'm not saying that, uh, they don't know that the importance of context. But I think that, uh, our design of the system is more suitable and, uh, general for those people who don't know how to code and embed it into their workflow, daily workflows, and provide even more all-in-one, uh, experience, not only for making prototypes or, or like web pages. Especially for those people like, I will give you another example. Like product managers can be a very important user groups of us, but there are a bunch of other very interesting group of people are using our product to make web pages, uh, like admins, like HRs. There are a bunch of companies HRs are using, regularly using our product to make announcements to their company. Because previously if, if they want to like announce something, like they write something in their, you know, internal communication apps, but, uh, sometimes it is not that good to use. Or they can share a file of Google Sheets so that the information or relevant links can be listed in the Google Sheets. But by using our products, they can put all the links or relevant information or just, just like if you want to organize your company to watch some movies and there are a bunch of Luma links, and you can allow your users or you want your company's people to choose that which movies you want to see, you want to watch. And, uh, the HR and admin people just put all the Luma link and the movie's poster into, onto the canvas and then select all of them and say that, "Make a pa- web page for me so that I can share the web page to all the people within the company, so that they can choose which movie they want to watch by clicking the Luma link in the web page." So that's not a typical like a prototype use case, but are regularly used by our users.In their cases

    19. AGAakash Gupta44:01

        Very cool.

    20. XWXiankun Wu44:01

        Yes. So I made someone here. So as you can see, uh, the results here can be put on the canvas along with all the initial context documents here. If you want to reuse the context, you need-- you-- all you need is select the final results and the previous documents or information and give another run to prompts or whatever. Uh, but, uh, it is kind of like a very different experience than the chat, streaming chat experience.

14. 44:43 – 56:12

    Advice for AI Founders

    1. AGAakash Gupta44:43

       So now people have seen the product, they've heard a little bit about the story. What is the advice you would give yourself if you could time travel back to the beginning of 2024? In other words, what is the advice you would give aspiring AI founders who wanna reach $10M ARR?

    2. XWXiankun Wu45:00

       This is a very interesting question. My tech will be very different 'cause very, uh, as you know, like at the very beginning, I was not planning to raise funds from VCs, and I was very patient. I just want to build something that I don't feel that if my company got killed by OpenAI's updates, it will be the end of my world. [chuckles] I was just like trying to build something and using the technology and try to figure out something that eventually can work. So my advice would be don't get, you know, too terrified or don't be afraid of like those amazing updates of AI progress. 'Cause I still remember in the early 2024, people was like, "Okay, in two years there are gonna be AGI. Like, uh, everything I do will be meaningless. Like, uh, all the things that I have done will be killed by OpenAI, and I don't know what to do." But I would advise you to focus on, you know, communicating with your users and don't feel the feeling, don't put the feelings of lose aversion too much, uh, along the way 'cause there are not, there are like bunch of things that you cannot control. If you cannot control those fac- factors, and actually you are willing to see the AI technology progress 'cause you wish that technology can in the end benefit the entire human beings. So don't feel the lose aversion, just focus on your products and your users. And as a, as, as we experienced that initially that we wanna build a design agent, but our users just use the canvas to upload, [chuckles] upload fo- uh, uh, documents and files so that we follow our users said that, "Okay, we hear you, we listen to you, and we make a better product for processing all your information." Today that we figure out kind of like processing documents and, and PDFs is not a, a small opportunity, but it is actually can be very big. If at the very beginning we say that, "Okay, chatting with files is, will be like absolutely, you know, covered by OpenAI or whatever," we'll be so formal and change our direction every day. So just stick what you want to really build and, uh, don't think about like going to IPO and raising funds too much. And I, I sometimes like in Chinese, like, uh, I sometimes talk to people and said that we're building a company that, uh, [in Chinese] xiang si er sheng. What does it mean, like [in Chinese] xiang si er sheng? It means like we're building this for dying. If you feel like what-- I, I, I want try to find a way to explain that what is [in Chinese] xiang si er sheng. I know that in the end, AI will take over the world, not really in a bad way, but, uh, most of people's job will be ta- you know, will be like replaced by AI. Actually, act- actually, I think that I'm not building a productivity tool, but I'm building a playground. In the end, I feel peop-- I want build up a, build up amusement park so people can still feel and enjoying the pleasure of, of working. Maybe in, especially in East Asian, people are so obsessed with working. If someday that, uh, AI take, like, take all of their work, they will be so sad, and some of them might not even have the willing to, to, to continue living. So actually, I'm building a playground of allowing people to pretend to work and still feel the fulfillment, satisfaction of delivering some values and contributing to the world. And it is actually, we are making it more productive, but in the end, it would destroy the productivity. We are not improve the efficiency. We are going to the different opposite direction of efficiency. We're going to pure pleasure and ent- entertainment in the end. So I don't know if I can go IPO with a productivity tool. I would say that I care about people's feeling in 2020, 2030 or 2035 or 2040. I wish they have the bravery or, uh, reasons to still live even though the AI will take over most of the jobs that normal people can do.

    3. AGAakash Gupta49:28

       So poetic. So I wanna just highlight a couple threads for people there. Don't feel FOMO around, well, OpenAI, are they gonna just obliviate us? Follow your users. Don't assume that your first idea is the best idea, but what are they actually using the product for? And then think about the future, like where AI is going and how you're gonna fit into that picture. These are some really, really important lessons. Now, a lot of people, they would be scared of the space you're in. You know, there's Miro. Miro's worth $17 billion. You could consider them a competitor. They're creating a lot of features just like you. How do you compete against these well-funded competitors?

    4. XWXiankun Wu50:07

       Yeah. If you see the Kuse 2.0 launch video, you will see that, as I mentioned before, that, uh, the canvas and Weibo part will be the secondary user interfaceFor us, the essential... Actually, people are c- start to compare us to AI Dropbox for the new-

    5. AGAakash Gupta50:26

       Yes

    6. XWXiankun Wu50:26

       ... [chuckles] for the new Kuse 2.0. So I think it's a dema- Uh, so there is, like, several points I want to address for a bit. First, I don't pay too much attention on the competition. As I mentioned, that, uh, I think all productivity tools might die in the, in the future. So it is pretty meaningless to you release this feature, they release that feature, so I need to catch up. Just build what really, you really want to build. So that's, uh, the first point I wanna cover. The second point is you can... If you, like, really build something, you can see the dyna- You know it's a very dynamic process of building a product. Essentially, people thought that we were competing with Miro, but right now people say that you are competing with, uh, Dropbox or Box. In China, there are a bunch of other, uh, products. Maybe, like, with, uh, in, like, three months or six months, with a deeper understanding of the AI or user's needs, we might be a different position or, uh, give people a different user experience. So just to focus on your, the value proposition and your users. Don't think too much about your competitors. Third, I think that, uh, it's... You need to accept the data, the probability of a startup becoming suc- like, typically successful is very rare. It's just the low. It's okay to fail. You... People might, like, if you, like, talk to a VC, like, will ask a question that why can you beat Miro, or why can you beat Dropbox? My answer would be like, uh, the probability of me beating Miro or Dropbox might be very low. That's the, the true instance that I'm in. But, uh, like, if you want try to find a formula of guaranteed success, I don't know the formula. I don't know if you know. If you know, like, pre- please tell me. But I don't know the-

    7. AGAakash Gupta52:12

       [laughs]

    8. XWXiankun Wu52:12

       ... guaranteed formula. What we can do is just to focus on the user needs and our vision and our mission, and it is a very dynamic process. And a lot of competitors just die without you interfering, without you attacking them. There are a lot of new competitors you don't even notice right now. Actually, in the end, they will win. A- and you are very small right now. You don't have much time sp- uh, s- you, you cannot spend a lot of time in, you know, doing this kind of work. You need to allocate your time and energy and, uh, ti- a- and, and, and, and your attention to the right place.

    9. AGAakash Gupta52:49

       Yes. As a founder, the battle is time. It's focus. It's mental energy. Put it into the right places. So we talked a lot about being a founder. We talked a lot about Kuse. There's also a lot of product builders listening. You know, they might be just in some big company, and they clicked on the episode to learn about visual context engineering. That's where I wanna end. What would be your roadmap if you knew nothing about context engineering and you wanted to ramp up to context engineering? What is the 90-day roadmap to learning context engineering?

    10. XWXiankun Wu53:22

        Actually, I want to talk about a bit a- about the company's structure. I think that might be even more valuable. I'm sorry, like, I'm not answering your question, but I think this part might be even more valuable for... 'Cause context, uh, visual context engineering might be, you know, the product part of our company, but, uh, what really builds up the culture and the essence of this company is our unique structure and the mindset of building this company. As you know that we don't raise money from VCs, but you didn't ask why. Uh, where the money come from? And, uh, we have a very special story here, and I think that might be a very interesting story for your listeners and our listeners. 'Cause every day, like, people claim that the... Get some, like, uh, technology breakthroughs and fantastic product design. I guess there, there are a few companies tell people that, uh, don't raise money from VCs, and try to make your own money. For us that, uh, as you know that previously I co-founded another company and got into YC and raised a lot of money. I know the capital markets. Later on, me and, uh, several partners, we actually tr- create a trading company. And our trading company, in a very right timing, the, the trading worked out, and we got our money, and we decide use the money we got from the trading company to fund Kuse. So at the very beginning, we said that, "Okay, we got our, our own money." And, uh, even though people will say that you can work hand in hand with VCs, and you can at the same time building great product while chasing money from VCs. That's true, but I think that, uh, entrepreneurship is a game of focus, being focused. And, uh, if you get a chance to spend a little bit more time forgetting raising money from VCs and being a part of that, you know, raising money game, take a little bit more time. Feel it. It probably give you a very fresh feeling of doing a startup. 'Cause some people will feel like the typical process of having or running a startup is like, I need to raise money first, then I build up some product, and I raise money from another VC. Uh, the process, in the process, you f- you sometimes feel like you spend too much time chasing money from VCs and modify your strategy too much for adapting the needs of VCs. If you got chance, if you got a little bit money, just to be patient and try to focus on building your product for a while. It's kind of a meditation of entrepreneurship. It give you a very special feelings of running a startup. I think it worth it.

    11. AGAakash Gupta56:01

        What a way to end it. XK, this episode dropped way more wisdom than I would've expected coming in, even though I had high expectations. Thank you so much for your time.

    12. XWXiankun Wu56:11

        Thank

15. 56:12 – 57:08

    Outro

    1. XWXiankun Wu56:12

       you so much.

    2. AGAakash Gupta56:12

       Context engineering is the future of productivity. Kuse proved it, $10M ARR in 60 days. Now you guys have all the inside secrets of how they built Kuse. Go use this knowledge to go build your own products, to use Kuse in your workflows, and I'll see you in the next episode. Bye. So if you wanna learn more about how to shift to this way of working, check out our full conversation on Apple or Spotify Podcasts. And if you want the actual documents that we showed, the tools and frameworks and public links, be sure to check out my newsletter post with all of the details. Finally, thank you so much for watching. It would really mean a lot if you could make sure you are subscribed on YouTube, following on Apple or Spotify Podcasts, and leave us a review on those platforms. That really helps grow the podcast and support our work so that we can do bigger and better productions. I'll see you in the next one.

Episode duration: 57:18