🐙 Useful Agents To Start Building Today | Tina Huang

0:00Hello friends, how's it going? Let me see if the stream is on. Could I get a mic check? Sound check? Hello friends, here are men. Yup, you can ask questions during a live. No problem. Hello Carlos. Hello Piero. The ancient order. Where am I? Look at my IG. I ask people where I am. To guess where I am. I'm on the east coast of the United States.

1:00It's my last stop. Good morning, good evening. Good day and night. While it was a little also plugged in my larger speakers. There is background music. Can you guys still hear the background music if I turn up the game? You cannot stop the background music. New York, yes, there you go. I was actually in Jersey though. Yes, close enough. Nice. Yeah, yeah, New York. It's been a while since I've been back. No background music? Okay, wonderful. Okay, I'm going to get started. We can chat as I go along. As per usual, feel free to just put the questions or comments or anything you want into the chat. I'll be monitoring it throughout as well. Yes. If you do want slides as well, please do sign up here.

2:04We'll send you the slides and stuff like that afterwards. Okay. So we're going to talk about AI agents specifically talking about all the different types of AI agents that people are building and then also the market and monetization strategies. Now that we've done, I mean, like almost a year and a half, almost like two years now since agents became like a thing. So people have been building a lot more AI agents. We've also seen what works, what doesn't work, what works for like monetization purposes and what has, what have been people have been doing with their AI agents. And there's also been more like categories of AI agents that came out as well. So I wanted to do cover all of those today. Okay. So we do have quite a lot of slides. So to get through them, so I'm going to just get started here. But please feel free to ask any questions in the chat. I will be monitoring as well. Okay. So this slide is going to, I'm going to go over pretty quickly. So just navigate a strategic landscape of AI agents and business transformations for market intelligence, the implementation strategies.

3:06I'm going to focusing on the market intelligence implementation stuff, which I haven't really focused on previously. So foundations at market will go over this quite briefly, in herrence of the definitions and core capabilities. And we're also going to be focusing on the very practical aspects of what's the market projection, real world implementations and how to go about doing this, including monetization models and pricing strategies. Okay. So this is just here. Like I've gone through these slides before in other live streams. I'm not going to go over it. It's just in case you want to have a refresh on it. What are AI agents? There's software systems that use AI to pursue goals and complete tasks on behalf of users. So I do have two videos here. If you haven't watched it, I highly recommend that you watch those videos to catch up on it. Okay. So the market growth and intelligence. Let's talk about that. So AI agents since 2024, they once it became like a big thing. We've seen that there has been like significant increase over time.

4:10And this is the market size projection. So you can see very much like a hockey curve situation that's going on. This is based upon research. You can also click into these research sources. You want to get more details about where we're getting this data from. But we can really see that it's just a massive amount of increase. So from 2024, we're looking at $5.43 billion industry. We're looking at $7.92 billion industry and going to 2026, $11.55 billion industry. So it's just massively growing. We're going to be very much here to stay right now. So key market metrics, 263 billion by 2034. 51% of large companies are implementing agent AI at this point and 100% return on investment. So expected return on investment is 100% and over. So some and they're here also some of the case studies that we've seen in the past year and a half in which they were able to reach some of these metrics, including Lumen technologies who were able to recover for hours per seller each week. Chat base, which was able to 65% production in support tickets, which dramatically improve customer satisfaction.

5:17You see that the market is shifting from automation to true agentic intelligence now. I can also see that corroborating corroborating that is the right term. What we see over here, we can see that agentic platforms like the platforms that people use to develop agents. There also has been massive increase in usage as well as valuations. So if I tap here, for example, make an a 10, which is my favorite no code, low code tool. Yeah, there's like a lot of different of these platforms that people can use now in order to build these agents. See if anybody has any questions. Let's learn high from my psychosity. Richard Steve Hacks has finally arrived. Hello, Richard Hack. Richard, this Stephen Hack. Any questions? Can I have a spot for the bootcamp? Oh, we are going to be launching our bootcamp next week.

6:19So let me send you to link as well. If you want to. We're launching only to the wait list. So I really recommend that you sign up for the wait list. Last time we sold out in. I believe less than an hour is like 55 minutes, something like that. And the wait list gets access first. So if you're interested in checking it out, here is the link for it. But yeah, I have like a lot to cover today. So I'm going to cover more of the content we can talk about the bootcamp in a bit as well. All right. So let's talk about AI agents that we built. This is an example of an AI agent that we built. So we as the lonely archipist team, we also do B2B consulting and also implementation. We do B2C as well. So you see like the B2C stuff that we do because it's on the internet. But we also do quite a few B2B stuff as well. So one of the examples that we've recently built is a document processing agent.

7:19It's an AI agent that automates document processing and transfers manual workflows into a intelligent automation. And by doing this project, we were able to like actually experience the reduced amount of manual processing time, standardized output consistency and scalability in terms of handling increasing amount of workload. And this is just an example of something that we've done the text stack that we use was a zero open AI for AI processing. And a 10 with the workflow automation and extremely as the user interface, the reason why we chose this text stack was because the client was concerned about privacy reasons, right. And the zero is something that they were able to trust. So we built it off things that they already trust, which is using a zero open AI as opposed to just opening it by itself or any other model. And then n 8n is the trusty little automation tool because the client also wanted to experience building it and the client, the client is non technical. So that's why we decided to choose a text stack in which they were able to build out this automation workflow by themselves with our mentorship as well.

8:25And then the user interface that we chose to stream with because it's a nice simple little user interface that is pretty easy for people to get a hang of. The key inside that we got from this process, it was a few months for us to finish the project with that AI agents are able to completely transform a workflow bottleneck into competitive advantages. Reporting is a really massive part of many different companies. So this document processing agent is something that was able to help the client and a lot of other companies also have very similar needs in terms of their operational stuff. So this is one of the ideas of we've implemented that there has been clear, proven track record of it being able to help clients and like real world clients, not just like making little projects here and there. See any questions. Okay, no questions. Cool. So that was an example by wanted to increase this a little bit more give you guys a bit more of an understanding of the entire profitable AI agent categories because obviously this is not the only agent that is profitable and useful.

9:33So I want to expand that a little bit more to give you more like understanding of more different cases that are out there. So these are some proven use cases with measurable business impact and clear profit levers. So let's start with these first. These are the high return on investment categories. So the first one is customer support. This includes like 24 seven coverage and faster resolution. Clarina, for example, is able to do 2.3 million chats per month and around like it's around the approximate equivalent of 700 full time employees by using their customer service agent. Another one is voice and ordering. So this is capturing miss calls is audio based functionalities audio based agents and allows for capturing miss calls and higher conversions. Donate host. I don't know how to pronounce it. Donate host. I think that's how you pronounce it was able to see an increase of 71% conversions versus 58% before where they just had like automated systems for doing things. Another example is another category is the workflow automation. By the way, the way I arranged the structure, I was quite I didn't do it in terms of the industry itself. I didn't in terms of the categories of the types of AI agents to build if that makes sense.

10:49So this is something that is applicable throughout a lot of different types of industries themselves because they're just they're very, very applicable right. Another one is workplace automation. So meetings to task fewer dropped items, auto AI, for example, 62% users save greater or more than four hours per week. This is like a transcription tool. These are more like automation things I run into background. It just becomes a lot more convenient. You don't need to like transcribe things. You don't need to email people as much in terms of action items that are there. The software would be the agent and the software would be able to make sure that you're not dropping as much items and everything is just a lot smoother. So this is another really important category for high return on investments agents. Let's see if I need questions so far before I continue on. Does 100% return on investment mean it's effectively pointless? No, it means 100% increase return on investment. Right. So it's like two times increase.

11:55Yeah, what database use we use the vector store that was built into Azure AI as well. What totally do you use to deploy so deployment is was was with streamlit and it also Azure's own own platform as well. So they have like a business enterprise platform that makes sure everything is completely safe. Now we did not use super base. No, we did not. What was the monthly spend on that B2B solution? What do you mean by that? You mean like how much they spend to maintain it where how much it was to build it? Okay. I will continue on the rest of these categories and then go back to answering you guys questions. So the other category. And high return on investment category is research and data. So this is just like higher str throughput better targeting just being able to do research and data processing a lot faster.

12:55So clay, for example, is another company and they found that they were able to do two to three times better coverage versus using like a suit tools to do research and data. Another category is IT operations. So just faster MCR reduced downtime. This is just something that in general, like you're detecting like issues that are popping up and just page your duty, for example, is able to opt to 70% noise reduction by using agents in this category. And then there is code and content. This one is like you're not really coding agents out there. You're able to accomplish faster shipping and more consistent quality is able to help developers a lot more in terms of whatever it is that they're trying to build also in terms of content as well, just producing more content code rap is an example was able to have 86% faster code delivery rate and people are just able to write code a lot faster. Any questions for these six different categories listed?

14:04Questions, questions. Wait, where did you guys go? Oops. Okay, question from Tonya. Do you think the price of subscription premium tool will going to be more expensive or cheaper in the future? I mean, I don't it should be cheaper like in terms of how economics works. It should be cheaper over time or it would remain the same with additional functionalities were slightly increased pricing with additional functionalities, but over time, the general trend is that technology matures things get cheaper. Caviette, looking to the privacy concerns about order before you dive in that is very true. So in terms of things that involve privacy concerns, I think we've also learned a lot as we build more and more agents.

15:09Like as we progressed doing that, so that's also why, like for example, the project that we were building, we chose to use like a ZER in order to manage a lot of privacy concerns and things like that. Because they were looking for like an enterprise solution. So these are definitely really important considerations to make when you're actually building for a company as opposed to just like building a solo project by yourself. Make sense? What skillset are most in demand in the AI agentic space? I will talk about that actually in a couple of slides. So hang on to that. Which category can you write everybody writing a chat? Which category do you find to be the most interesting to you, whether it be something that you have experienced in or something that you're just interested in building? I'm curious. So customer support voice and ordering, where flawed information, research and data, IT operations, record and content.

16:11Well, category is most in demand. I mean, they're honestly all very much in demand. There's a lot of different startups and consulting and like freelancers that are all in all six of these categories, which is why like I decided to map up these categories the way that they are. Yeah, a lot of them, they really are. For example, coding content ones, you see so many like content generation, like vibe coding tools, these all belong that category. A lot of the research and data like clays and example, but there's also a lot of research data that's more B2B consulting base. So freelancers were consultants will be working with companies directly and you also get really big companies like Deloitte, for example, who would like McKinsey, I do a lot of these like big consulting companies. They would also be focusing on this area and improving. Then you have yeah, like voice and orderings, I did all of them do have a significant amount of interest and money that's going in.

17:13Let's see what people are saying. IT operations, where flawed information, IT, cool. Question from Leonardo, hello, are these categories meant for large enterprises, where can they also apply to vertical agents or micro small businesses, both they are very much both. So yeah, I've seen like so many different types of companies. I'll go into a little bit more detail about exactly how you people usually build these solutions from like a business perspective. But yeah, it's really for both. They're across different companies, different sizes of enterprises, different companies, different verticals as well. So looks like we got a lot of work flow automation, customer support and work flow automation research and data work flow, a lot of work flow automation, some IT as well, customer support voice and ordering code voice and ordering cool. Okay, looks like we have a pretty good spread of different ones. I'm very happy to hear that because it means that you guys have a distribution of different types of different types of AI agents that you're interested in building.

18:22Yeah, all of these have a lot of opportunity in them. They genuinely do. Okay, so hard to know what to focus on for me because by the time I employer adopts AI to landscape is going to be vastly different. That's fair. That is fair. I'm not going to deny that. Okay, so take a note of that. I think this is like very interesting because I haven't really heard people talk about this as much in terms of what kind of categories of agents seem to be useful these days. And then I also want to talk about when you're building this, the opportunity framework, right? How does to approach finding the kind of agent to be building in a very specific industry, Renish. I don't have too much time to cover all of this. But I just wanted to like actually give a framework about this. So the first one is the look is important things to look for pain points that are tedious, slow, expensive or error prone. So you usually start off with industry trends, like pick a market or industry that you're familiar with and something that is also trending upwards, right?

19:22You don't want to be working in a market that is actually decreasing. And then you want to identify the user pain points within that market, like where in the workflow are people are doing, like where is it that people are having problems? And then this is the part that's like very specificity. I these two are just common for whenever you want to build any solution, like any startup, any company, any freelancing role. But this is the part that is really important on the AI agents portion of it, which is looking at profit, leave your validation. So you want to look for places that AI is able to give a specific advantage. So the pain points themselves are looking for tedious slow, expensive or error prone areas. And then you are looking at the key profit levers. So these are the general levers that AI is able to do. So 21 seven availability low cost increased speed, scalability and personalization. So since AI is able to do this, you just think to yourself, okay, what's the industry trend? What are the user pain points that are tedious slow expensive or error prone? And how can I use these key profit levers that AI has in order to deal with these user pain points?

20:28So that is the framework for it. And then after you figure this out, you want to validate the return on investment of your solution by using a single use case before scaling it. So put into the chat opportunity framework if you understood this approach. If you don't understand it, also please put I don't get it. I will explain more in detail. This is also very important. So I want to spend a little bit of time here. Put opportunity framework. If you understand everything that I just talked about, you do not put I don't get it. Opportunity framework. Okay, perfect. Personalization is key. Just got here. What did I miss so much? I miss so much. I'm kidding. Don't worry about it. Opportunity framework. Yeah. Okay. Good. Good. Sounds reasonable. Okay. Cool. Okay. J9 you. I don't get it. Okay. I'm happy to explain what it is. Can you explain what part of it you do not get?

21:40So here are the different high return on investment categories of AI agents. And here is how you can approach figuring out what exactly is industry and niche it is that you should be building in by going through this process. Looking at where it is that AI is able to have the key profit levers and then choosing a type of agent to build in the high return of investment categories because these are proven categories where if you build an agent that is based upon this kind of solution, you would be able to get good results. Opportunity framework. Cool. Alright. Looks like you guys get it. Yay. I'm very glad. So good. Good to pay attention to this slide. I mean, you should pay attention all of them. This one is particularly good to pay attention to. I don't really see anybody like actually addressing it from this perspective that I'm aware of at least.

22:40Okay. So now let's talk about how to actually implement this. Right. So from this process, you're like, okay, like I kind of get it. What it is I should be building. So how do I actually go and build it? Well, the builders pass. I split this into like the no code where like builder kind of vibe coding solution for people who want to take a no code approach. And then we all I also cover decoding approach, which is. This is how we run our boot camps to, by the way, we always have a no code path and we also have a code path as well because the implementation in terms of the tools are different, but the concept and what you're solving is still going to be the same. Right. So for the builders path is the no code pathway of doing things. The this is actually like it's great. It honestly is amazing that people who don't know how to code are able to actually build these days. Like I think that's just crazy. Like just even a few months ago to like a year ago, it would have been very difficult for you to build agents without any coding background. But now you can by using visual workflows and natural language programming. Okay, prompting. So we call it creator friendly platform. So we're just using like builder and creator kind of interchangeably here. So just just FYI, if you're confused.

23:49The platform is that we generally would use would, for example, be na 10 in terms of visual flows. So dragging and dropping word flow automation connecting multiple servers with visual node based interfaces, especially useful for the word flow automation type. So like this category or flood animation type stuff. I also do like other categories as well, but it was kind of like built more for the stuff over here. So yeah, na 10 is my favorite go to when it comes to building with a no code approach. One that using one of the vibe coding told we pull up Replay here. I mean, it could be like other stuff. Honestly, you can use like Replay. You can also use like other vibe coding software as well. But why do we use like this combination? It's because like na 10 would give you the workflow of something, but the like actual like user interface and having it on the web like host day on the web and then adding like other parts to it. This is na 10 is rather limited when it comes to that. So it's actually quite nice if you can use a vibe coding tool like Replay, bolts, you know, like any of the vibe coding tools that you would like to use depending on the nature of it.

24:59Not going to go into much detail about like which to choose, but depending on the nature of what you're trying to build. And you're also your skill level. You can create like a very good infrastructure and also like a very good user interface surrounding the na 10 workflow. So this would be serving as kind of like your back end. And then this would be like your front end if you're more familiar with web application language. So an example of a success story is like Cynthia Chen, for example, she was able to build like doggy decks app, which is live on the app store in two months using cloud plus Replay despite not no engineering training. The ideal is shelf for five years until vibe coding made it possible. So this has opened up the doors for a lot of these different workflows to be done. An example of a no code email automation examples is like a super simple example would be asking your workflow automation to summarize my Gmail inbox daily and send updates to slack. And then you have a get email trigger AI summary formatting and messaging sending it to slack. You can build this with like 30 minutes with drag and job interface, no coding required if you know what you're doing. Any questions.

26:05Which AI are you using to build this? Yeah, the good thing about na 10 is that you can choose any type of model that you want, right? And also like other parts of it as well. You choose different models based upon the solution that you're trying to build. So I mean, like we I know I'm kind of glazing over it a little bit. I know that because like our book has 28 days long, right? It takes us like that going through that process and like going to the exact details of it. So sorry if I'm not going to like to so much detail because simply do not have enough time to do that. But you can like choose different models based upon your different use cases, and quite important. And then also things like tool use and things like that. Within na 10 there's a lot of different choices that you can make. It's not like a blanket thing like this is definitely the best thing for everything. It's really not like that. It can be very, very varied when it comes to what it is that you're building and what kind of things that you're choosing to build with. Hi, Tine first time joining your live. I love your content. Oh, thank you. Thank you. Thank you. And thank you so much for joining.

27:09Her lives are always available on the channel. Indeed it is. Yes, you can always check it out afterwards. You can click it. Yeah, it'll be it's always live. So don't worry about it. If you cannot pay attention, we got to go to something else right now. Where can I see info for the bootcamp? So I'll just, you know, I will just pin it. So I'm just a check out details. So we have it. Yeah, just join a wait list. We'll be sending out Tuesday at 11 a.m. EST is when we're opening up to the wait list only. So if you're interested, just sign up for the wait list so you'll get the invitation to see like the full information about it. So check out details. For bootcamp and sign up for wait list to get access. First, yeah, so you can check it out. If you want, if you're interested.

28:16I'm just going to pin the message. So yeah, feel free to check it out. If you would like to. Okay. There's a bootcamp. Yeah, we do have a bootcamp. So the reason why you might not have heard about the bootcamp by way is because we just sell out. It's like, I mean, thank you guys so much for this. It's honestly. Yeah, it means a lot to us. We only have a hundred spots available because we don't want to have more than that with our current team because we want to give people as much, you know, all the attention that they deserve that they need in order to do well in the bootcamp. So that's how we limited only a hundred spots. But we've been selling out like under an hour on the wait list. We never even like announced it to the public. So that's maybe why you haven't heard about it. But yeah, thank you so much for everybody that has joined the bootcamp and it's interested in it as well. Okay, anyways, so this is the no code builders path for at a high level. How it is that you should approach in. Here's like some suggestions. There's anybody have any questions about that. Do you use any agents for YouTube content creation? I do. I do use agents in parts of my workflow. That is true.

29:24I joined a wait list. Yay. How much would a bootcamp cost approximately a ballpark number? Maybe honestly, you do just wait until it comes out because we haven't like settled it completely yet. I don't want to give out a number that I'm not sure if it's going to be the final number or not. So set up your wait list. You'll get the details about that. Sorry. I don't want to say anything that potentially might be incorrect about it. Oh, congratulations, triage. It's birth bootcamp first come first serve. Yes, it's first come first serve. So we just open up to the wait list first 100 people who set up and pay an enroll. Then you get it. Okay, let's move on to the building AI agents with open AI multi agent famous. This is the code implementation of doing this. So you're still implementing the same solution. The framework over here is that we really like to use the open AI agents SDK specifically in the bootcamp and just in general as well. We're doing our BTV stuff and building things like ourselves as well. So building agents involves assembling components across several different domains like models, tools, knowledge memory, audio speech, guardrails and orchestration. Do you open agents SDK is a lightweight and powerful framework for building multi agent for flows. So this is the approach that we like.

30:39So the SDK I think it's the most comprehensive when that's out there. Like there's other competitors like like Google's version of it, for example. But I just feel like we open AI agents SDK really just has all of the components that are there. Like it has different models that are there. You don't even yeah, there's like different models. Okay, now you have models that are outside of the open AI ecosystem as well. Like you can do that too. And then there's like different tools, I function calling, web and file search. There's guardrails that are there for safety validations. There's orchestration, like deployment monitoring and improvement. There's ways for knowledge and memory. So they have vector stores and embeddings for you to use. And then there's also audio and speech. So real time API and voice support as well. So all of this is really easy and already implemented within the SDK. And then some of the SDK like key features would include having agents like core entities with specific rules and capabilities. Being able to handle like handoff. So handing off in multi agent systems like one agent handing off to another agent system super easy to implement in SDK different sessions like persistent conversations, state and context management, which is also very important.

31:43And also allows you to have tracing. So you're able to monitor agent performance and decision workflows as well throughout the process to see what is happening and how to improve your agentic system as well, especially for multi agent systems. This is really, really handy. So yeah, that's the reason why like we personally we really like to use the open as agents SDK. That's what we primarily teach as well as the tool. But like I said, you know, these tools always keep changing maybe next year or next month is going to be something different. That's totally fine because we're learning about like these framers and foundations, right. So the tools themselves may change right now. It's like high preferred and attend for no code mostly and then agents SDK for decode implementation. But even if it changes is okay. Because if you know these principles, you can use any tool really. So an example of a multi agent customer support system would be like the customer will say something like my subscription isn't working and I was charged twice angry. And then that you could have a triage agent, which would categorize this and then handed off to a technical agent. So it would probably categorize this as a subscription issue. And then it would the techno agent would assess it and be OK, like hand it off to like a building agent and the building agent would maintain a system.

32:53So this is kind of a workflow that you would be on a very high level. What it would look like if you're using the SDK to implement it. And you have how many agents is one, two, three, yeah, three different agents that they're performing different things between each other. And then you would be specialized in the collaborative each other in order to have the full conversation continuity and to make sure that the customer was able to get a response and to fix this situation. Let's see any questions. Pimp Paul, do most production workflows involve rag and vector databases? I can't imagine the business rules and content required for real world flow would fit in a model's regular context window. I want to say it depends. It really does depend a lot of them do involve rag and vector databases like in which you have to store certain context and information.

34:01There are a few that can come to mind in which you don't need to do that. If it's something like an automation tool, for example, that does something very specific between tasks. I don't think you would need rag and vector database per se, but something that you should definitely think about whenever you're building something. What I always say is like with these different components that are here, you don't need to implement all of them, but you should definitely at least think about whether you should implement some of these or not and go through it. In sequence, they figure it out. Hyper to workshop. 28 days on how many hours per day training. Yeah, so it's 28 days. It's two hours. So it's one hour. We're talking about the bootcamp. So yeah, it's 28 days. It's a one hour for the actual workshop presentation. And then we also have one more hour, which is the Q and A, but in between we do have a circle space where you guys are going to be building out projects. So throughout the agents bootcamp, you would be building minimum of four different agents starting from simple to more advanced. And then you would be building it. You'll be talking with each other.

35:09And we're also going to be helping you throughout the process as well. So it's constant asking questions and then be able to help you through that process too. That's the structure of the bootcamp. Yep. From Nicole, if you use Azure, you want to go with a code approach, what tool do you use that similar to open AI agents SDK? You can still use Azure because Azure has open AI, right? Azure is more just for making having the was a call like privacy stuff, right? So it's like, for example, you can get the opening I key from Azure to using Azure opening as opposed to like opening I directly. And you can still plug that in and use the SDK while you're doing it. It doesn't stop you from doing that. Do you test agent three? No, what is agent three? Oh, wait, do you mean here agent three? Could you rephrase your question? Do you mind clarifying what you mean by agent three is this like a tool or product? What do you mean like a third agent? What is agent three?

36:18Do you use an agent? Yeah, I do. I do. I use an agent. Yeah. Tight and it has a self-hosted origin. I'm a third on the home server. Okay. That was a comment. What would you focus on if you had to choose between aries for bodies for AI implementation? That's a hard question. I don't know how to answer that question because it depends on so many different things. I mean, I'm like super biased, right? Like I would personally choose AI implementation because that's what I'm doing right now. But you know, depending on what your factors are, that might be different for you. Yeah, I would go with like AI research where I am implementation personally just because I'm very biased in those. What day next week is the book and vice I can buy it? It is 11 a.m. EST on Tuesday. Yes. Yes. I believe so. Yeah. 11 a.m. EST on Tuesday. 4D wait list only. Yeah.

37:23You will get email. Is the book and for beginners where you need to have some knowledge? Great question. So no, you do not. We designed the book camp where you can, you're coming in. You don't need to have any knowledge, but people who do have a little bit of knowledge are also able to be our fine because you will have optional assignments on top of the four agents as well. So we do like do it so that everybody learns about the framers and principles that's always been my my kind of model. Like I don't believe in focusing purely on tools. I believe in focusing on understanding the big picture and the frameworks and what's happening in the industry. And then using different tools to implement these solutions that are being cut out that we do. And right now we use an A10 combination with vibe coding for the no code people and then for people who are doing the code implementation. We use age agents SDK for that. Yeah. And they would have the each project each of the four agents and optional projects as well. You have the option of doing it with a no code version and also a code version where you can do both.

38:25You know, you can do that too. You also have lifetime access to everything. So some people like do one and then they do the other. Maybe they try to code when they think it's too hard for now. No problem that you can go back to the no code first. So you can switch between them. What type of pre-qualification do you need to join a book and I have no experience with the agents. You do not need any experience with the agents. Do not worry. Is this like going to stay on safe forever. Yes, it is. It is going to stay forever. No replete agent. Huh? Not sure what we're talking about here. Oh, oh, from replete agent. Okay, God it. No, I did not test it. No. Replay agent three. No, I have not where am I half and I forgot. That's also possible. It has done a lot of times. I don't think so. When we wanted this come out. What are the subscription required to build the app? What? To build this? I mean, it's not a subscription for agents SDK. You're just paying for the API stuff. And then for any time you're paying for any time.

39:26You can host it. You can pay for any time in cloud. If you want to host it, or you can do it locally as well, which is free. And you can pay for replete for hosting. If that was the question. Any other questions? Any other questions? Okay, let's see 200 minute auto auto noem dev with agent three. Okay, it seems like I have to go check out agent three. Thank you for that. I will go check it out. Let's see. Okay, this is a great question, dark. How were you able to confirm privacy concerns with AI agents LM not getting outside or leaking data of the customer tenant domain using enterprise license as your compliance. That is a great question. So my simple answer to this question is that we are depending on a zero like we're literally using tools like a zero, for example, precisely because they are supposed to be trustworthy in her handling the privacy. Right.

40:29So why is it that we're not just like building our entire privacy system because we do not trust our ability to build up an entire privacy system in a way that is as good as what's ready out in the industry. So that is like your simple answer as to how it is to confirm this the additional thing would just have to honestly be trial and error at that point. Like a lot of testing whenever we did is that we build something we always go through like a lot of processing of evaluations and testing, which is like these components of e-vals and testing is stuff that people don't talk about, especially on YouTube. Like I don't know why people don't really talk about it. They just kind of talk about like building the agent and then just leave it, but kind of like the testing and the evaluation is just as important. If not more important. Before you actually plan an agent into production. Yeah, I don't know why people on YouTube don't talk about that part, but yes, we do that a lot. Let's see. I was fortunate. Yes, thank you, Mark. I'm very happy that you like the bootcamp. I really recommend.

41:32Definitely want to do both code and no code. Yeah, a lot of people do like both implementations where they start with one and then to do the other when they switch around as well. So people, I think, yeah, like a lot of people start off with something and then they just eventually do the other version as well. The thing is, I actually think that people who do like code implementation, this might be a hot tape because a lot of coders are like, oh, you should always use code for everything. And yes, I think the advantage of code is that it's a lot more flexible. It's a lot cheaper. You can build a lot more things that you cannot do with no code tools. It's also a lot faster. You also have access to new things. This is the truth. Right. That is the complete truth. If you know how to code, you always have those advantages. And it's very good for you to know these things. But with that being said, even for myself, like I do know how to code, but do I sometimes build things using N A 10? Yes, I do because N A 10 is sometimes it's like very specific. It's like a specific thing I'm building. I know N A 10 will be really easy to do that. And then when I'm communicating with stakeholders who don't know how to do code, for example, it's a lot easier for me to build an N A 10 workflow and explain that process as opposed to having to explain an entire code base.

42:34So I do think like there's reasons to do both and know how to do both as well. Okay, I'm going to continue on because we have 15 minutes left. Oh, no, I have so many more slides to cover. Okay. Yes, so let's talk about monetization strategies and value creation now. So stitching gears a little bit, you know, we talked about all how it is that you can build it. And we also talked over here about what are the profitable AI agent categories, the things that you should be building that we know how to do it. And we talked about the opportunity framework now, right. So let's talk about it from a different approach, looking at monetization, how it is that you monetize this sort of thing if you want to make into a business that you're not just doing it for for fun. So we find if you want to do for fun, these two, you know, definitely go do that. But assuming that you also want to make into like a money making thing. So there are like different ways of doing this. So multiple paths to value from service based models of scalable products, you can choose between immediate revenue streams or long term scalable solutions.

43:40Here are the three business models that we've seen seem to do very well when it comes to like proven track record. That's saying that these are the only business models that work right like if there are many, many different business models with these three are like the ones that we've seen have proven track record. The first one is freelancing. So a project based agent development using no code tools, especially we see this quite a lot of using code tools as well. Honestly, like both, yeah, definitely both. So not not just no code tools, but most people who use no code tools would go with the freelancing option. That's how that's better way of putting it. If you do know how to code, you can do all three. If you don't know how to code, usually people end up doing the freelancing option or working with other people to do consulting or product as well. So freelancing is just as his name suggests. It's just like project based agent development. Yeah, normal freelancing. And then for consulting is strategic guide us an AI agent integration and compliance. Sometimes like it's a combination of consulting plus freelancing.

44:42So you're consulting in terms of explaining to them like how it is that you should be building the agent. Like what's the tech stack that you should use? I'm sure that you're complying to all of your compliance things like, for example, maybe you're like building something in the finance field. There's like a lot of compliance things there. If you're building something in a medical field, there's a lot of compliance things there as well. So telling them, okay, like you need to use like as your, you know, for example, in order to comply to your thing, you cannot just do like a random API generation, right. Like stuff like that would be included in consulting and you would also need to like implement that sometimes depending on how it is that you do it. So those are kind of like agency based of freelancing consulting. And then there's not the product side, which is scalable AI agent solutions with recurring revenue. So the projects that this is the one that is kind of the most public, right. And then these are, you don't really hear about them as much because they're like B2B stuff. It's not publicly announced. The product stuff is like the VC backed where like lifestyle stuff, lifestyle business where people will build a product and then launch it online.

45:48And then people can see it. So all the like for like all the examples that were listed here, for example. Yeah, like these companies, right, they're all like product base where they build a thing. Maybe they get funding for maybe they don't and then they launch it and they get people to buy their service as like a software service. So we don't really hear about the other two as much, but this is still like very much happening. And it doesn't mean that there's not a lot of money that people are making from these two. These two, they actually don't take venture capital funding. So oftentimes they're making like a lot of money that they just directly pocket while for this one, if you're building a product with venture capital back, you might hear like things with like massive evaluations. But not necessarily you actually make that money and put it in your pocket. They're another type of product if you're not going for venture capital is going for like lifestyle product, which is actually our team is building right now. So we are looking into building like scalable agents as like a lifestyle product where we're not taking venture capital and just launching it directly.

46:49That's another way of approaching it. So we've actually done all three. So we've done these two a lot. And we're also building out this one as well as we speak fingers crossed. Let's see if any questions before I move on to the price. Oh, you know, let me just cover the pricing models as well. So in terms of the pricing models there, you can price it differently. Sometimes it's usage based like charging per API call or person transaction. So this is kind of tied mostly to the product size. So you would be charging it per call or it can be like per subscription as well. So fixed monthly or annual pricing. So this can be based upon all of these as well. And then outcome base as well, which is for paying for measurable results like able if you're able to decrease, I don't know, like increased speed by 20% or like decreased mistakes by 30% something like that. I don't know. So for like a specific deliverable, this is also quite common. So the outcome based one is more common for freelancing and consulting as opposed to product.

47:59So if you are doing freelancing and called consulting, it's again, this is not something that you must do. So I want to say caveat here is what we prefer to do and something that I feel like works the best from my experience specifically is having outcome based pricing. So I'm going to increase the strongest value alignment because customer success is going to be equal to your revenue and it's going to be equal to sustainable growth. So when clients pay for measurable business results, everybody usually wins through that. There's a lot of other ways of doing this, but I find that this is brings about the most amount of alignment. So it comes to everybody just winning in the situation, you don't really get conflict of interest in this kind of situation. Of course, the downside for this is like sometimes if you're very new into field, you might want to start with hourly, for example, because you simply don't know how long it's going to take for you to build it. And you also simply don't know how much it is that you can charge as well. So you might want to start off with hourly, but then maybe at some point once you figure out, get a better understanding of how the industry works, how your own services work to switch over to more like outcome based pricing.

49:05Let's see if anybody has any questions. Let me drink some water. It feels like my voice is kind of dying. Is my voice extra quiet today? We're not really. I feel like my voice is kind of dying. So I was also speaking at the HubSpot conference. I was just like talking to a lot of people recently, so my voice is dead. Does anybody else have a naturally really quiet voice? Because my voice is naturally, I don't know if it's like the tone of it or literally is just a quiet voice. I have a very difficult time being heard in a crowd. So I have to feel like I'm shouting for other people to hear my voice. Does anybody else have that problem? Let's see. Freelancing. Yeah, let me know in the comments. Which one do you find the most interesting to you? Like what one do you think makes the most sense for you to do if you're interested by freelancing, consulting or product?

50:24What type? Your voice is. Yeah. Okay, good. As long as you can hear me. I'm talking really close to my cuz well, I'm a mumbler. Same here. Right. Yeah. People are always asking me to speak up like chest voice, throw away type thing. I don't know. What's the difference between chest voice and throw voice? I just know very bad at singing. I'm not a very loud talker, but I am a little better at it than before. Okay, yeah, I really have that problem. It's like every time I talk to someone, they just cannot hear what I'm saying. And the more comfortable I am with people, the more I default to my natural voice. And then my natural voice is like, then people can't hear anything I'm saying. Anyways, great at that. I do have a mic. You're being super fast. Oh, sorry. I shall slow down a little bit more. It's because I'm so excited. Okay. Um, looks like we have mostly product, product and freelancing. Okay, now we do have a good distribution as well. I'm seeing a lot of product and freelancing and some consulting.

51:32Is the price of the book, I'm a stacy. No, it's not. It's just that I'm not exactly sure. Oh, which that we're going to be charging. So I don't want to like say something that's not right. But last time we were charging $1,000. No, no, it wasn't. It was less than dollars. It was nine. Actually, I see what I mean. I think it was nine, nine, seven, four, 28 days plus lifetime access to everything. That was, yeah, that's what we were charging last time. So I don't want to like say something that I don't exactly know what it is. The final number we decided on for this time around. Yes. Thank you. Consulting and freelancing. Thank you guys. I should know this number. I do. I do. Yeah. That was our previous amount. Um, consulting and freelancing, but product is very important. Okay. So looks like consulting and freelancing is the important one.

52:35Uh, is because, oh, okay, consulting and freelancing is the one you're gravitating towards a product is very important. As well as a product product consulting and consulting. Yeah. I really think there's a lot of advantages for both. It really is. Um, I don't think there's like one that's particularly better for the other. Like for some people having a product based solution is actually easier to start off with. Some people consulting is easier. Some people freelancing is easier. So in my specific case, I found freelancing to be the freelancing and consulting. Is the easiest for me to start off with because freelancing is kind of like implementing stuff. Right. And then consulting is kind of like, yeah, like strategy in addition to implementation. So I would often do both together as well. And then we started building products more recently. So I would build like products. Um, four companies as part of freelancing, but I haven't like built like an entire product and then scaled it and then like actually launched anything. So yeah, for a lot of engineers, um, were like product managers that were coming from industry, they usually find that product is the easiest to do because they've already gone through that process while freelancing and consulting.

53:42They need to build more of a reputation for that. Right. For me, it's like, I have the reputation online. So that's why it's easy for me to do those two. And the product takes a little bit more effort from my side. Um, worth every penny. Thank you guys. I'm very glad I'm honestly the feedback for the book. Have has been really, really good. And thank you so much for everybody that has joined and just worked so hard on your projects as well. Like we have, you know, on the LinkedIn group. Um, we have that as well. It's just like, it's really cool seeing everybody is building. So I'm really happy and thank you so much for saying that it's good because you enjoyed it. I didn't pay anybody to say this but no paid actors here. I really glad that we worked really hard. And that's why we keep it out of 100 people, right? Because I, I mean, because we sold out so fast, like we were talking as I should we increase the number, but I don't want to increase the number. Because with our current team structure at least, I just, I just want to make sure that everybody has the best experience that they can. So that's why we're not increasing the number, not because we're just

54:51being mean or something or like not letting people join, but that's the reason. Um, okay. So kind of a little summary for everything. So the agent take feature strategies strategic opportunities in the AI agent revolution. Um, so yeah, kind of a summary of what we cover is we talked about the market reality. We know that the market is very much growing. There's like 51% enterprise adoption now in 100% return on investment expectations, which is why so many companies are doing this. So we're going to start investing into building AI agents. The implementation is multiple past from doing this. If you're going from a code code perspective using like SDKs like the open AI agents SDK and for the no code version, you're using things like by coding tools with N a 10 for the automation workflow. So you're able to build, um, even if you don't know how to code, or if you do know how to code monetization. We also talked about that, which is on different types of monetization methods. If it comes to freelancing and it comes to consultate outcome based pricing usually creates the strongest value alignment assistance, but growth on the product side, you're usually looking at something that has recurring revenue.

56:00Either, um, you're doing it per usage basis, word subscription basis. So your strategic next steps at this point is to identify your opportunities. If you want to build something, I think it's really important to first identify your opportunities. Remember we talked about the opportunity framework. About identifying where it is that you could be creating it and across referencing it with proven different categories of AI agents to build that you can leverage the the AI advantage, the profit leverage choosing your path. Then you need to decide how you're actually going to build it like if you're technical, you want to use like SDK, do you want to use a no code platform. What combination you might want to use as well. And then you have to figure that out. Then you can start building it. So building that me people, focusing on outcome based value creation and then monetization. Yay. That's the whole process. So throughout the book, and we actually do go through like these steps as well. That's why again, like I cannot cover all of it right now because the book, I'm literally 28 days where we would go through these and you'll be building D agents.

57:01But that's okay if you don't want to take the bootcamp as well. Like I do want to like lay this out for you because this is how it has been shown to work well when you are building an agent and have it go from conception all the way to the ending part of it when you're like monetizing it and improving over time. So this is the full process of doing it. Cool. Yeah. And here's the if you want to join the lonely Archipelago agents book can wait list feel free to do so if you want. Yeah. Cool. Any questions. I want to join your team. We're not currently hiring. We have been like thinking about that. Like because we want to give more people the opportunity to join the bootcamp because we do limit it to like 100 people each time. So we have been talking about hiring potentially. So look out for that if we do hire. I just want to make sure that we don't like we're training people. We don't change the quality of the bootcamp. That is very important to me.

58:10Let's see. I'm not being paid by Tina. I just really love that bootcamp. Thank you. And kale, I don't know who you are. I can't tell from your name. Who are you? So I think like honestly if you guys. This is a really important slide. Yeah, I think like this is the a lot of times like people talk about things, but this is kind of the missing piece when it comes from a business perspective in terms of figuring out what it is that you should be building and how to think that your building is actually worthwhile or not. I think a lot of people like don't really think about this step for some reason. So yes, you should pay attention to this. And then of course the implementation part is very important as well. Yeah, and then like monetization of this obviously really important to.

59:12Does anybody have any questions for any of this? Because I do feel like this was a high level overview of a lot of different of the things about the different process. The big picture process if you want to be building agents that are useful in industry. Do you have anybody have any specific questions for any part of this process? So I don't qualify for your team. Can you hire me to protect you from non-dangerous animals? But you are very good at protecting from dangerous animals. Are you not seeing as you live down under? Oh, Nicole. Okay. Hello, Nicole. I do know you. Okay. Yes, Nicole is part of the bootcamp. Thank you so much. And thank you so much for being here as well. I'm really glad that you enjoyed it. We have you as I think we have your testimony on the website as well. Test the money, testimonial was difference testimonial. Yes, I think that is the right term.

1:00:13Thanks for your support. Is each iteration of the bootcamp different we'll be getting up today I info and trans after the 28 days. So the way that we framed a bootcamp that's a really good question. Unlike maybe like some other I know like the way that people do it. People have different versions of doing it the way that we do it is that because we teach by principles and frameworks and structures. So when you're learning the theory behind it's like how to implement these steps will make up an AI agent. It is not something that goes out of date like the tools themselves that we're teaching could potentially change that is true. Right. It does it could potentially change. But because you understand how to go about it. It's okay because even if the tools change you're still you're just using a different way to implement it. Right. Even if you're using a no code version or code version of doing it. They're just changing out a different tool for another tool. That's why it is not like you learn something and it becomes obsolete afterwards is not it will not.

1:01:13Yeah. So with people who do go through the bootcamp they do like get certain access to things like last times agents bootcamp got the AI sprint workshop for free. Actually was anybody here from the AI app sprint by any chance that we did like a few days ago anybody put AI app sprint if you were in the AI app sprint. I'm curious. I think yeah. Yeah, there should be a few right. Um. Not in the chat almost. Also bootcamp alumni. Oh, hey Daniel. Yay. I'm so glad. Do you need any app for the bootcamp? No. No, you do not need any app. We walk you through the full process of doing it. Well, okay. I guess you need the app to develop it. So you need to use an attend or agents SDK and vibe coding. Oh, cool. Yeah. Okay. That's awesome. Hey. So many of you guys were in the app sprint. Okay. That's pretty cool. Hello again. I hope that was a useful.

1:02:24That was so stressful to be honest with you guys. I was super stressful. Could you guys tell I was really stressed because we were doing the entire workshop like the app sprint doing it live completely with. And you know, we were just going with the vibes and like debugging and fixing things as it goes. So we were like super stressed. Could you guys tell. Okay, it was perfect. Oh, I'm so glad to hear that. What is streamlit streamlit is a application that allows you to have a user interface and a dashboard and be able to host things among other things. You could tell we were really nervous. A little I like the scrappiness of it. Okay. Okay. I'm glad you liked it. I was like try. I was like, we must remain calm. Like, this is really a good idea to do this entire thing live and go through the entire process and have hundreds of people watch us try to fix the bugs. I'm like, I don't know.

1:03:27But we feel like we should show you the entire process. Yeah, I was like, like, remain calm. At some point I was just like, breathe. Sorry, this is barely the ass. What is the cost for the bootcamp? So the bootcamp from a previous cohort was $1,000. Sorry, $997. I keep saying it. It's not $997. For this, we don't have the exact pricing yet. So do sign up for the whatever it is that is there. I don't know if we're increasing it more remaining the same or like what's happening to it. I don't have the exact details yet. So that's why I don't want to make any promises right now. But if you want to know the information, it should be in the email that we sent out. Yes. Yes, I noted that you were speaking really, really fast. Was I? Okay, maybe that's my coping mechanism by talking really, really fast. If I talk really, really fast, maybe they won't notice that I am saying things.

1:04:35Okay. That sure was great. Love the grind. I am glad. I'm glad. Like $997 is different from $1,000. It's $3 different. It's $3 different. Oh yeah, here's the thing was not the thing. Here's the he's scannity joint away list. I also pinned it if you want to like have the details out. First come for serve. 100 people really hope to see you guys. Some of you guys in the AI not the AI apps for me. Oh my god. In the agents bootcamp. In the agents bootcamp. Do you guys? Yeah, we might be hosting like more workshops as well. If you guys like the AI apps print, I think we asked for feedback as well. If you like that kind of thing hosting like a two hour workshop, maybe like a two day workshop, things like that. Yeah, and then the people in the agents bootcamp would also get free access to the upcoming workshop as well. That's like a nice little bonus. So do you guys like that format?

1:05:35Like this is just something that's like two hours on three hours along with two day long as a sprint. Any idea when will the bootcamp be available? So we are launching to the wait list only on Tuesday 11 a.m. Yes, he is turned standard time. Yeah. Yes, okay, cool. Yeah, I think in that case, we might do more of these workshop style things. Yeah, if my nerves can handle it. Okay, well, I think yeah, we're a little bit over ready. So I hope this is helpful for you guys this entire structure. You can sign up for the workshop link here if you do want to slides for what we talked about today. Yeah. The workshops are informative. Okay, great. You know, I just feedback. You know, we listen to the feedback. I listen very much listening to feedback you guys give us. So if you're like, I like workshops, I will do more workshops.

1:06:41Yeah. Thank you all so much for joining today. I hope this was really helpful and I hope to see you guys soon. Next video or live stream or in the bootcamp. Okay. See you guys. Have a good one.