Hiring Employable Agents + Startup Battlefield Finals — Howie Liu, HyperAgent

Please give a round of Please give a round of applause to applause to HyperAgent CEO, Gouyi HyperAgent CEO, Gouyi Liu. Liu. OK. Greetings. Who OK. Greetings. Who here is excited about here is excited about here is excited about the end of this the end of this the end of this event? It was an event? It was an event? It was an incredible few incredible few incredible few days, great performances, days, great performances, days, great performances, and by the way, just an and by the way, just an and by the way, just an incredible performance from incredible performance from incredible performance from Gary. And, by the way, I've Gary. And, by the way, I've Gary. And, by the way, I've never been never been never been presented with such a presented with such a presented with such a bass before. I am very, bass before. I am very, bass before. I am very, very inspired right now. I very inspired right now. I very inspired right now. I consider myself consider myself consider myself primarily a person who primarily a person who primarily a person who thinks in terms of products, thinks in terms of products, thinks in terms of products, meaning that no matter how much I meaning that no matter how much I meaning that no matter how much I love the technical side of love the technical side of love the technical side of development, development, development, architecture, or architecture, or architecture, or algorithms, at heart I still algorithms, at heart I still algorithms, at heart I still think in terms of think in terms of think in terms of product forms. So to product forms. So to product forms. So to wrap up wrap up wrap up today, I today, I today, I want to give a brief and want to give a brief and want to give a brief and general overview of general overview of general overview of how I think the how I think the how I think the world of agents will evolve world of agents will evolve world of agents will evolve in the context of in the context of in the context of product forms. I product forms. I product forms. I called it “Hiring called it “Hiring called it “Hiring Efficient Agents Efficient Agents ,” and it’s a reflection on ,” and it’s a reflection on ,” and it’s a reflection on how models are getting how models are getting how models are getting better and better—we better and better—we better and better—we heard from great heard from great heard from great speakers about the latest speakers about the latest speakers about the latest cutting-edge models like cutting-edge models like GLM 5.2 and others, and GLM 5.2 and others, and GLM 5.2 and others, and how to squeeze the how to squeeze the how to squeeze the most most most performance out of them. And it's performance out of them. And it's performance out of them. And it's not about technicalities not about technicalities not about technicalities , but about how to , but about how to , but about how to think about it from a think about it from a product creation perspective. So who am I product creation perspective. So who am I ? I am the founder and ? I am the founder and ? I am the founder and CEO of Airtable, as well as HyperAgent, CEO of Airtable, as well as HyperAgent, CEO of Airtable, as well as HyperAgent, which is a product which is a product which is a product within Airtable. As many within Airtable. As many within Airtable. As many of you know, Airtable of you know, Airtable is an almost ridiculously

is an almost ridiculously is an almost ridiculously horizontal horizontal horizontal platform. When we platform. When we platform. When we started building the started building the started building the company almost 12 company almost 12 company almost 12 years ago, we years ago, we years ago, we got negative got negative got negative feedback from almost feedback from almost feedback from almost every investor: “ every investor: “ You can’t come out You can’t come out You can’t come out with a horizontal with a horizontal with a horizontal platform and platform and platform and try to make try to make try to make something that’s for something that’s for something that’s for everyone and everything.” everyone and everything.” everyone and everything.” Like, you can't Like, you can't Like, you can't solve all these solve all these solve all these use cases use cases use cases for every type of for every type of for every type of application that people application that people application that people want to build. But want to build. But want to build. But I think we ultimately did a I think we ultimately did a I think we ultimately did a good job with good job with good job with many of them and many of them and many of them and helped a large helped a large helped a large community of community of community of developers. Now we developers. Now we developers. Now we apply a very apply a very apply a very similar philosophy to similar philosophy to similar philosophy to agents. And I will tell you exactly how agents. And I will tell you exactly how agents. And I will tell you exactly how we do it. we do it. So when I look So when I look more broadly at the progression of more broadly at the progression of more broadly at the progression of models and the forms of models and the forms of models and the forms of products they products they products they allow allow allow to be created, I see it to be created, I see it to be created, I see it as a kind of spectrum, as a kind of spectrum, as a kind of spectrum, right? So we've right? So we've right? So we've essentially gone from essentially gone from essentially gone from augmentations, where you augmentations, where you augmentations, where you give a model, one give a model, one give a model, one of these LLMs, a of these LLMs, a of these LLMs, a sequence of sequence of sequence of tokens, and it tokens, and it tokens, and it just keeps just keeps just keeps generating new ones until it generating new ones until it generating new ones until it reaches a stop reaches a stop token. And, in general, token. And, in general, token. And, in general, it just it just it just complemented your complemented your complemented your opinion, right? Or opinion, right? Or opinion, right? Or was completing, uh, a was completing, uh, a was completing, uh, a pattern. And, you know, before pattern. And, you know, before pattern. And, you know, before InstructGPT came along, which InstructGPT came along, which InstructGPT came along, which provided a provided a provided a chat format, it was perhaps chat format, it was perhaps powerful, but it had little powerful, but it had little powerful, but it had little narrower application.

narrower application. Once we got to a Once we got to a higher level of higher level of higher level of model intelligence, model intelligence, model intelligence, and with and with and with the innovation of InstructGPT and the innovation of InstructGPT and the innovation of InstructGPT and tuning these tuning these tuning these models for real models for real models for real conversational interaction conversational interaction conversational interaction with users, we with users, we with users, we got chatbots, got chatbots, got chatbots, right? And of course, we right? And of course, we right? And of course, we all know that ChatGPT all know that ChatGPT all know that ChatGPT has become an incredibly has become an incredibly has become an incredibly successful product. successful product. successful product. We are all now very We are all now very We are all now very familiar with this familiar with this familiar with this format of AI. The format of AI. The format of AI. The next level I next level I next level I call an agent, call an agent, call an agent, right? And, you know, I right? And, you know, I right? And, you know, I think the term " think the term " agent" has become too agent" has become too agent" has become too overloaded. Uh, overloaded. Uh, overloaded. Uh, we've probably we've probably we've probably talked about this a lot talked about this a lot talked about this a lot over the last over the last over the last few days, but my few days, but my few days, but my definition of an agent definition of an agent definition of an agent actually coincides actually coincides actually coincides with Anthropic's definition, with Anthropic's definition, with Anthropic's definition, which is different which is different which is different from workflows. from workflows. from workflows. An agent is really a An agent is really a An agent is really a model that model that model that recursively recursively recursively refers to refers to refers to itself, right? And it has, itself, right? And it has, itself, right? And it has, you know, an open you know, an open you know, an open set of tools set of tools set of tools that it can invoke, that it can invoke, that it can invoke, reasoning steps reasoning steps reasoning steps that it can take, and that it can take, and that it can take, and decisions that it can decisions that it can decisions that it can make. So, he is not make. So, he is not make. So, he is not limited to a linear limited to a linear limited to a linear workflow, but workflow, but workflow, but can make an can make an can make an unlimited number of unlimited number of unlimited number of choices. Uh, still choices. Uh, still choices. Uh, still , you know, I , you know, I , you know, I like to think like to think like to think of agents as being of agents as being of agents as being a little different from the a little different from the a little different from the next level of next level of next level of agents, which I'm agents, which I'm agents, which I'm using the using the using the term Claw for here, because term Claw for here, because term Claw for here, because Open Claw obviously Open Claw obviously Open Claw obviously popularized that popularized that popularized that concept. But, you know, concept. But, you know, concept. But, you know, I see a market shift away I see a market shift away I see a market shift away from agents that from agents that from agents that were still largely were still largely were still largely reactive, uh, to reactive, uh, to reactive, uh, to user input. You user input. You user input. You forced the agent to go forced the agent to go forced the agent to go and complete the and complete the and complete the task. He could task. He could task. He could spend 20 minutes, spend 20 minutes, spend 20 minutes, maybe up to an hour, on a maybe up to an hour, on a maybe up to an hour, on a task, and task, and task, and then he would finish then he would finish then he would finish the job. Claw, uh, as I the job. Claw, uh, as I the job. Claw, uh, as I call them, are, you know, call them, are, you know, call them, are, you know, agents that do agents that do agents that do the job for a much longer period of time, the job for a much longer period of time, the job for a much longer period of time, right? They might right? They might right? They might even have a mechanism, even have a mechanism, even have a mechanism, like the like the like the heartbeat mechanism in Open Claw, heartbeat mechanism in Open Claw, uh, and now in other uh, and now in other uh, and now in other products like Hyper products like Hyper products like Hyper Agent, that allows them

Agent, that allows them Agent, that allows them to wake up on their to wake up on their to wake up on their own and almost own and almost own and almost display this display this display this constant activity. constant activity. constant activity. That is, what happens under the hood is that That is, what happens under the hood is that each awakening each awakening essentially leads to essentially leads to essentially leads to many different many different many different steps, so the agent steps, so the agent steps, so the agent loops loops loops back to itself. back to itself. back to itself. It executes It executes It executes tool calls. Each tool calls. Each tool calls. Each response after the response after the response after the tool call tool call tool call re-accesses re-accesses re-accesses the LLM until a the LLM until a the LLM until a breakpoint is reached breakpoint is reached breakpoint is reached . But then this . But then this . But then this breakpoint is breakpoint is breakpoint is reactivated or reactivated or reactivated or restarted using a restarted using a “heartbeat” mechanism where you “heartbeat” mechanism where you can wake up the agent again and again can wake up the agent again and again can wake up the agent again and again , , , right? So this idea of ​​a right? So this idea of ​​a right? So this idea of ​​a constantly operating constantly operating constantly operating agent that agent that agent that actually actually actually moves relentlessly towards its moves relentlessly towards its moves relentlessly towards its goal and creatively goal and creatively goal and creatively finds ways to finds ways to finds ways to unblock itself unblock itself unblock itself where previous where previous where previous agents got stuck. I agents got stuck. I agents got stuck. I think this is a very, think this is a very, think this is a very, very important very important very important next leap in next leap in next leap in form factor that we're form factor that we're form factor that we're seeing. And seeing. And seeing. And finally, if finally, if finally, if you think about the models you think about the models you think about the models that are now capable of, that are now capable of, that are now capable of, you know, very, very you know, very, very you know, very, very intelligent intelligent intelligent coordination of complex coordination of complex coordination of complex tasks, right? tasks, right? tasks, right? Personally, I quite Personally, I quite Personally, I quite often use the often use the often use the workflow capabilities workflow capabilities workflow capabilities in Claude Code, where in Claude Code, where in Claude Code, where you can distribute a you can distribute a you can distribute a large amount of work large amount of work large amount of work between a massively between a massively between a massively parallel set of parallel set of parallel set of agents. But this agents. But this agents. But this concept of, you know, an concept of, you know, an concept of, you know, an agent that can agent that can agent that can interact with other interact with other interact with other agents, right?

agents, right? agents, right? Interaction between Interaction between Interaction between agents to agents to agents to perform even more perform even more perform even more complex tasks, complex tasks, complex tasks, right? right? right? Longer-running tasks Longer-running tasks are, in my opinion, the are, in my opinion, the are, in my opinion, the next important next important next important step. All of this is possible step. All of this is possible step. All of this is possible because because because models are getting models are getting models are getting smarter, smarter, smarter, reaching a level where reaching a level where reaching a level where they can they can they can maintain maintain maintain consistency even consistency even consistency even in longer and in longer and in longer and open-ended tasks, open-ended tasks, open-ended tasks, right? If you think back to right? If you think back to right? If you think back to the days of Baby AGI and Auto GPT, the days of Baby AGI and Auto GPT, the days of Baby AGI and Auto GPT, they were pretty interesting they were pretty interesting they were pretty interesting science experiments science experiments , but in the end, while , but in the end, while , but in the end, while they were they were they were fun to watch fun to watch fun to watch for hours, they for hours, they for hours, they inevitably veered off inevitably veered off inevitably veered off course into something course into something course into something not very useful, not very useful, not very useful, right? They right? They right? They would sink into the jungle would sink into the jungle would sink into the jungle and just get stuck. and just get stuck. And I think that models are now smart enough to not only smart enough to not only ensure that ensure that ensure that long-term long-term long-term tasks are completed, but also to tasks are completed, but also to tasks are completed, but also to allow us to allow us to allow us to apply apply apply organizational organizational organizational principles to principles to principles to break down problems into break down problems into break down problems into tasks of varying tasks of varying tasks of varying scales that agents scales that agents scales that agents can delegate to can delegate to can delegate to each other; it's all each other; it's all each other; it's all about managing about managing each agent's context window and each agent's context window and the ability to the ability to the ability to coordinate work coordinate work coordinate work between them. So, this all between them. So, this all between them. So, this all sounds sounds sounds pretty abstract so far. I pretty abstract so far. I pretty abstract so far. I want to make this want to make this want to make this as as as specific as possible and specific as possible and specific as possible and show how we show how we show how we applied design applied design thinking to thinking to thinking to create a create a create a horizontal horizontal horizontal product, simplifying the product, simplifying the product, simplifying the complexity of complexity of complexity of agent work, similar to how agent work, similar to how agent work, similar to how we did it we did it we did it for databases or for databases or for databases or Airtable-based applications Airtable-based applications , in HyperAgent. So this is a , in HyperAgent. So this is a , in HyperAgent. So this is a completely real case study completely real case study completely real case study based on based on based on the experience of the experience of the experience of our real our real our real customers. One of the customers. One of the customers. One of the interesting things we interesting things we interesting things we found out after

found out after found out after launching HyperAgent is launching HyperAgent is launching HyperAgent is that typical that typical that typical HyperAgent users are HyperAgent users are HyperAgent users are not at all who not at all who not at all who you would think, right? you would think, right? you would think, right? We certainly get We certainly get We certainly get our share of purely our share of purely AI-focused software companies, AI-focused software companies, or even AI or even AI companies themselves, but we also get companies themselves, but we also get companies themselves, but we also get approached by other approached by other approached by other interesting businesses that interesting businesses that interesting businesses that are often traditionally are often traditionally are often traditionally offline companies. In offline companies. In offline companies. In this case, we this case, we this case, we show a very show a very show a very realistic realistic realistic scenario: a client scenario: a client scenario: a client is engaged in is engaged in is engaged in landscape design landscape design landscape design and uses and uses and uses agents to conduct agents to conduct agents to conduct almost all almost all almost all stages of the business—from stages of the business—from stages of the business—from finding clients and finding clients and finding clients and proposing projects proposing projects proposing projects to physically to physically to physically renovating the garden or renovating the garden or renovating the garden or yard and yard and yard and managing the entire managing the entire managing the entire process, including process, including process, including costing, costing, costing, presentation, and the presentation, and the presentation, and the work itself. I think work itself. I think work itself. I think this is a very interesting this is a very interesting this is a very interesting discovery I discovery I discovery I came across while came across while came across while working on HyperAgent: working on HyperAgent: working on HyperAgent: the ingenuity of the ingenuity of the ingenuity of agent developers is agent developers is agent developers is not limited to not limited to not limited to AI companies or AI companies or software developers. I was software developers. I was really inspired by really inspired by really inspired by Gary's speech, which was Gary's speech, which was Gary's speech, which was literally a minute literally a minute literally a minute ago. I believe that the ago. I believe that the ago. I believe that the main idea of ​​this main idea of ​​this main idea of ​​this speech is applicable speech is applicable speech is applicable to any industry to any industry to any industry and any sector and any sector . There are a lot of really creative people in this room . There are a lot of really creative people in this room who can go and who can go and transform transform transform their own industry, their own industry, their own industry, whatever it is.

whatever it is. whatever it is. It doesn't have to It doesn't have to It doesn't have to be exclusively a be exclusively a be exclusively a software business. software business. software business. But I believe that But I believe that But I believe that all it takes is all it takes is all it takes is the ability to imagine what the ability to imagine what the ability to imagine what work will look like work will look like work will look like when you become not when you become not when you become not just an AI- just an AI- centric centric centric company, but, as I company, but, as I company, but, as I like to say, a like to say, a like to say, a company that company that relies entirely on a fleet of relies entirely on a fleet of agents; where you agents; where you agents; where you go beyond go beyond go beyond automating automating automating individual narrow individual narrow individual narrow tasks and start tasks and start tasks and start using an entire using an entire using an entire network of agents with network of agents with network of agents with clear rules of clear rules of clear rules of behavior, behavior, behavior, context, and context, and context, and tools for tools for truly useful truly useful truly useful and fully and fully and fully autonomous work. autonomous work. So let's So let's take a look at what it take a look at what it take a look at what it looks like. This is a looks like. This is a looks like. This is a realistic example realistic example realistic example of how this of how this of how this landscape designer landscape designer landscape designer wants to attract wants to attract wants to attract clients, right? clients, right? clients, right? They receive They receive They receive a lot of incoming a lot of incoming a lot of incoming inquiries from inquiries from inquiries from interested interested interested customers. Someone customers. Someone customers. Someone sends a request: " sends a request: " Hi, here's my Hi, here's my Hi, here's my current current current backyard, it's a little backyard, it's a little backyard, it's a little neglected." "Here is my neglected." "Here is my neglected." "Here is my information, can information, can information, can you provide an estimate and you provide an estimate and you provide an estimate and suggest your suggest your suggest your design option?" design option?" design option?" This automatically This automatically This automatically goes to goes to goes to one agent one agent —the sorting agent —the sorting agent , right? So, this , right? So, this , right? So, this agent actually agent actually agent actually takes the request and takes the request and takes the request and determines the scope determines the scope determines the scope of the project: should we of the project: should we of the project: should we take it on, is take it on, is take it on, is it quality lead, and is it quality lead, and is it quality lead, and is it a real person?

it a real person? He does some He does some research, but research, but research, but ultimately hands ultimately hands ultimately hands the job off to another agent—an the job off to another agent—an appraiser agent— appraiser agent— appraiser agent— to actually to actually to actually look at the physical look at the physical look at the physical landscape of landscape of landscape of this person's backyard and come this person's backyard and come this person's backyard and come up with a plan for how to up with a plan for how to up with a plan for how to redevelop it. redevelop it. So, here the assessing agent takes on the task of assessing agent takes on the task of assessing agent takes on the task of reviewing the reviewing the reviewing the original original original images or even images or even images or even videos provided by videos provided by videos provided by the applicant. Our Hyper-agent the applicant. Our Hyper-agent the applicant. Our Hyper-agent works with a works with a works with a fully functional fully functional virtual machine sandbox. virtual machine sandbox. So every agent, So every agent, So every agent, in fact every thread in fact every thread in fact every thread with an agent, is capable of with an agent, is capable of with an agent, is capable of writing full-fledged writing full-fledged writing full-fledged code. It can code. It can code. It can manipulate files manipulate files . It can . It can . It can use use use tools like FFmpeg tools like FFmpeg tools like FFmpeg to slice individual to slice individual to slice individual screenshots from screenshots from screenshots from a video file, a video file, a video file, analyze them, analyze them, analyze them, find all the find all the find all the necessary information, necessary information, necessary information, and create a fairly and create a fairly and create a fairly high-quality and high-quality and high-quality and personalized personalized personalized offer, which offer, which offer, which was previously impossible for was previously impossible for was previously impossible for such a small such a small such a small business, right? business, right? If you think about an If you think about an elite elite elite interior designer or a very interior designer or a very interior designer or a very prestigious prestigious prestigious landscape landscape landscape architect, perhaps architect, perhaps architect, perhaps they could they could they could make such a make such a make such a customized customized customized proposal for a proposal for a proposal for a client with a budget of client with a budget of client with a budget of over $5 million over $5 million over $5 million if they if they if they were working for a were working for a were working for a large hotel. But it would be large hotel. But it would be large hotel. But it would be unheard of to unheard of to unheard of to do something do something do something so detailed, so detailed, so detailed, to create actual to create actual to create actual visualizations and a visualizations and a visualizations and a presentation for a presentation for a presentation for a client just for a client just for a client just for a small small backyard landscaping project. But backyard landscaping project. But this person did it this person did it this person did it because they figured out how to because they figured out how to because they figured out how to use use use agents extremely agents extremely agents extremely efficiently and efficiently and efficiently and autonomously to autonomously to autonomously to do work that would do work that would do work that would otherwise be beyond the

otherwise be beyond the otherwise be beyond the reach of humans. So what reach of humans. So what reach of humans. So what happens next: the happens next: the happens next: the appraiser agent appraiser agent appraiser agent created this created this created this offer and offer and offer and presentation. And now a presentation. And now a person needs to intervene again. There's a person needs to intervene again. There's a stage where people stage where people stage where people still have to still have to still have to unlock unlock unlock agents, right? Sarah Goh agents, right? Sarah Goh agents, right? Sarah Goh published a great published a great published a great post a few weeks post a few weeks post a few weeks ago describing where ago describing where value will accumulate in this AI landscape as landscape as landscape as technology technology technology and model intelligence evolve. and model intelligence evolve. and model intelligence evolve. I completely agree with her I completely agree with her I completely agree with her opinion that the opinion that the opinion that the main obstacle main obstacle main obstacle to deploying to deploying to deploying useful agents useful agents useful agents remains the human remains the human remains the human factor. You know, even as factor. You know, even as factor. You know, even as models get models get models get smarter, smarter, smarter, the ability for agents the ability for agents the ability for agents to get help to get help to get help from humans who still from humans who still from humans who still control control control strategies, decisions, and strategies, decisions, and , importantly, block , importantly, block , importantly, block critical, critical, critical, high- high- high- impact actions, like in this impact actions, like in this impact actions, like in this case, case, presenting a presenting a presenting a binding offer to a customer binding offer to a customer binding offer to a customer . Or, . Or, . Or, for example, consent to for example, consent to for example, consent to perform work—this is perform work—this is perform work—this is where a person has to where a person has to where a person has to intervene. Just like a intervene. Just like a intervene. Just like a great manager great manager great manager who gets reports or who gets reports or who gets reports or updates on updates on updates on work from their work from their work from their team to then team to then team to then help them remove help them remove help them remove obstacles. So we obstacles. So we obstacles. So we have Sage going back have Sage going back have Sage going back to the real to the real to the real owner of this owner of this owner of this business and saying, “Hey, business and saying, “Hey, business and saying, “Hey, here’s a new client. I have here’s a new client. I have here’s a new client. I have already analyzed already analyzed already analyzed the project. Here's a great the project. Here's a great the project. Here's a great presentation and presentation and presentation and proposal, I've put together a proposal, I've put together a proposal, I've put together a realistic realistic realistic estimate based estimate based estimate based on all the factors on all the factors on all the factors I've already studied for I've already studied for I've already studied for this particular this particular this particular project, so you can project, so you can project, so you can approve sending it to approve sending it to approve sending it to the client and see the client and see the client and see if they accept it, right?

if they accept it, right? » And you see that » And you see that » And you see that even the even the even the proposal itself—it's proposal itself—it's proposal itself—it's actually a nice actually a nice actually a nice interactive interactive interactive webpage, which again would be webpage, which again would be unthinkable for a unthinkable for a business of this type— business of this type— to send to every to send to every to send to every client, even for client, even for client, even for small assignments small assignments small assignments or jobs like a or jobs like a $10,000 landscaping project, right? $10,000 landscaping project, right? You know, even better, You know, even better, You know, even better, when interacting with when interacting with when interacting with these agents, as these agents, as these agents, as Gary talked about G Brain Gary talked about G Brain Gary talked about G Brain and creating this and creating this and creating this very persistent very persistent very persistent contextual layer contextual layer contextual layer that learns with every that learns with every that learns with every interaction, I totally interaction, I totally interaction, I totally agree. I think agree. I think agree. I think this is a kind of this is a kind of this is a kind of universal universal universal principle and format for principle and format for principle and format for useful agents in the useful agents in the useful agents in the future—they future—they future—they shouldn't be shouldn't be shouldn't be static. They static. They static. They shouldn't just be shouldn't just be shouldn't just be static static static LLM calls or static LLM calls or static LLM calls or static agents who agents who agents who only have a clue and only have a clue and only have a clue and a few fixed a few fixed a few fixed skills. They really skills. They really skills. They really need to learn with need to learn with need to learn with every interaction, every interaction, every interaction, right? And you can't right? And you can't right? And you can't wait for the next wait for the next wait for the next stage of retraining stage of retraining stage of retraining your models, when your models, when your models, when you have a you have a you have a bunch of data or bunch of data or bunch of data or criteria to criteria to criteria to retrain the model. retrain the model. retrain the model. They have to learn in They have to learn in They have to learn in real real real time, right? They time, right? They time, right? They must accumulate must accumulate must accumulate memories and memories and memories and skill updates, take into account skill updates, take into account skill updates, take into account user feedback, user feedback, user feedback, and learn from real-world and learn from real-world and learn from real-world results to results to results to constantly constantly constantly adapt and adapt and adapt and become better, become better, become better, just like humans. So just like humans. So just like humans. So in this case, you in this case, you in this case, you see that in Slack or, in see that in Slack or, in see that in Slack or, in our case, through our case, through our case, through Slack, email Slack, email Slack, email or other or other or other call methods, you can call methods, you can call methods, you can communicate with the communicate with the communicate with the agent like you would with a agent like you would with a agent like you would with a real person and real person and real person and say, "Hey, say, "Hey, say, "Hey, remember that remember that remember that next time." It's next time." It's next time." It's as if you're pointing out, "Here's a as if you're pointing out, "Here's a as if you're pointing out, "Here's a differently differently differently formatted formatted formatted document, or a different document, or a different document, or a different view of the proposal, view of the proposal, view of the proposal, or you missed this or you missed this or you missed this additional additional additional cost lever on this cost lever on this cost lever on this particular project

particular project ." So, you know, ." So, you know, ." So, you know, remember that for remember that for remember that for next time. And I next time. And I next time. And I think it's critical that think it's critical that think it's critical that in this in this in this next generation, next generation, next generation, agents can agents can agents can remember remember remember things like that, right? They things like that, right? They things like that, right? They must be flexible must be flexible must be flexible and evolve like and evolve like and evolve like humans to fully humans to fully humans to fully realize realize realize the potential the potential the potential afforded by the afforded by the afforded by the latest advanced latest advanced latest advanced models for models for models for achieving achieving achieving autonomy. And autonomy. And autonomy. And ultimately, I think what it ultimately, I think what it ultimately, I think what it comes down to is comes down to is that with all these that with all these that with all these agents, each of agents, each of agents, each of which is capable and which is capable and which is capable and even interacting with the even interacting with the even interacting with the others, the human job is others, the human job is others, the human job is increasingly to increasingly to increasingly to remove the remove the remove the obstacles to the obstacles to the agents working effectively. So I agents working effectively. So I compare it to the compare it to the development workflows that have development workflows that have evolved: evolved: evolved: in the beginning, before in the beginning, before in the beginning, before any AI any AI coding, we just coding, we just coding, we just sat in front of sat in front of sat in front of a computer. We humans are a computer. We humans are pretty single-minded pretty single-minded , aren't we? You , aren't we? You , aren't we? You look at one look at one look at one code file, think about a code file, think about a code file, think about a problem, write problem, write problem, write some code, maybe some code, maybe some code, maybe take a break, and take a break, and take a break, and then come back then come back then come back to it. Then we to it. Then we to it. Then we got advanced got advanced got advanced coding with the first coding with the first coding with the first generation of GitHub Copilot generation of GitHub Copilot generation of GitHub Copilot and similar and similar and similar tools, which I tools, which I tools, which I think are just think are just code autocompletion tools.

code autocompletion tools. You type the code, and it You type the code, and it You type the code, and it can automatically can automatically can automatically complete a few complete a few complete a few extra lines. extra lines. extra lines. Then came Then came Then came tools like Cursor tools like Cursor tools like Cursor Composer 1.0, not the Composer 1.0, not the Composer 1.0, not the Composer model itself, but a chat where Composer model itself, but a chat where Composer model itself, but a chat where you could talk to an you could talk to an you could talk to an agent and he would make more agent and he would make more agent and he would make more complex changes. And that complex changes. And that complex changes. And that gave even more gave even more gave even more benefits, didn't it? benefits, didn't it? benefits, didn't it? We're now moving We're now moving We're now moving to a stage where the to a stage where the to a stage where the best best best agent system developers agent system developers agent system developers I know, and many of I know, and many of I know, and many of you sitting in this room, are you sitting in this room, are you sitting in this room, are simply managing an entire simply managing an entire simply managing an entire fleet of agents, fleet of agents, fleet of agents, right? In fact, right? In fact, right? In fact, going to bed without going to bed without going to bed without assigning your assigning your assigning your agents to do agents to do agents to do useful work at night useful work at night useful work at night feels like a feels like a feels like a big loss, as if big loss, as if big loss, as if your team is your team is your team is idle because you haven't idle because you haven't idle because you haven't removed the obstacles removed the obstacles removed the obstacles to their work. to their work. to their work. So the UX of how you So the UX of how you So the UX of how you control these control these control these agents, whether in agents, whether in development workflows or in development workflows or in general tasks general tasks general tasks like this landscaping example like this landscaping example , also needs to change. , also needs to change. So we thought about So we thought about So we thought about it like this: you need it like this: you need it like this: you need this this orchestration control plane where you orchestration control plane where you can can can see at a glance what see at a glance what see at a glance what all all all your agents are working on, what's your agents are working on, what's your agents are working on, what's blocking them, and who's handing off blocking them, and who's handing off blocking them, and who's handing off work to each other. And work to each other. And work to each other. And your job becomes like your job becomes like stepping back, as if you're stepping back, as if you're looking at the entire looking at the entire looking at the entire SimCity landscape, and SimCity landscape, and SimCity landscape, and managing everything that's managing everything that's managing everything that's happening at a happening at a happening at a macro level between macro level between macro level between your various your various your various agents, instead of agents, instead of delving into delving into each individual project each individual project each individual project or task. And I or task. And I or task. And I think this is a very, think this is a very, think this is a very, very exciting very exciting very exciting moment, because we've moment, because we've moment, because we've moved from not just moved from not just moved from not just supplementing or supplementing or supplementing or slightly improving slightly improving slightly improving existing human existing human existing human work, but to a work, but to a work, but to a radical leap radical leap where humans become where humans become where humans become orchestrators of orchestrators of orchestrators of agents that can also agents that can also self-organize . And finding the right

. And finding the right . And finding the right way to engage with way to engage with way to engage with these agents as these agents as these agents as they become they become they become increasingly capable increasingly capable will become not just will become not just will become not just an opportunity for an opportunity for an opportunity for great accomplishments, but, great accomplishments, but, great accomplishments, but, in my opinion, a in my opinion, a survival imperative for every survival imperative for every company and industry. company and industry. company and industry. This will be a way to This will be a way to This will be a way to stay ahead of your stay ahead of your stay ahead of your competitors and competitors and competitors and succeed in succeed in succeed in your industry. And if your industry. And if your industry. And if you don't, you don't, you don't, then, as Jensen said, it's then, as Jensen said, it's then, as Jensen said, it's not artificial intelligence that will take your job not artificial intelligence that will take your job not artificial intelligence that will take your job , it's , it's , it's the person the person the person using artificial using artificial using artificial intelligence who will take it, or it intelligence who will take it, or it intelligence who will take it, or it will take your business. will take your business. will take your business. So I'm really excited about this So I'm really excited about this So I'm really excited about this opportunity, and for opportunity, and for opportunity, and for everyone in this room to be a everyone in this room to be a everyone in this room to be a part of this part of this part of this transformation and transformation and transformation and this bold this bold this bold leap into the leap into the leap into the next form of next form of next form of AI. All of this being said AI. All of this being said AI. All of this being said , we want to , we want to , we want to suggest you suggest you suggest you try HyperAgent. try HyperAgent. try HyperAgent. So, at hyperagent.com/aie you So, at hyperagent.com/aie you So, at hyperagent.com/aie you will receive $1,000 will receive $1,000 will receive $1,000 in in inference credit. They can be inference credit. They can be inference credit. They can be spent on all our spent on all our spent on all our different model different model different model offerings, offerings, offerings, including the latest including the latest including the latest Opus 4.8, Fable 5. You can Opus 4.8, Fable 5. You can Opus 4.8, Fable 5. You can use them use them use them on GLM 5.2, GBD 5.5, etc. And on GLM 5.2, GBD 5.5, etc. And on GLM 5.2, GBD 5.5, etc. And we would be interested we would be interested we would be interested to see what you to see what you to see what you create. With agents create. With agents create. With agents on HyperAgent or elsewhere. on HyperAgent or elsewhere. on HyperAgent or elsewhere. Thank you all for being here. Thank you all for being here.

Thank you for joining Thank you for joining us. us. us. Thank you. Thank you. To wrap up our To wrap up our events, a few months events, a few months events, a few months ago I saw that ago I saw that ago I saw that HyperAgent was launching this HyperAgent was launching this HyperAgent was launching this startup competition. And startup competition. And startup competition. And we at AIE have always wanted to we at AIE have always wanted to we at AIE have always wanted to organize some kind of organize some kind of organize some kind of platform for platform for platform for competitions and contests competitions and contests competitions and contests for people. But we didn't for people. But we didn't for people. But we didn't have any have any have any mechanism by mechanism by mechanism by which we which we which we could could could evaluate it, evaluate it, evaluate it, fund it, or fund it, or fund it, or anything like that. And then anything like that. And then anything like that. And then HyperAgent appeared with HyperAgent appeared with HyperAgent appeared with its 500 its 500 its 500 founders. founders. founders. Yes. Yes. Yes. Of course? What was the Of course? What was the Of course? What was the inspiration for this? inspiration for this? inspiration for this? You know, when I think You know, when I think You know, when I think about the multi-level about the multi-level about the multi-level accumulation of change, we accumulation of change, we accumulation of change, we go, as I go, as I go, as I said, from people said, from people said, from people doing the same doing the same doing the same job with a little bit of job with a little bit of job with a little bit of AI support, to AI support, to AI support, to people running people running people running entire fleets of entire fleets of entire fleets of agents doing agents doing things that were impossible before—like that things that were impossible before—like that landscaper doing landscaper doing landscaper doing business completely differently, business completely differently, business completely differently, to this new to this new to this new economy of companies economy of companies economy of companies built on built on built on agents. So the " agents. So the " founding founding founding five hundred" is a five hundred" is a five hundred" is a bet on the most bet on the most bet on the most advanced part of advanced part of advanced part of this economy, this economy, this economy, right? These are the companies with right? These are the companies with right? These are the companies with the highest the highest the highest use of use of use of agents, who are not only agents, who are not only agents, who are not only changing their field, changing their field, changing their field, but also creating but also creating but also creating entirely new entirely new entirely new rules of the game for rules of the game for rules of the game for doing business in this doing business in this doing business in this new era. And I think the new era. And I think the new era. And I think the effect that we're already effect that we're already effect that we're already seeing from these seeing from these seeing from these companies operating in companies operating in companies operating in this way, as this way, as this way, as Gary said, is just Gary said, is just Gary said, is just amazing. This is significantly amazing. This is significantly amazing. This is significantly more than even more than even more than even early companies that early companies that early companies that relied on relied on software or the software or the internet. So I truly internet. So I truly internet. So I truly believe that this sector of believe that this sector of believe that this sector of the economy will have the economy will have the economy will have explosive growth, explosive growth, explosive growth, much higher than much higher than any other, and this will any other, and this will any other, and this will apply not apply not apply not only to the tech industry or the only to the tech industry or the only to the tech industry or the software industry, but to all software industry, but to all software industry, but to all industries.

industries. Yes. In my Yes. In my London talk, London talk, London talk, I often talk about I often talk about I often talk about how worker agents “ how worker agents “ eat” the rest of the eat” the rest of the eat” the rest of the work, making it work, making it work, making it available to available to available to non-technical teams. non-technical teams. non-technical teams. By the way, my team is By the way, my team is By the way, my team is back there right now, because AI back there right now, because AI back there right now, because AI is working on AirTable, and they is working on AirTable, and they is working on AirTable, and they really want really want really want to meet you and to meet you and to meet you and work with HyperAgent. work with HyperAgent. work with HyperAgent. So, out of 500 founders, So, out of 500 founders, So, out of 500 founders, you pre- you pre- you pre- selected 20. Today selected 20. Today selected 20. Today they competed on the " they competed on the " startup battlefield." startup battlefield." startup battlefield." Three became Three became Three became finalists. Let's finalists. Let's finalists. Let's introduce them. I don't introduce them. I don't introduce them. I don't know if I should know if I should know if I should press this. But press this. But press this. But first, I just first, I just first, I just want to introduce want to introduce want to introduce our esteemed our esteemed our esteemed judges. I think judges. I think judges. I think Gowi, Theo and Gowi, Theo and Gowi, Theo and Joshua will be with us. Thank you from Hajan. Joshua will be with us. Thank you from Hajan. And, well, I think they And, well, I think they 'll judge us. 'll judge us. But we will also But we will also talk about the format talk about the format talk about the format and backstory. So, and backstory. So, and backstory. So, essentially, the essentially, the essentially, the HyperAgent team has kindly HyperAgent team has kindly HyperAgent team has kindly allocated $100,000 allocated $100,000 allocated $100,000 in prizes, and it's a in prizes, and it's a in prizes, and it's a competition for whoever competition for whoever competition for whoever emerges as the most emerges as the most emerges as the most impressive impressive impressive winners. So, 50 winners. So, 50 winners. So, 50 thousand for the winners, 30 thousand for the winners, 30 thousand for the winners, 30 thousand for second place, thousand for second place, thousand for second place, and 20 thousand for third. We and 20 thousand for third. We and 20 thousand for third. We will show a hype video, will show a hype video, will show a hype video, courtesy of courtesy of courtesy of Hyperframe and Hajan. We have Hyperframe and Hajan. We have Hyperframe and Hajan. We have speeches from speeches from speeches from the founders and a the founders and a the founders and a Q& Q& Q& A session with the A session with the A session with the judges scheduled, and then we'll judges scheduled, and then we'll judges scheduled, and then we'll move on to the move on to the move on to the next next next presentation. So, presentation. So, presentation. So, three quick three quick three quick presentations one after the presentations one after the presentations one after the other. We decided that we

other. We decided that we other. We decided that we wanted to implement wanted to implement wanted to implement an evaluation to an evaluation to an evaluation to understand what exactly is understand what exactly is understand what exactly is impressive about impressive about impressive about using this using this using this technology and why it is technology and why it is technology and why it is important. This is not important. This is not important. This is not so much an investment so much an investment so much an investment presentation, you know, presentation, you know, presentation, you know, not a pitch to venture not a pitch to venture not a pitch to venture capitalists, but rather capitalists, but rather capitalists, but rather an answer to the an answer to the an answer to the question: why question: why question: why should you be interested? Why is should you be interested? Why is should you be interested? Why is this an interesting this an interesting this an interesting application of application of application of technology and why are you an interesting technology and why are you an interesting founder founder founder that people should know about that people should know about that people should know about ? I am very happy ? I am very happy ? I am very happy to announce the to announce the winning teams. I think winning teams. I think winning teams. I think we'll we'll we'll introduce Spiro from introduce Spiro from introduce Spiro from Komodo first. Please come on stage. Komodo first. Please come on stage. Thank you. I think you Thank you. I think you will take the floor and will take the floor and will take the floor and start the presentation, start the presentation, start the presentation, right? Good. Do you have right? Good. Do you have right? Good. Do you have clickers or will you clickers or will you clickers or will you start your own? OK. start your own? OK. start your own? OK. Hello everyone, I'm Spiro Hello everyone, I'm Spiro Hello everyone, I'm Spiro Anakos, one of the Anakos, one of the Anakos, one of the founders of Komodo. This is founders of Komodo. This is founders of Komodo. This is Dom. And I'm Nick Anakos. We are Dom. And I'm Nick Anakos. We are the founders of Komodo. Komodo the founders of Komodo. Komodo the founders of Komodo. Komodo has introduced an agent has introduced an agent has introduced an agent approach to trading approach to trading approach to trading physical goods.

physical goods. From the rice you eat to the gold to the gold to the gold chain around your neck, the chip chain around your neck, the chip chain around your neck, the chip in your smartphone, the in your smartphone, the in your smartphone, the fuel in your car, or the fuel in your car, or the fuel in your car, or the copper pipes in our copper pipes in our copper pipes in our data centers— data centers— physical goods physical goods physical goods run the world. But the run the world. But the run the world. But the world world world trade system is broken. trade system is broken. Trillions are traded every year, but hundreds of Trillions are traded every year, but hundreds of billions billions billions are lost due to are lost due to are lost due to fraud, fraud, fraud, fragmentation, and fragmentation, and fragmentation, and the incredible the incredible the incredible complexity of complexity of complexity of coordination. Komodo coordination. Komodo coordination. Komodo took this complexity and took this complexity and took this complexity and turned it into an turned it into an turned it into an agent-based agent-based agent-based execution layer. Our execution layer. Our execution layer. Our proprietary CFC state machine proprietary CFC state machine proprietary CFC state machine runs runs runs specialized specialized specialized agents that agents that agents that fully validate fully validate , execute, and , execute, and , execute, and close deals. close deals. People often ask, " Why is everything so complicated?" Why is everything so complicated?" Why is everything so complicated?" And if you think about it, that's And if you think about it, that's And if you think about it, that's every country in every country in every country in the world, every resource, the world, every resource, the world, every resource, all their different currencies all their different currencies all their different currencies and banking laws. and banking laws. and banking laws. So of course, this will So of course, this will So of course, this will inevitably lead to inevitably lead to inevitably lead to disaster, right? disaster, right? disaster, right? But thanks to AI, this But thanks to AI, this But thanks to AI, this can finally be can finally be can finally be solved. We are very solved. We are very solved. We are very happy to see you here.

happy to see you here. Our first client from the Our first client from the UAE is also in the room, a trading company UAE is also in the room, a trading company UAE is also in the room, a trading company from Dubai that from Dubai that from Dubai that does does does hundreds of millions of transactions hundreds of millions of transactions hundreds of millions of transactions every year, and he has already every year, and he has already every year, and he has already tried our tried our tried our product. We are happy to be product. We are happy to be product. We are happy to be here today and here today and here today and thank you for this thank you for this thank you for this opportunity. opportunity. $17 trillion worth of physical goods move around the world every year. $17 trillion worth of physical goods move around the world every year. The world is held together by The world is held together by The world is held together by them, yet it still them, yet it still them, yet it still moves on paper, moves on paper, moves on paper, email, and email, and email, and handshakes. No handshakes. No handshakes. No one knows who is one knows who is one knows who is real and whether the real and whether the real and whether the documents are real. documents are real. documents are real. The result is billions in The result is billions in The result is billions in losses. Commod losses. Commod losses. Commod integrates into your integrates into your integrates into your stack: Gmail, Slack, Stripe, stack: Gmail, Slack, Stripe, stack: Gmail, Slack, Stripe, Salesforce, WhatsApp and more. Salesforce, WhatsApp and more. Salesforce, WhatsApp and more. The deal begins. The deal begins. The deal begins. Agents verify Agents verify Agents verify both sides: KYC, both sides: KYC, both sides: KYC, identity, and sanctions in identity, and sanctions in identity, and sanctions in seconds. seconds. seconds. Forensics Forensics Forensics instantly detects instantly detects instantly detects forged documents. A forged documents. A forged documents. A private private private room for transactions is being created. room for transactions is being created. room for transactions is being created. Delivery Delivery Delivery and logistics partners and logistics partners and logistics partners are connected with are connected with are connected with up-to-date prices. up-to-date prices. up-to-date prices. Then a swarm of agents Then a swarm of agents Then a swarm of agents structures the deal structures the deal structures the deal in parallel. in parallel. in parallel. Compliance, legal, Compliance, legal, Compliance, legal, currency, banking currency, banking currency, banking , shipping, , shipping, , shipping, billing—weeks of billing—weeks of billing—weeks of work in minutes. The work in minutes. The work in minutes. The swarm swarm swarm obeys a obeys a obeys a single protocol—CFC, single protocol—CFC, single protocol—CFC, our patented our patented our patented state machine. Agents state machine. Agents state machine. Agents advance only after advance only after advance only after passing each passing each passing each prohibition condition. He prohibition condition. He prohibition condition. He cannot move cannot move cannot move until the evidence is until the evidence is until the evidence is real and real and real and the calculation is possible the calculation is possible . He never . He never . He never touches the money.

touches the money. touches the money. He only gives a signal He only gives a signal . Your bank transfers the . Your bank transfers the . Your bank transfers the money. 30 days— money. 30 days— closed in minutes. closed in minutes. closed in minutes. Commod—a product you Commod—a product you Commod—a product you trust. trust. What is happening What is happening now? now? now? I don't know. I don't know. I don't know. Is it my turn? Is it my turn? Is it my turn? Yes. Um, I I I realized Yes. Um, I I I realized that we didn't give you a that we didn't give you a that we didn't give you a microphone. So, are you microphone. So, are you microphone. So, are you still connected to still connected to still connected to the microphone at the other the microphone at the other the microphone at the other table? Oh, you're already with table? Oh, you're already with table? Oh, you're already with microphones. Okay, microphones. Okay, microphones. Okay, great. Um, I think we'll have a great. Um, I think we'll have a great. Um, I think we'll have a little little little question and question and question and answer session. Yes. answer session. Yes. So, should I ask So, should I ask questions or what? questions or what? questions or what? That's right. That's right. That's right. My team My team My team agreed to this, not agreed to this, not agreed to this, not me. me. Wait, why are you Wait, why are you turning this on? turning this on? turning this on? Joshua, do you want to start? Joshua, do you want to start? Joshua, do you want to start? Well, of course. I am very Well, of course. I am very Well, of course. I am very interested in the second interested in the second interested in the second project, common.io. And project, common.io. And project, common.io. And you know, of course, we you know, of course, we you know, of course, we need a common need a common need a common space for space for space for agents to collaborate. agents to collaborate. agents to collaborate. I wonder if you've considered I wonder if you've considered I wonder if you've considered HTML format instead of HTML format instead of , say, a , say, a , say, a markdown file? What did you markdown file? What did you markdown file? What did you think about that?

think about that? Oh, sorry. Well, of course. Well, of course. Yes, no need. Yes, no need. Yes, no need. Yes, I was just saying that, Yes, I was just saying that, Yes, I was just saying that, you know, the second project you know, the second project , common.io, it's about , common.io, it's about , common.io, it's about This is the next one, yes. This is the next one, yes. This is the next one, yes. Oh, next one. Oh, next one. Oh, next one. ask them about the ask them about the ask them about the first one. first one. first one. Well, so Well, so Well, so I'll ask a question. I'll ask a question. I'll ask a question. This is one of those This is one of those This is one of those products that is very products that is very products that is very difficult to create because difficult to create because difficult to create because it requires so much it requires so much it requires so much highly specialized highly specialized highly specialized knowledge, as well as knowledge, as well as knowledge, as well as developing all these developing all these developing all these elements to gather the elements to gather the elements to gather the information information information needed for the needed for the needed for the product. A lot of my product. A lot of my product. A lot of my fear about fear about fear about modern startups modern startups modern startups is that what is that what is that what they build will they build will they build will eventually become eventually become eventually become a function of a sufficiently a function of a sufficiently a function of a sufficiently intelligent intelligent intelligent model. Do you see a model. Do you see a model. Do you see a future where models future where models future where models become become become powerful enough and powerful enough and powerful enough and tools from Anthropic tools from Anthropic tools from Anthropic and OpenAI can and OpenAI can and OpenAI can compete with you compete with you compete with you if they become if they become if they become smart enough smart enough smart enough and develop in the and develop in the and develop in the right direction? right direction? right direction? Do you want to answer? Do you want to answer? Do you want to answer? Yes, I will answer. So, Yes, I will answer. So, Yes, I will answer. So, in commodity trading, it is in commodity trading, it is very important what kind of very important what kind of data is data is data is involved. And what involved. And what involved. And what we did with our we did with our we did with our orchestration layer orchestration layer is we set up a " is we set up a " gate" for each gate" for each gate" for each agent. Therefore, they agent. Therefore, they agent. Therefore, they access access access real databases and real databases and real databases and have this have this have this context themselves. Whether it's context themselves. Whether it's context themselves. Whether it's gold, gold, gold, oil, grain, or oil, grain, or anything else, there are anything else, there are anything else, there are certain procedures that certain procedures that certain procedures that only only only industry experts know about, and industry experts know about, and industry experts know about, and they are the ones who have the they are the ones who have the they are the ones who have the documents and documents and documents and information to information to information to obtain them. That's why obtain them. That's why obtain them. That's why we're working with we're working with we're working with Connor and a few Connor and a few Connor and a few other partners other partners other partners to collect this data, and to collect this data, and to collect this data, and I have the industry I have the industry I have the industry experience to understand experience to understand experience to understand what's real and what's not what's real and what's not what's real and what's not and train agents and train agents and train agents based on that. So, based on that. So, based on that. So, to answer your to answer your to answer your question, can this question, can this question, can this happen? They happen? They happen? They could, of course, learn could, of course, learn could, of course, learn more from the open more from the open more from the open internet, but this is a internet, but this is a internet, but this is a private industry with a lot of private industry with a lot of confidential confidential information. And they information. And they information. And they must learn from must learn from must learn from certain data to

certain data to certain data to develop and develop and develop and become experts in become experts in become experts in this field. this field. this field. So, it sounds like So, it sounds like So, it sounds like something that's quite something that's quite something that's quite individual, in the individual, in the individual, in the sense that every sense that every sense that every company is a little company is a little company is a little different. You different. You different. You need to get need to get need to get everything you need for them everything you need for them . How many customers does it . How many customers does it . How many customers does it take for a take for a take for a company to become company to become company to become worth a billion worth a billion worth a billion dollars or reach, dollars or reach, dollars or reach, say, 100 million in say, 100 million in say, 100 million in annual revenue? annual revenue? annual revenue? How many How many How many customers do you think this would customers do you think this would customers do you think this would require? require? That's about 100 That's about 100 customers in total. customers in total. customers in total. Huge scale. It is Huge scale. It is also important also important to mention that the size of to mention that the size of to mention that the size of the deals in this industry is the deals in this industry is the deals in this industry is colossal. It's not colossal. It's not colossal. It's not like Stripe, for example, which like Stripe, for example, which like Stripe, for example, which processes processes processes microtransactions. microtransactions. microtransactions. One transaction could One transaction could One transaction could be, you know, be, you know, be, you know, $200 million $200 million $200 million or even more. And or even more. And or even more. And one partner, one partner, one partner, for example, an for example, an for example, an oil oil oil refinery, conducts refinery, conducts refinery, conducts operations worth operations worth operations worth $9-12 billion per $9-12 billion per $9-12 billion per month. So entering at month. So entering at month. So entering at this level with a this level with a this level with a commission on every commission on every commission on every deal fundamentally deal fundamentally deal fundamentally changes how much we changes how much we changes how much we can earn and can earn and can earn and what our valuation can be what our valuation can be what our valuation can be . I want to add one more . I want to add one more . I want to add one more point. Our point. Our point. Our solution is solution is solution is comprehensive, so it comprehensive, so it comprehensive, so it serves the entire serves the entire serves the entire supply chain, supply chain, supply chain, but the “wedge” for but the “wedge” for but the “wedge” for entering the industry is entering the industry is entering the industry is traders and banks.

traders and banks. traders and banks. Banks got out of Banks got out of Banks got out of trading physical trading physical trading physical goods goods goods years ago when years ago when years ago when fraud became fraud became fraud became too rampant. So this too rampant. So this too rampant. So this opens up huge opens up huge opens up huge liquidity because of the liquidity because of the liquidity because of the audit trails audit trails audit trails and the immutability of the data and the immutability of the data and the immutability of the data that we provide that we provide that we provide for the product and the for the product and the for the product and the transaction itself. Therefore. transaction itself. Therefore. transaction itself. Therefore. I think there's a I think there's a I think there's a lot to lot to love about this business. In love about this business. In love about this business. In many ways, it many ways, it many ways, it meets all meets all meets all the criteria of an ideal the criteria of an ideal the criteria of an ideal vertical vertical vertical agency business agency business agency business to build. If to build. If to build. If you look at YC's request you look at YC's request you look at YC's request for vertical for vertical for vertical agents, it's a agents, it's a agents, it's a perfect fit in the perfect fit in the perfect fit in the right industry. right industry. right industry. You can You can You can earn a lot here, and there are many earn a lot here, and there are many earn a lot here, and there are many specific specific specific industry tasks industry tasks industry tasks that create, in that create, in that create, in my opinion, a reliable my opinion, a reliable my opinion, a reliable protective barrier. My protective barrier. My protective barrier. My view is view is what to do to what to do to what to do to keep up with changes keep up with changes keep up with changes in the market, and how to in the market, and how to in the market, and how to evolve the evolve the evolve the product along with the product along with the product along with the development of models development of models development of models to remain to remain to remain competitive competitive ? It's not so much a ? It's not so much a ? It's not so much a question of, "Is this question of, "Is this question of, "Is this business relevant and business relevant and business relevant and great today?" And great today?" And great today?" And more about what more about what more about what will change in two or three will change in two or three will change in two or three years, or even years, or even years, or even next year, that will next year, that will next year, that will affect you in a way that affect you in a way that affect you in a way that Anthropic or OpenAI won't " Anthropic or OpenAI won't " eat your lunch." But eat your lunch." But eat your lunch." But what does it take what does it take what does it take to stay ahead of to stay ahead of to stay ahead of the curve and be a the curve and be a the curve and be a better product?

better product? better product? Yes, I think the only thing Yes, I think the only thing Yes, I think the only thing you can really you can really you can really protect is being the protect is being the protect is being the most creative most creative most creative in your category, in your category, in your category, no matter no matter no matter what product you what product you what product you create. In fact create. In fact create. In fact , nothing but , nothing but , nothing but creativity provides creativity provides creativity provides protection, right? protection, right? protection, right? Everyone has a product these days Everyone has a product these days Everyone has a product these days , but we , but we , but we really really really like the “ like the “ eat your eat your eat your competitor’s lunch” model, like competitor’s lunch” model, like competitor’s lunch” model, like Anthropic. We Anthropic. We Anthropic. We like the like the like the release model, after which release model, after which release model, after which ten companies ten companies ten companies become irrelevant become irrelevant become irrelevant the next day. So the next day. So the next day. So yes, it's an aggressive yes, it's an aggressive yes, it's an aggressive release and release and release and product excellence, but ultimately product excellence, but ultimately product excellence, but ultimately we will win we will win we will win through distribution. through distribution. through distribution. What excites you most What excites you most What excites you most about your about your about your roadmap, if roadmap, if roadmap, if you can share? you can share? Become a bank. Yes. This is fascinating. Yes. This is fascinating. Thank you very much. Thank you very much. Thank you very much. Great, thank you. Great, thank you. Great, thank you. Class. Class. Class. Support Komodo Support Komodo Support Komodo with a round of applause. And now we with a round of applause. And now we with a round of applause. And now we will invite the next will invite the next will invite the next team. You can team. You can team. You can come here. come here. come here. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. I appreciate it. I appreciate it. I appreciate it. Yes, that's cool. Yes, that's cool. Yes, that's cool. Thank you. Thank you. Thank you. Next up is Max Next up is Max Next up is Max Weinbaum and the team from Weinbaum and the team from Weinbaum and the team from Comet IO. Come on, Max. Comet IO. Come on, Max. Comet IO. Come on, Max. Uhu. Oh, are you going to Uhu. Oh, are you going to Uhu. Oh, are you going to show a video? Okay show a video? Okay , come here. , come here.

, come here. Okay, we'll show Okay, we'll show Okay, we'll show the video. It was the video. It was the video. It was This is music. This is music. This is music. A. A. That is correct.

Okay, come in. Okay, come in. Come on, guys. Oh, Come on, guys. Oh, Come on, guys. Oh, thank you. thank you. Um, could you... Um, could you... we're having a little we're having a little we're having a little technical difficulty here, technical difficulty here, technical difficulty here, the judges couldn't the judges couldn't the judges couldn't see the video. see the video. see the video. Good. Good. Good. We need to We need to We need to describe everything a little to, well, get describe everything a little to, well, get describe everything a little to, well, get started. started. started. Before you... Before you... Before you... Well, yeah, before you Well, yeah, before you Well, yeah, before you start the whole start the whole start the whole presentation, yeah. presentation, yeah. presentation, yeah. Oh, I understand. Oh, I understand. Oh, I understand. Imagine we are not a Imagine we are not a Imagine we are not a multimodal multimodal multimodal model. We model. We model. We only accept text. only accept text. only accept text. We'll give you all the We'll give you all the We'll give you all the context. context. context. Yes, yes, yes. Yes, yes, yes. So, Comet.io is a So, Comet.io is a multi-user multi-user multi-user markdown editor. It markdown editor. It markdown editor. It focuses on focuses on focuses on creating the best creating the best creating the best documents for documents for documents for modern work modern work modern work involving people and involving people and involving people and agents. But what you would agents. But what you would agents. But what you would see in the video see in the video is a familiar is a familiar document editor interface, where document editor interface, where people and people and people and agents appear, they agents appear, they agents appear, they notify each notify each notify each other, anyone can other, anyone can other, anyone can call anyone, call anyone, call anyone, and that's really the gist of it and that's really the gist of it and that's really the gist of it . . Perfectly. Perfectly. OK. So... OK. So... Friday afternoon, Friday afternoon, you're ready to go you're ready to go you're ready to go home, packing your home, packing your home, packing your laptop, and suddenly laptop, and suddenly laptop, and suddenly you hear beeps all over the you hear beeps all over the you hear beeps all over the office. The server office. The server office. The server is down, a massive outage is down, a massive outage . You put down your . You put down your . You put down your bag, walk to the bag, walk to the bag, walk to the conference room, and conference room, and conference room, and take out your laptop.

take out your laptop. take out your laptop. You have only You have only You have only one question in your mind. What's one question in your mind. What's one question in your mind. What's going on, Claude? You going on, Claude? You going on, Claude? You enter a query, enter a query, enter a query, wait a few wait a few wait a few minutes while it minutes while it minutes while it processes the ideas, and processes the ideas, and processes the ideas, and finally get finally get finally get an answer. And the first thing an answer. And the first thing an answer. And the first thing you do is you do is tell everyone tell everyone tell everyone else in the room these else in the room these else in the room these ideas so they can ideas so they can ideas so they can incorporate them into their Claude. incorporate them into their Claude. incorporate them into their Claude. Then someone else Then someone else Then someone else comes up with another idea, comes up with another idea, comes up with another idea, tells you, and you tells you, and you tells you, and you enter it into your Claude. enter it into your Claude. enter it into your Claude. And you realize: how did And you realize: how did And you realize: how did we get to this point? we get to this point? we get to this point? Is this work? There has to Is this work? There has to Is this work? There has to be a better way, and there be a better way, and there be a better way, and there is. These are agents, the is. These are agents, the is. These are agents, the people who do people who do people who do the work and those who are the work and those who are the work and those who are responsible for the responsible for the responsible for the problem, sharing problem, sharing problem, sharing context in context in context in real time. But real time. But real time. But how? Think about the how? Think about the productivity software productivity software you all you all you all use, the use, the collaboration tools. Can collaboration tools. Can you just add you just add you just add your your your coding agents there so they coding agents there so they coding agents there so they work together? No, you can't work together? No, you can't work together? No, you can't , because , because , because they are made for they are made for they are made for people, not people, not people, not agents, and that's why we're agents, and that's why we're agents, and that's why we're making Comet. This is a making Comet. This is a making Comet. This is a multi-user multi-user multi-user markdown editor. We are markdown editor. We are markdown editor. We are fully focused fully focused fully focused on making on making on making it the best it the best it the best document document document platform for platform for platform for modern work, for modern work, for modern work, for people and for agents.

people and for agents. people and for agents. It's open, It's open, It's open, modular, which is modular, which is modular, which is key for key for key for enterprises, for enterprises, for enterprises, for developers, and all developers, and all developers, and all wrapped up in a wrapped up in a wrapped up in a super-familiar super-familiar super-familiar user user user interface. I started interface. I started interface. I started my career in my career in my career in productivity by productivity by productivity by implementing implementing implementing collaboration features in Microsoft collaboration features in Microsoft collaboration features in Microsoft Word. Max and I Word. Max and I Word. Max and I worked together on Textio worked together on Textio , the first , the first AI-powered writing software, AI-powered writing software, six years before six years before six years before ChatGPT. We've ChatGPT. We've ChatGPT. We've implemented implemented implemented enterprise enterprise enterprise AI solutions for writing at AI solutions for writing at Fortune 500 companies. We've been doing Fortune 500 companies. We've been doing this for a long time. The market this for a long time. The market this for a long time. The market for commenting tools for commenting tools is expanding thanks to is expanding thanks to agents. As agents. As agents. As more people more people more people use use use agents, there will be a growing agents, there will be a growing agents, there will be a growing need for need for need for them to be able to them to be able to them to be able to share share share context in a context in a context in a product built product built product built for agents but for agents but for agents but designed for designed for designed for people working people working people working together in even the together in even the together in even the most demanding most demanding most demanding situations. situations. Perfectly. I Perfectly. I like the like the like the structure of this structure of this structure of this business, because it's the business, because it's the business, because it's the opposite of what opposite of what opposite of what we just heard we just heard we just heard from the guys at Commod, from the guys at Commod, from the guys at Commod, right? They went into a right? They went into a right? They went into a narrow vertical, narrow vertical, narrow vertical, trying to get the trying to get the trying to get the most value out of a most value out of a most value out of a very deep, you know very deep, you know , sphere, and you're going , sphere, and you're going , sphere, and you're going as broad as possible and as broad as possible and as broad as possible and maybe a little less maybe a little less maybe a little less deep, but that's deep, but that's deep, but that's intentional, right? And intentional, right? And intentional, right? And I don't consider this a I don't consider this a I don't consider this a disadvantage at all. I think, disadvantage at all. I think, disadvantage at all. I think, you know, one you know, one you know, one way to assess way to assess way to assess the landscape here is to the landscape here is to the landscape here is to realize that the realize that the realize that the agent economy is agent economy is agent economy is going to be huge. There's going to be huge. There's so much so much money going through agents, literally money going through agents, literally money going through agents, literally token spending, token spending, token spending, right? And if you right? And if you right? And if you can become a can become a can become a product that product that product that even even even one or two percent of those one or two percent of those one or two percent of those tokens touch, you can tokens touch, you can tokens touch, you can become very valuable, become very valuable, become very valuable, even if the margin or even if the margin or even if the margin or revenue per call revenue per call revenue per call or usage is or usage is or usage is relatively low. You

relatively low. You relatively low. You can be a very can be a very can be a very cheap, very fast, cheap, very fast, cheap, very fast, and very broad and very broad and very broad product and product and product and still achieve still achieve still achieve great scale, great scale, great scale, right? So I think it's a right? So I think it's a right? So I think it's a very, very very, very very, very interesting form of interesting form of interesting form of product. What were product. What were product. What were your most your most your most unexpected findings unexpected findings unexpected findings in terms of in terms of in terms of product design or product design or functionality to functionality to make it better make it better make it better for agents than for agents than for agents than counterparts built for counterparts built for counterparts built for humans? For example, humans? For example, humans? For example, why not just why not just why not just let agents let agents let agents collaborate in Google collaborate in Google collaborate in Google Docs, right? Or even Docs, right? Or even Docs, right? Or even try try try using GitHub using GitHub using GitHub as a repository not as a repository not as a repository not only for code but also only for code but also only for code but also for markdown? for markdown? for markdown? All modern All modern writing software is terrible for agents to use. Software like Google Docs has a like Google Docs has a very poor format. very poor format. very poor format. It has MCP. You can It has MCP. You can It has MCP. You can send a send a send a document there. You can document there. You can document there. You can let your let your let your agent read a agent read a agent read a document there, but you can't document there, but you can't document there, but you can't force them force them force them to collaborate in to collaborate in to collaborate in real time. And this is real time. And this is real time. And this is critically important for critically important for critically important for any real any real any real work that needs to be work that needs to be work that needs to be done. And as done. And as done. And as more agents more agents more agents join the join the join the workforce, they workforce, they workforce, they need to be there, need to be there, need to be there, along with your along with your along with your documents. documents. documents. Talking to our Talking to our Talking to our users who users who users who love working with love working with love working with documents and documents and documents and rely on them in their rely on them in their rely on them in their daily lives, we daily lives, we daily lives, we realized that they realized that they realized that they need to involve need to involve need to involve their agents because their agents because their agents because they are also starting to they are also starting to they are also starting to use them.

use them. use them. It's not just It's not just It's not just developers anymore. developers anymore. Okay, I think Okay, I think so. so. so. A. Final words? I A. Final words? I just want to say one more just want to say one more thing: I was really thing: I was really thing: I was really surprised when I surprised when I surprised when I talked to people at talked to people at talked to people at this conference when this conference when this conference when a lot of them said, "I a lot of them said, "I a lot of them said, "I created my own created my own created my own visual Markdown visual Markdown editor so that other people editor so that other people editor so that other people could look at it could look at it could look at it , but then , but then , but then I gave up." I just I gave up." I just I gave up." I just rendered an HTML rendered an HTML document. This is a complex document. This is a complex document. This is a complex problem that problem that problem that needs needs needs focused focused focused attention to make it attention to make it attention to make it great. And that's exactly what great. And that's exactly what great. And that's exactly what we do. we do. we do. Perfectly. OK. Good Perfectly. OK. Good Perfectly. OK. Good job, guys. job, guys. job, guys. Thank you very much. The next and Thank you very much. The next and Thank you very much. The next and last team to last team to last team to compete— compete— compete— we'll now show a video from we'll now show a video from we'll now show a video from Foundry. Foundry. Content creators don't Content creators don't have time to have time to have time to build a business, build a business, build a business, so they have to so they have to so they have to hire an agency. It's hire an agency. It's hire an agency. It's expensive, slow, and expensive, slow, and expensive, slow, and risky. Built by Foundry risky. Built by Foundry risky. Built by Foundry replaces the agency and replaces the agency and replaces the agency and allows creators allows creators allows creators to launch their own to launch their own to launch their own products faster and products faster and products faster and cheaper. Our AI cheaper. Our AI agents build agents build agents build turnkey businesses.

turnkey businesses. turnkey businesses. The creator only The creator only The creator only launches it. We launches it. We launches it. We continue to continue to continue to increase revenue. increase revenue. increase revenue. Enter the nickname of Enter the nickname of any creator, and our any creator, and our any creator, and our agents will become agents will become agents will become their biggest their biggest their biggest fans: fans: fans: they will watch videos, they will watch videos, they will watch videos, read comments, read comments, read comments, and find the " and find the " painkiller" that the painkiller" that the painkiller" that the audience audience audience will actually pay for. We've will actually pay for. We've will actually pay for. We've done this for done this for done this for creators of all creators of all creators of all sizes, and over sizes, and over sizes, and over half a million people half a million people half a million people use use use products products products created by our created by our created by our agents. The cost of agents. The cost of agents. The cost of software development software development is approaching zero. is approaching zero. All that remains is All that remains is All that remains is trust and taste, and no trust and taste, and no trust and taste, and no one has them one has them one has them more than creators. We are more than creators. We are more than creators. We are building an agent building an agent building an agent product team product team product team that will turn that will turn that will turn every creator into a every creator into a every creator into a founder. founder. founder. Woohoo! Foundry team, Woohoo! Foundry team, Woohoo! Foundry team, come up here. come up here. Thank you. Thank you, Max. Thank you. Thank you, Max. Hello, AIE. I'm Weston Law Hello, AIE. I'm Weston Law Hello, AIE. I'm Weston Law Gettys. I am the founder and Gettys. I am the founder and Gettys. I am the founder and CEO of Built by Foundry. We CEO of Built by Foundry. We CEO of Built by Foundry. We are going to are going to are going to turn every turn every turn every content creator into a content creator into a content creator into a founder of their own founder of their own founder of their own company. And I have company. And I have company. And I have great great great news today. I think we news today. I think we news today. I think we have found the formula for a have found the formula for a have found the formula for a successful business. You successful business. You successful business. You take your take your take your unique expertise unique expertise , turn it into a , turn it into a , turn it into a product, and solve the product, and solve the product, and solve the cold cold cold start problem with start problem with start problem with built-in built-in built-in distribution. Theo, you distribution. Theo, you distribution. Theo, you did it with T3 chat.

did it with T3 chat. did it with T3 chat. Yes, that Yes, that Yes, that sucked. This doesn't sucked. This doesn't sucked. This doesn't work. work. work. This works great for This works great for This works great for our our our users, but users, but users, but Gary did it with G stock Gary did it with G stock Gary did it with G stock and G brain. And we see and G brain. And we see and G brain. And we see that if we that if we that if we help creators help creators help creators take a step back and take a step back and take a step back and solve a real solve a real solve a real problem that problem that problem that their users have. their users have. So, we're not just So, we're not just creating another creating another creating another Patreon to put Patreon to put Patreon to put content behind a paywall. We content behind a paywall. We content behind a paywall. We help them help them help them solve the real solve the real solve the real problem problem problem behind it, and our agents behind it, and our agents behind it, and our agents are experts at are experts at are experts at finding pain finding pain finding pain points in points in points in user user user queries. Our queries. Our queries. Our companies have worked with companies have worked with companies have worked with many different many different many different niches and creators of niches and creators of niches and creators of varying sizes. And varying sizes. And varying sizes. And really, we're building really, we're building really, we're building this to give power back to this to give power back to content creators. They're content creators. They're stuck...I stuck...I stuck...I worked at TikTok for 2 years worked at TikTok for 2 years worked at TikTok for 2 years and met hundreds of and met hundreds of and met hundreds of creators who were creators who were creators who were stuck on stuck on stuck on advertising contracts advertising contracts , affiliate , affiliate , affiliate links, or links, or links, or selling $20 PDFs selling $20 PDFs selling $20 PDFs in their in their in their profile bio. And they are just one profile bio. And they are just one profile bio. And they are just one step away from starting step away from starting step away from starting their own business. They their own business. They their own business. They just need just need just need someone to help, and that's someone to help, and that's someone to help, and that's exactly what our exactly what our exactly what our agents do. At Built By Foundry, we agents do. At Built By Foundry, we agents do. At Built By Foundry, we engage teams of engage teams of engage teams of agents to build a agents to build a agents to build a recurring recurring recurring revenue business for authors revenue business for authors revenue business for authors so they can so they can so they can focus on focus on focus on publishing content publishing content publishing content while we continue to while we continue to while we continue to grow their grow their grow their revenue.

revenue. revenue. I have many thoughts I have many thoughts I have many thoughts on this matter. on this matter. on this matter. Good. Good. Good. So, the problem I So, the problem I So, the problem I see as a creator, see as a creator, see as a creator, even with the brands I even with the brands I even with the brands I work with, is that work with, is that work with, is that there is a there is a there is a saturation point where saturation point where saturation point where talking about a certain talking about a certain talking about a certain product with your product with your product with your audience becomes audience becomes audience becomes pointless. Like with Vercel, pointless. Like with Vercel, pointless. Like with Vercel, for example, it quickly for example, it quickly for example, it quickly got to a point where got to a point where got to a point where everyone in my audience everyone in my audience everyone in my audience was either already a was either already a was either already a Vercel user and did Vercel user and did Vercel user and did n't want me to n't want me to n't want me to "pair" it to them, or they weren't going to "pair" it to them, or they weren't going to become one and didn't want me to become one and didn't want me to talk about it. It's talk about it. It's talk about it. It's very easy to reach very easy to reach very easy to reach saturation point, and the saturation point, and the saturation point, and the best solution is to best solution is to have a much wider have a much wider have a much wider range of things to range of things to range of things to show your show your show your audience. One of the audience. One of the audience. One of the best things best things best things I've done for myself -- I've done for myself -- the single moment that has the single moment that has the single moment that has had the biggest had the biggest had the biggest positive impact on positive impact on positive impact on me, my career, my me, my career, my me, my career, my business, and my income -- business, and my income -- is going from is going from is going from four sponsors four sponsors four sponsors to 80. Because now to 80. Because now to 80. Because now I have a wider I have a wider I have a wider variety of things to variety of things to offer my offer my audience. And even audience. And even audience. And even if 80% of it is not if 80% of it is not if 80% of it is not relevant to them, the 20% that is relevant to them, the 20% that is relevant to them, the 20% that is relevant gives relevant gives relevant gives a conversion that I would a conversion that I would a conversion that I would not have otherwise received. not have otherwise received. not have otherwise received. The diversity of what The diversity of what I offer to I offer to I offer to the audience is where the audience is where the audience is where I find the most I find the most I find the most value. And every value. And every value. And every creator I've creator I've creator I've worked with has ultimately worked with has ultimately worked with has ultimately come to a similar come to a similar come to a similar conclusion. However, you are conclusion. However, you are conclusion. However, you are touching on some touching on some touching on some really important really important really important points here. Creators points here. Creators points here. Creators really do have better really do have better really do have better taste than most taste than most taste than most founders. Creators founders. Creators founders. Creators really do have better really do have better really do have better distribution channels distribution channels distribution channels than most than most than most companies. Creators companies. Creators companies. Creators know their know their know their audience better than audience better than audience better than most businesses. And most businesses. And most businesses. And you built a you built a you built a system that system that system that defines it. But it doesn't defines it. But it doesn't defines it. But it doesn't offer a solution to offer a solution to offer a solution to another problem, which is that another problem, which is that another problem, which is that you have to you have to you have to offer offer offer diversity to diversity to diversity to your audience. I can your audience. I can your audience. I can see how this would see how this would see how this would work great as a work great as a work great as a tool for tool for tool for businesses when they businesses when they businesses when they see that the see that the see that the author's audience aligns with author's audience aligns with author's audience aligns with their their their customer archetype, helping the customer archetype, helping the customer archetype, helping the business create business create business create better products for better products for better products for Theo. And I've said this to Theo. And I've said this to Theo. And I've said this to many companies

many companies many companies that want to work with me that want to work with me that want to work with me . Watch . Watch . Watch my streams. Watch my streams. Watch my streams. Watch my videos. Find my videos. Find my videos. Find my problems, then my problems, then my problems, then show how your show how your show how your product solves them. product solves them. product solves them. This is a much better This is a much better This is a much better offer than offer than offer than trying to start your trying to start your trying to start your own company as a own company as a own company as a YouTuber. As the YouTuber who YouTuber. As the YouTuber who YouTuber. As the YouTuber who has created the most has created the most has created the most companies companies companies after anyone else, after anyone else, after anyone else, as far as I know, this as far as I know, this as far as I know, this doesn't work. It just doesn't work. It just doesn't work. It just exhausts you and exhausts you and exhausts you and exhausts your exhausts your exhausts your audience too. audience too. audience too. The best thing you The best thing you The best thing you can do is can do is can do is partner with partner with partner with companies to companies to companies to teach them what teach them what teach them what your audience wants, and your audience wants, and your audience wants, and let them come let them come let them come to you with money to to you with money to to you with money to show it to those show it to those show it to those viewers. viewers. viewers. I understand. And I will say that I understand. And I will say that I understand. And I will say that your niche as a techno- your niche as a techno- blogger is very complex, blogger is very complex, blogger is very complex, because there are many options because there are many options . We found that for . We found that for . We found that for all our authors—I'll all our authors—I'll give just a few give just a few give just a few examples. One examples. One examples. One author is called Brandon, author is called Brandon, author is called Brandon, he creates content he creates content he creates content about farming and about farming and about farming and always forgets the always forgets the always forgets the best best best settings for settings for settings for his dryer, which his dryer, which his dryer, which sounds sounds sounds silly to us, but for silly to us, but for silly to us, but for his 2 million his 2 million his 2 million followers, it's a followers, it's a followers, it's a real problem. And real problem. And real problem. And now they pay now they pay now they pay him a monthly or him a monthly or him a monthly or annual subscription to annual subscription to annual subscription to solve this solve this solve this specific problem. specific problem. specific problem. So I think, Theo, So I think, Theo, So I think, Theo, unfortunately you've chosen a bad unfortunately you've chosen a bad unfortunately you've chosen a bad niche. I'm sorry. But niche. I'm sorry. But niche. I'm sorry. But if you want to change if you want to change if you want to change the vector to the vector to the vector to farming or farming or farming or gardening, maybe gardening, maybe gardening, maybe we could we could we could work together.

work together. But I would say that in But I would say that in any any any user niche, except for user niche, except for user niche, except for technology and maybe technology and maybe comedy, we've been able to comedy, we've been able to comedy, we've been able to take a step take a step take a step back. Our agents back. Our agents back. Our agents find a problem find a problem find a problem that that that people are willing to pay to solve, people are willing to pay to solve, people are willing to pay to solve, and not a single author in the entire and not a single author in the entire and not a single author in the entire history of our history of our history of our company has left company has left company has left us because they always us because they always us because they always remain remain remain profitable. profitable. You are very convincing. You are very convincing. Okay, great. Thank you. Okay, great. Thank you. Okay, great. Thank you. Applause. To the founder. Applause. To the founder. Applause. To the founder. Yes, thank you. So, we'll have a Yes, thank you. So, we'll have a Yes, thank you. So, we'll have a short break short break short break for discussion. The judges for discussion. The judges for discussion. The judges will go backstage for a will go backstage for a will go backstage for a few minutes. In the few minutes. In the few minutes. In the meantime, excuse me. meantime, excuse me. meantime, excuse me. OK. In the meantime, we've OK. In the meantime, we've OK. In the meantime, we've prepared this prepared this prepared this short video about the short video about the short video about the World's Fair. So take World's Fair. So take World's Fair. So take your time, judges. your time, judges.

This is our biggest This is our biggest event for AI engineers event for AI engineers . We are constantly adding . We are constantly adding . We are constantly adding new techniques for you to new techniques for you to new techniques for you to try. try.

Thank you very much. Thank you very much. OK. I think we're OK. I think we're OK. I think we're inviting everyone inviting everyone inviting everyone back to the stage. back to the stage. Applause for Startup Applause for Startup Battlefield. Congratulations, guys. Battlefield. Congratulations, guys. Yes. Okay, so we have a Yes. Okay, so we have a few prizes that few prizes that few prizes that we're going to we're going to we're going to give away. I think give away. I think give away. I think people people people are doing some are doing some are doing some checks back there. So, checks back there. So, checks back there. So, third place goes to third place goes to third place goes to Komodo. Theo, it seems you're Komodo. Theo, it seems you're Komodo. Theo, it seems you're handing over, Theo, yes. handing over, Theo, yes. Okay, we'll Okay, we'll take some take some take some team photos too. team photos too. OK. Next, and I think OK. Next, and I think this also reveals the this also reveals the this also reveals the order, we have Foundry in order, we have Foundry in order, we have Foundry in second place. Wonderful second place. Wonderful second place. Wonderful presentation. This is a presentation. This is a presentation. This is a great presentation. great presentation.

Congratulations. Perfectly. Okay, Congratulations. Perfectly. Okay, no secrets about the no secrets about the no secrets about the grand prize, Comment grand prize, Comment grand prize, Comment IO. Congratulations. Great IO. Congratulations. Great IO. Congratulations. Great video too. Congratulations. video too. Congratulations. video too. Congratulations. Well deserved. Do you Well deserved. Do you Well deserved. Do you also want to take a also want to take a also want to take a group photo? I think group photo? I think group photo? I think we'll all do it together? we'll all do it together? we'll all do it together? Ahem. Yes. OK. Ahem. Yes. OK. Ahem. Yes. OK. OK. These are the founders OK. These are the founders . We are just here to . We are just here to . We are just here to comment. comment. comment. Thank you all. Okay, we'll Thank you all. Okay, we'll Thank you all. Okay, we'll get there with get there with get there with a few people. a few people. a few people. Catch them. Joshua, Catch them. Joshua, Catch them. Joshua, come here. Yes. come here. Yes. come here. Yes. OK. Thank OK. Thank OK. Thank you very much, guys. Do you want you very much, guys. Do you want you very much, guys. Do you want to talk about to talk about to talk about developing with Hyper Agent? Yes developing with Hyper Agent? Yes , of course. Are there any , of course. Are there any , of course. Are there any final words you final words you final words you would like to add? would like to add? would like to add? I think you're exaggerating a bit I think you're exaggerating a bit I think you're exaggerating a bit , yes. , yes. , yes. Yes, I just want to Yes, I just want to Yes, I just want to thank everyone here at thank everyone here at thank everyone here at AIE. We received great AIE. We received great AIE. We received great feedback feedback feedback throughout, thanks to throughout, thanks to throughout, thanks to Hyper Agent for the invitation Hyper Agent for the invitation Hyper Agent for the invitation and Hyper Fans for helping and Hyper Fans for helping and Hyper Fans for helping with the video. Overall, with the video. Overall, with the video. Overall, thank you very much. Come visit thank you very much. Come visit thank you very much. Come visit us. It's free.

us. It's free.