Day 1 Q&A: AI Content Mastery Summit

Hello and welcome to the day one Q&A of the AI content mastery summit. I am so excited to be here. My name is Simone Leah and I have my co-host with me, Mitch Assa, and we are both so excited to hear what you thought of day one. But firstly, hello Mitch.

>> Hello. Great to be here.

>> So excited. And honestly, guys, we just can't wait to hear all of your questions and what you thought of day one. So, jump into the chat and let us know what you thought. But what did you think of day one, Mitch?

>> Yeah, I mean I every every day is going to be incredible. Like today especially. I I'm definitely a systems guy. So today we're diving into a whole bunch of different systems trainings and we've had a lot of great comments come through via email. I saw a comment um there by Key as well. Thanks for tuning in Key. Uh about Rosco's presentation all around app building which is very popular right now. So, uh, yeah. So, it's hopefully it's been a great day for everybody.

>> Oh, I hope so. And yeah, for me, like Rosco and Dr. Nikki were both standout sessions today. I just thought on a systems level and a business and automation level, both of the both of those guests just provided so much value and so much input that people could really dive into and create tangible outcomes in their business. So, yeah, what a fun day. So love it.

>> Well, if there's no question

>> absolutely.

>> So if there are people tuning in live, then make sure you add your questions into the question box. And if you're catching up uh on the recording, then do what you can to make it to the future live. So we'd be doing these every day for the rest of the week. And we're here to answer questions and go deeper. like that that is the the one thing I'd love to do is like come with some questions based on the sessions and then we can go deeper and and figure out how that relates to you and your business and what you're trying to achieve. So that would be my suggestion and uh and yeah, really make sure that you can get here.

>> Absolutely 100% totally totally totally agree with that. So, what is some things that people could take away from someone like Rosco's session that you think they could implement straight away? Like within the next 24 hours, what are some things you think they could implement?

>> Yeah. Well, I mean, I wanted to talk about something slightly differently. Right. Like I just want to compare two sessions in today's uh day one AI systems and strategy which is I mean there's two different types of sessions there with Simone Ferretti the way that you went through that is basically like his content system right which is a it's a process it's a strategy that you that he does in order to generate new new followers and views and so forth for his content brand whereas if you look at some of the other ones like Rosco and Ellen and um and potentially probably even Nikki Sweeney as well is hers is or their presentations are a lot more about the the structure behind that the foundation the building the technical components and the that's like the two key ingredients that you need in order to get started. So you it's first of all you need to understand what are you going to automate the the strategy component and then you need to understand the systems layer or the technical aspect of how you can then automate and build these systems and so I use automate a lot it does you don't necessarily have to automate it but you do need some sort of a a system in order to to structure all of this and a system isn't just using claude like I I think the the third part of this is you need to answer What what type of business am I building? Am I going to be a soloreneur and not take on any help from anybody else and not having teams and employees? If that is the case, then just building in Claude and adding all your your connectors is probably okay because you're the only one operating. But if you are somebody who wants to build a business and hire people and um train others how to use the system, then you need like a a single source of truth somewhere where you can go and everybody on the team can see what is getting done, where each part of the the system or each part of the strategy is at for all of the different projects or the content that you're creating. So that you it's a yeah, like I said, it's a single source of truth. So, you need to figure out all of these questions before you start building because so many of the content videos that I see, especially on Claude, are built for your soloreneurs, which is nothing wrong with that. Um, you can make really good money these days being a soloreneur, but you just need to answer those questions first. So, that's like a foundational um starting point that I wanted to just lay out here first so that we can begin to think about that. So, if we do dive into some of the other things I'm building, you just have an understanding that we're building for your bigger companies, the the the companies that want to hire staff and have all of the these different um people inside of a single system, which means we need to build it a little bit more in depth than if you're just building it for yourself. So, I hope that makes sense.

>> Absolutely. I love that. And that's where I think Dr. Nikki's session really blew me away because she was thinking on that bigger level, but still only as a soloreneur. So, absolutely love that you've brought that up about knowing what you're trying to build, knowing what your outcomes are and all of that sort of stuff so you can build the system around it. And as you've said, a lot of people are those solo entrepreneurs. I would love to know just you don't even have to go deep on this if you want to, but for all of those people who are kind of sitting there and they're more on that um just beginner or a little bit over beginner getting to that mid mid range, people are starting to talk about NAN. Is that something that your soloreneur is going to need? What is it? And should we even be talking about it at this level?

>> Yeah, absolutely. Um [snorts] so yeah if you've got traction 100%. So the way that I think about this I I was I was wondering whether to pull up N and and different automation uh systems and concepts today or not. Uh so there's a couple of people online if you are interested in that then help me steer in that direction if you want to learn a little bit more about like automation systems and database building and so forth. I don't want to go too deep on that but the there is an important uh trend that you probably need to be aware of. So the the the the scary concept for nontechnical people when you start talking about automations and systems are probably two words databases and uh and APIs. But if you just have a they're not that hard to learn and if you have a solid foundation in understanding databases and APIs then you will probably understand how to build with AI a whole lot better because I think the next trend that we're moving to is uh what a lot of different platforms used to do or do today is they would some of the bigger ones would provide an API for free and then if there was a huge amount of cost to doing certain processes in them. You would just add some credits into their API uh dashboard and you would you then get access to the API which Claude has done the best branding around APIs and they've called them connectors. It's all it is. You're connecting one piece of software with another piece of software and that's all you really need to know. We don't need to get get too technical about it. But I definitely believe that more and more brands are going to offer APIs and it'll be an API as a service. So it'll be a new income stream for them because everybody can see that we're moving to a place where either the soloreneurs are going to be using claude and APIing into tools via claude or the companies are going to be building their own custom systems like I mentioned yesterday where you have where you're using where you're pulling all of the data into a single single system like air table or a vibe coded tool or a a full proper coded tool. Um, so just having a little bit of understanding as to what they are, then that would definitely help build. Now that's a long way of getting to N. N is simply just a tool to connect APIs and and manipulate data through those APIs and send them into different directions. That's all it does. It's just an automation tool to help connect all of your different tools. So um yeah, there exactly where I was about to go next. So in the comments said is API equals a Zapia. Exactly. Zapia is probably the most well-known tool in this space. The they're doing a lot of great things. Um I think the reason why N took off recently is because they they're a little bit they're not entirely no code. They allow you to add in code which it just allows you to to build so much faster. Maybe Zapia's added that type of node in recently as well. I'm not sure. I haven't checked them out for a while. But just uh I I don't know how to code, but we use a code node inside of nan almost every day. And it's the reason is is because you you'll have different types of data from one you want to get from one point to the next and you need to structure it in a certain way in order to get it into the next API and sometimes these tools don't make it easy. Whereas now you can simply pull all of the the data from one side and then the other say I want to get it into this node on this side and then throw that into chatgpt or claude and say, hey, this is what I want to do. Can you write me some JavaScript in order to manipulate this data and it does it. So like that that's the reason why people are building with van is because you can combine code with no code and it's just a fast way to build these types of automations and agents and so forth. So, um, how's that sound so far? Am I going getting too far off track?

>> No, no, no. I love this. So, just to clarify because I know you were talking about building big big for big businesses and things like that. So, a big business is obviously more likely to use something like Naden where they can control the code for example. Would your soloreneur necessarily not need that level of detail? And would your bigger business like you're building for lift different companies, they would want that sort of level of coding and maneuverability or is it just interchangeable and depends what you like the best?

>> Um, no. I don't think those two are related actually because the bigger businesses don't care about the tools that you use. They just want the end result. Um so the yeah in in many ways it's more important for the soloreneur type person the person who's going to be doing it themselves to use the the tool that they are most comfortable with and sometimes it's a progression. So we started building in make.com which is again a tiny little bit more advanced than um than Zapia but still not as advanced as nan and the things you could do in there. So it was like a progression and I used Zapyard years and years ago when we were trying to connect just little tools and I I knew it could it could automate different tools back then and then once AI came out then everything adapted to to be ready for AI as well. So yeah I mean I I probably wouldn't complicate it. Nan looks scary for about the first seven days, but if you're just in there just testing things, pulling in some templates from the template library, understanding it, asking you can you can just grab the JSON code for the automation, plug it into Claude and say, "What does this do? How could I learn this? How what what else could I do for this automation? These are the tools that I use. How can I connect those tools to this?" And it can it'll it just train you on exactly what you need to do for for your business. So, um, yeah, I would just go straight to NAN these days. Like, yeah, spend seven days there and you'll you'll be up to speed pretty quickly.

>> Amazing. Does NAN replace anything that we're already using or just aid with anything that we're using?

>> Does it replace anything that we're using? No, I don't think it replaces anything. I guess the other big distinction that is probably important to know is when you start using APIs, your AI accounts are separate to the API account. So your you would still be paying for your chat GPT or Claude um as a let's call it like a retail account. That's your $20 a month depending on what subscription you're on. That's your $20 a month and that's for you to use their dashboard and and their internal connectors and their GPTs and projects and whatever else they have. When you start building you you now need to open up a developer account and these are basically pay as you go and that was actually quite a big draw for us in in the early days. So let me let me take you back to how all of this started. We started I started building these automations because we were using a tool called cast magic. That was they were very early out when um GPT3 started to take off and uh basically you would plug in your podcasts and then they would create all of the different things like your titles and descriptions and social media posts and all of that type of stuff. And back then I had enough knowledge to know that um we could pull that data in via API already. And then I just asked myself the question like is there a chat GPT API because if there is surely I can just build this quickly myself. And that's how everything started and led me down uh many deep rabbit holes since. And

>> it took me a few days to to figure it out and understand it. I hired somebody on Upwork to teach me a little bit about how to structure a database. But by the end of it, I now had a new skill. And then instead of paying cast magic, I think we're paying them $59 a month in order to do all of our podcasts, we or summit sessions at the time, we were paying chat GPT as we go for I think to do I I don't know the exact numbers, but let's say I was doing 15 different podcasts per month that we were running through this system in order to generate the content and we're paying $59. What I realized is that the amount of AI that we were using was less than like $6 at the time. So, they were making like a $50 plus margin on that tool that was not too hard to replicate. And there's there's a lot of tools out there. I mean, they've done a lot of development since and um they're probably way better now, but back then that's that's exactly where we were. So, the you with the development tools, you're only paying for what you use. So, I I I guess that's another perk is you can have so many more tools and rather than paying $59 a month for all of them, you're just paying like maybe a dollar here and maybe $2 there and then $10 for this other one. And the the costs are actually not too bad um if you if you're using so many different tools. So, I that's the that's where we're going. But then obviously I think we're going to have API costs that that will pop up over time as well. So, maybe it's going to equal out. But yeah, that that's a just another important um distinguishment to make so that you understand there's a bit of a nuance there.

>> Yeah. No, love this. Thank you so much. We do also have a question from Zippy if you're able to answer that one. If you've set up one system like Makos APR, can you take what you have and move it over to NAN or do you have to set it up from the start?

>> Great question.

>> You do have to set it up from the start. However, I mean, we thought about this. we're building inmate.com. We had a whole bunch of templates we were selling and we're thinking about this uh like a year and a half ago and you can I mean I imagine that the tools are absolutely better now. So like Claude and so forth, you would be able to if you if you trained Claude with enough uh of the N uh articles and also some example uh automation files, JSON files trained to like a project or something with all of that data. There's probably a GitHub repo that you can pull into there as well, so it understands it all. And then you say, hey, this is a make.com version of what we were doing previously. Um, we're moving it into nan. This is the these are all refer to your knowledge in order to um use best practices for nan. It might be able to get that. uh we we just built everything from scratch again because the other thing that I realized is the way you have to build certain thing certain workflows inside of make uh can actually be 10 times more efficient inside of nan so it's going back to what I said before whereas where inside of make you used to have to use things like um iterators and aggregators and like send data in all of these different directions now it's just one code node to do like 10 processes that make make.com used to used to need. So yeah, we just I mean we just built them all from scratch. The

>> Yeah, I was going to go into something else, but it's not relevant. So yeah.

>> No, I love that. Thank you. And I think once you've built something once, you you know, it is going to be quicker and faster and easier for you to replicate that in something else and more tailored to that thing anyway. So I think rebuilding it the second time does allow you to uh just refine what you already had there to get a better output in the end anyway because you've been using it and testing it. So yeah, I love all of that answer. Do you guys have more questions about N8N? Because we could probably like talk about this for a while. I know Mitch loves it, but we want to make sure that we are answering your guys' questions. Like you guys have said, yes, you'd love to know more about NAND. Yes, up here um just again in relation to NAND. So yeah, what sort of questions do you have regarding NAN and also the first sessions of the day? So fire them at us because we will keep talking about this forever if you let us.

>> [snorts]

>> Yeah, I I want to point out one other thing that is probably more important than nan which is the um it's the the understanding of the system that builds it. So like nan is good in many ways for connecting data from one point to another point but when we start talking about AI we want to understand like what is the system that we're building. So let me let me share my screen. I actually have a few concepts.

>> We're going we're going into this. I wasn't sure. I had it ready just in case, but I wasn't sure if we get here or not.

>> So, I try to re record a lot of my concepts as I'm building and learning. The most important I mean there's a few important ones. Let's just start with the five AI uh skills essential for success when it comes to automation. So, number one, like I said, which is API integration. understanding that having like a base level understanding again like Claude have made this so easy where they have rebranded APIs to uh why can't I just click off somewhere uh rebranded APIs to connectors that because that's all they are so just think of APIs as connectors and every single tool that has an API almost has really clear documentation of how to actually use that and all you need to do is plug in their documentation into claude saying this is what I'm trying to do helps helps step me through this process if you're ever unsure. So that's having a base understanding that would be great obviously then prompt engineering and prompt engineering when it comes to automation is slightly different because uh as many of you probably know like when you start talking about automation you have variables and the the variables are the things that change the the data. So, for example, if we're trying to create content based off our summit presentations, then the variables in that prompt are going to be the speaker name. They're going to be the transcript from that that speaker uh session. It's going to be the title of that session. So all all three of these things are variables which we will plug in which means every time the automation runs the same prompt is used but it plugs in those different variables which means we're going to get a different output every time based on that speaker session. Uh the next one next skill would be AI workflow automation and so much of this can be handled by Claude now is just like tell it what you're trying to do and it'll help step you through that. AI agent development is where we're moving to. So um that would be things like I mean open claw would be 10 times easier and safer within months. Um so those types of tools menace is a is a another version and then AI assisted app development which is that's what Rosco talked about today right which is your replet bolt base 44 even Google uh if you don't know Google if you're paying for a Google subscription then start building apps inside of AI studio you'll get some credits included for that and you can build some pretty cool stuff they're not things you can deploy externally but you can deploy uh and use some internal apps which are pretty cool. So you can get a feel for how to actually develop apps um just using your your Google subscription. So yeah, there's some of the the skills that I would recommend learning if building these types of workflows and systems are your thing, which is another important point, right? Like the if they're not your thing, you don't want to be uh hating life every day while building these. Um but I would say even if you don't think it's your thing, just get in and start just give yourself a week. start prompting uh like Google AI Studio or Replet or Bolt and just see what you can come up with because I've I've seen so many people over the last six months especially say I I'm really didn't think I was somebody who would want to do development stuff but bringing an app to life or bringing like something really cool to life is very rewarding and you start to get addicted to it. So maybe just give yourself seven days and and see if it's for you or not. Um so do those make sense, Simone?

>> Yeah, absolutely 100%. And like even me watching you build out the back end of this summit in real time, I haven't even been the one building it. I'm like this is so cool watching this come to life and that even that has just inspired me to be like now I need to learn more things like this is so cool. So, I totally agree with what you're saying there of give yourself seven days to to have a play and look at all of the things because that seven days is going to be amazing. You're going to grow. You're going to learn some stuff like you might not think it's for you. I never thought it was going to be for me. Like I was like, I'm the creative one. Um, but yeah, no, like it's been so exciting to watch and now I'm so inspired to do it myself. So, for you guys watching, like just dive in and have a go. And everything you said then was so clear and well articulated. So thank you Mitch for showing us that.

>> Thank you. And then so now let's get to the the database structure because this if you're going to build this that there's also an important uh understanding at least a a base level understanding of how to structure a data a database would be important. So I just typed this into Gemini to see if we could explain this in the in this in the clearest way. So it goes through all of this, but I think it's just better if I open up this uh graphic here. So when building a a database, it's important to understand the that there's relational records. So the best way that it's come up with to explain this is think about if you have any type of checkout tool or um CRM, when you add in your customers, you only add them in there once. If even if they purchase two, three, four, 200 times, right? The the main the the key record here is the primary key is then re in relation to the orders. So the orders always attached to that primary record for customers. Then within those orders, you have your different products. So, a single order might have 10 different products in it, but it's only still one order. You're not creating 200 different new customers every time a new customer every time the same customer orders. You're not creating um 10 different orders if you've got 10 different products in that single order. So, you we then need to just apply that to what we're building for AI. And if we come back to here, the way that I think about this is what are the foundational databases that we need for pretty much everything that we build. Number one is of course we need our AI models. So we can use any type of model that we want. We can use claude in our dashboards. We can use chat GPT. We can use grock. We can use um basically any model we want through different routers. We have our image models. So that would be another foundational table that I would create within this this database. We could have a prompts library where we can pull in different prompts and use them in our workflows. Uh our brand assets there. So we the the other way we build these systems is we build them so that you can plug in any type of brand. So some systems might have five different brands in them. You don't have to build new systems or new automations for your brands when you understand that you could have a brand assets table. So these are things like this uh this content here is wrong. But these are things like your brand voice when it comes to AI. So that's going to be uh unique to your brand obviously. Then the avatar for that brand would fall into brand assets. If you're connecting your social media channels, then it would be your social media profile URLs. So all of these different types of things fall into the brand assets table which means um we can then like let's pretend that this is brand assets. So this would be the big moves AI brand and then we want to create content. We would send in all of the brand voice from big moves the avatar we would uh do what's called a a lookup record. We would send that that record into all of these here that are linked to that brand. So, this is where I can start to lose people. So, I don't want to go too deep on this, but hopefully people are following is you you create your brand assets once and then as long as you link up your database correctly, every time you select that you're creating content for that top brand, then it's going to pull in the brand voice and the avatar details and that'll feed that into the automation every single time. So that's how we can build these these types of systems that are ready for all of the the different um all the different companies and brands that we use. So we just got a comment there. It looks like vookup or xookup. Exactly like it's pretty similar. The you can kind of do a lot of these things in inside of uh sheets as well, but the once you go to tables like air table air table make it super easy. So air tableable is a good middle ground between a proper database like superbase and a spreadsheet like um Google sheets. So like air table is the is the middle ground. Let me just give you a quick look at it. So this is the the back end of our sessions. So all of our sessions that we create. Let me just show you this. So we have here the summit name is linked to all of these different lookups here. So if I take that out, it's going to remove all of these things because these are in another table. Whereas if I add it back in now, all of these are these fields here are available for me to use inside of any automation that I trigger from this workflow or from this table. So yeah, I'm happy to go further into this, but I do recognize that for some people they could be tuning out or it could be getting a little bit complicated. So I don't want to go too far, but this this is like just having an understanding that this is important if you want to build these systems is a good enough start to then begin to explore that yourself. Um

>> I'd love to know from from the back end, guys, this is how it worked for me. I would find a speaker, have a chat with the speaker, get them to lock in on a session title, all of that sort of stuff. Are you self-populating that sheet, Mitch, or is that populating through something else with the information we give it? Like when I was giving the information about the speakers, was that self-populating in the background?

>> Uh, yes, correct. Um,

>> I may

>> can you just unshare the screen for two seconds and let me just pull up

>> how that worked.

>> So, so I mean I'm not going to open the website page because that's pretty self-explanatory. It's just a form on a page. Um, so what would happen is people would fill that in and sorry, let me just find where I need to be. All right.

>> Yeah, you can share again now.

>> Okay.

>> So, this is what nan looks like. So, basically the form would get filled in. So, it was a form on a website which would trigger a web hook. web hook is simply just a a URL that sends all of the the the data from that that form into this workflow which allows us to then use it throughout the rest of the workflow. So it the when someone filled that out, it would come in and then it would go through um well we had a filter here. Sometimes there was a head shot, so like an image that we could use for the person um and then sometimes there wasn't. if there was an image, we had to like manipulate that that data from Dropbox somehow. Um, and so we just built a different workflow, but that that's why it looks a little bit more complicated than it should, but let's just stick to the top row. So, it would come through, it would search for that contact in the database already. So, search by email. Is this person's email in our database? If it uh does that person exist? If yes, then we would just update the contact with any of the new uh information that they've provided that we don't have already. If it didn't exist, then that should then we would create that record. So we can see, let me zoom in. So this one is an update record. This is air table. This is update record. Or if they're not there, then it would create a record. So it would add that contact to air table. Then we would just merge those two uh two different um routes. Then what I was doing was pulling in all of their information. So there were when they were filling it out, they were adding in things like um uh any important notes for the talk that we should know about. They were adding in a suggestion of a title for their talk, maybe a couple of key points. I would pull that into this is a um an AI node. So, we're using GPT5 for this. And we're pulling in all of the variables from that contact record. We're pulling in the name from that person. Uh we're pulling in the topic ID, notes, if they will promote this summit or not. And then the record ID so we could attach it back into the correct air table field. So, you can see on the left hand side here, these are all variables. And then on the right hand side, it shows you what those variables look like. So this was a test in here. Uh and then we're just asking the AI to structure that data a little bit more so we could create a session that was then ready for you Simone to to begin using. So um yeah like if if AI was not available we couldn't build an automation like this to this degree. We could add people to the table for sure or the contacts to our CRM but we couldn't then structure the data in the way that we need to create a new session record so that you could come in trigger a button to then you know begin your workflow. So um yeah once all of that that so that AI was sent in as a prompt and then it would create a record in the sessions table map that sessions table back to the context contact. So let me just explain that better. So it would come into here. It would click and add a new field here. It would grab the test person. Add those that that person there and it would attach it to then the AI content mastery summit and then fill in one more field and then then it shows up in the in the interface. So then it shows up. So we're working actually from pages that look like this. It's not like an app, but it's good for an internal tool, right? Like we can then come in and this is how we then control all of the data and the content for each of the sessions inside of here.

>> And look how pretty that is, guys. All I had to do is click a button and things would just magically happen. It was amazing. [laughter]

>> Yeah. So like I mean there's so many tabs and and things going on here in this one. It's quite a detailed workflow. I wouldn't recommend starting here if you want to start here. But when we started I pulled up here as well. We just built little of the tools that we needed. So I mentioned that we took cast magic and what they were doing and built that tool. So that tool turned into like project uh sorry into it's not even here but it turned into like content automation OS where we had where we're feing feeding transcripts and then we're getting other content on the other side but we just did some for podcasts, webinars, lead magnets, memes, offers um like news posts. So, we would scrape newsletters each day and then uh find the most important news updates and then take those news updates, create a nano banana image for how we wanted it to appear on Instagram and then it will post on social media. So, like there's there's so many different things that you can do once you start understanding um what these these system and or automation and system tools can do and understanding database structure and so forth. So yeah, I hope that was somewhat valuable. Um, any questions?

>> I think this is great. It just really shows everyone like the behind the scenes and also what it actually looks like when we put it into practice because I didn't know any of that backend stuff and it was just magically happening. So yeah, I would also love to know from you guys in the audience, how are you following with this? Is this stuff that's um too advanced? Is this would you like more detail? Where would you like us to go from here? Because like to me I'm like this is super cool and you know we could keep going on this but where are you guys at? Obviously Zippy saying 100% valuable and it's so valuable like Mitch are just like h amazing but where is everybody else at? You want more detail, less detail, simplify it, go more advanced. Where are we at?

>> Yeah, while while those are coming in then also let me just show you quickly as well. So I mentioned yesterday that we save 52 hours um with the workflows that we were running in the last week. So the other thing that we're doing now for our automations which we never used to do we we didn't worry about tracking usage and so forth but now we we're tracking usage as well for all of these workflows inside of the summit system. So this $60 I mean it wasn't even this is like total that was spent as we've been building this system. I mean, that's not true. We only started adding in the usage tracking probably midway through using like building this system and it's still not complete where I would say we're still only like 60% complete in ter terms of what we want to do. But to save that 52 hours, we're spending less than $60 in in AI usage. So obviously there's development time and there's tools like N that we have to pay for which is maybe $40 a month and then we're paying Air Table and so forth. But we're not paying um huge amounts of money when when it comes to using the AI component and that's processing content for 20 plus virtual summit sessions and helping us write emails and so forth and and create images like thumbnail images. So it's not a crazy amount, right? And we now we can see this over time like what what models are going to be the most popular that we're using. We can see what workflows we're uh are using the most money. So if we have a workflow that is uh using a ton of money, we can then maybe look at that and say, okay, if we're if we use some cheaper models from up here, do we get the same output? If we do, then why not make that the default for some of these workflows? Like we don't have to be using claude for everything all the time. Um because that's the one that costs the most. Maybe we can use GPT like or Grock or GPT even GPT 4 sometimes. Um so like having this is also quite useful for when you start to scale up as well. And yeah, you can see per per brand, per summit, per workflow, per model. All of the this this data helps you build a um a a a system that allows you to make really clear decisions based on the the data what the data is showing you and which is super important.

>> Absolutely. And super interesting as well because you can just adjust as you go.

>> Yeah. All right, I'm gonna stop that share. [clears throat]

>> So Dwayne said, "The token usage dashboard is fire," which absolutely it is. I think that's super cool. I don't want to be like the person to ask, but how cool would it be to see you build something like that for us or even show like how you actually did that? How'd you do that? Or is that too technical? Can we not ask?

>> [laughter]

>> No, I mean I I have uh probably 10 videos on my personal YouTube channel that almost nobody has watched and they're just over- the-shoulder looks at how I built all of those systems. So if um yeah, just go to Mitch Assa YouTube channel, everything is there for free. A look over my shoulder as I build the database structure, as I build the NAN automations. Um, everything is there and yeah, it's it's it's super valuable if you want to get a bit of an insight into how to actually build this stuff because I I noticed a lot of the videos that are out, they show you the end result and then they show you how it works and then they say sign up for school and which is great. It's a good funnel for them to get you into their school communities. But I just I gave every I like just showed every step of the way the mistakes that I was making. Um, we do have an offer as well if you want to get the templates. It's like $50 or $60 or something, but you don't even need to. You could you could just watch over my shoulder as I'm building and build the same stuff. Um, so it's all for free. Yeah.

>> Amazing. We also have a question from Zippy as well, so I'll bring that up. But everybody jump on Mitch's channel and check that out.

>> So the question,

>> if you have different clients, would you have different NAN accounts or can you use one? I'm thinking to have different Yeah, it's a good question. There's no right or wrong way. The way that we prefer to operate is we ask the client to start their own account and then we work in their account. Um we we're not interested in trying to lock people in from uh like a business point of view. Like some some people like to operate that way. We've never really operated that way. Even years ago, I was running Facebook ads for clients and some some agencies used to um start a new account if you were working with them and then if you left them, you would lose all of that that that entire account and everything they've done. I've just never worked that way. So, yeah, if we're working with clients, we just say, "Hey, I mean, you can get sign up to being an affiliate for a lot of lot of these tools." You say, "Okay, you need to open all of your accounts. Here's my links where I get a little bit of a commission for you signing up to those um those tools." and I'm going to work inside of your account, which means you own everything. If you're not happy with what I'm doing, then you get to keep it all. Um, so yeah, that's the way that we build. One caveat to that is sometimes we because we have a lot of the um the integrations done inside of our own account and we have uh like we we've we've now we we have a way that we build that is kind of replicatable across all of the different projects that we do. Even though the projects are different like there's a certain way that we build. So, it's just easy for us to copy one workflow and then duplicate it into um the next one to build quickly and get like a prototype up for our customers and then once we go to deploy it once it's ready, we will then move it into their accounts. So, we we sometimes do that as well.

>> Super cool. Yeah, it's very exciting. Um Zippy's just said, "Makes sense. You got it where I was going with that. You're so damn smart." Yes, [laughter] you're very very very good at like all these sort of it's like a little puzzle and you're just like I'm going make this work. So yeah, thank you for sharing your wisdom with us.

>> You're welcome.

>> Any other questions?

>> Where else? Where from here, Simone?

>> Well, do you guys want to hear about tomorrow or do you want to stay on NN? I'm happy to share with you about tomorrow if you're excited. I'm excited because tomorrow's obviously in a little bit of a different direction. We're going into content creation tomorrow, which is why obviously today being systems and strategy, we wanted to go a bit deeper on on these sort of things with you, but tomorrow we're moving into content creation. So, you guys ready to hear about that? Have we got any more questions?

>> Let's go.

>> Questions coming in, we can still answer them, but yeah, let's dive into tomorrow.

>> Tomorrow is going to be super fun. We've got Ali Ma who's talking AI strategies for non- techy content creators. This was cool because he's got some super cool stuff that like I've never seen or stuff that I heard about a few years ago and it wasn't at the level where I wanted to use it and now

It's amazing. So, um, Zippy's just said, "I love the demos. More to come, right?" Of course, we can do demos as much as that. We're totally keen. [laughter] Um, yeah.

So, tomorrow we've got Ali Ma. So, that's going to be awesome. and he shows you some really cool tools that pretty much completely automate all of your content creation for things like Facebook, things like that where you can just set the set the tool to go and it'll just make the make the content for you and plug it in. Also, he does share um some apps as well which will take away from having VAS. So, he's got some really cool stuff to share in his session.

Abby Dia Ammonette is super cool. She does 30 days of content in 30 minutes, which you know, I when I first was thinking this, I was like, "Yeah, cool. Anyone can do 30 days of content in 30 minutes." But the quality of the output that she was getting out because of the way she structured her prompts, I was like, "Man, that's so cohesive." So, that's also going to be a mind-blowing session for anyone who's wanting to understand how to create better prompts that are going to create a cohesive social media plan for you that's going to achieve the outcomes you actually want for that month. So, that's a super cool session.

Renee Remick, who generates viral content in 60 seconds. He's amazing. He's the creator of Viral Sky. Um, literally fantastic guy. Has like got the most funnest channel. Like you look at his Instagram and all his videos and they're not repetitive. They don't get boring. So, he's managed to make content in a way where it doesn't feel like you're just watching the exact same thing every time. So, that's also going to be a super cool session.

And then we've also got Jonathan Ma Mast who's uh all about how to create AI content. So he also goes a bit deeper on things like your prompts and the way you're actually actually um building the things that are going to get you that output. So tomorrow is going to be super super super fun with the content creation side of things. So you guys will have to tune in for that.

Um we've just got a >> and Jonathan Master also has the it talks about how to make it sound like you as well. um which was one of the most popular >> email responses that we've got. Uh so when we asked everybody the the email like what what you're looking to learn, that was one of the most popular responses which is how do I make it sound like me? And I mean I I mean I think everybody touches on that to some degree when it comes to content creation because everybody's focusing on that. Um but yeah, Jonathan Mast especially as well.

>> Absolutely. If you guys combine what Jonathan Mast teaches with Abby Dyier um and with her system for creating the content, that's going to be a really winning formula for creating really good content that does sound like you, but also 30 days of content really really fast. So that those two there will work really well uh in conjunction with one another.

>> Absolutely.

>> So we've got a couple questions here.

>> Yeah, I can see them. They're they're for the members area. I've just screenshot those. We'll get those fixed up straight away afterwards. Thanks for that.

>> Okay, perfect. Um, Zippy is asking, "Did you hear about Open AI is no longer going to have Sora? This came out a few hours ago."

>> I didn't yet. No, I I I because of the time zone difference. I woke up, had a shower, and did a little bit of movement and logged into here. So, I've not heard about that so far. Uh, I mean curious to know is it they're just shutting it down which I think could be what they're doing because I don't know many people who use Sora and it's not that they maybe once they launched Sora and the new versions people used it for a day or two but I don't really know if they if that's the direction that um they need to go. Also, I think Open AI is probably feeling like they're getting behind from I mean that that would that's my assumption. I could be totally wrong. So, it could be like all hands on deck to get their what is what is the one thing that they want to specialize in. I think that's probably where they're at is instead of trying to do everything like Claude is now Claude for code is basically they're taking over. Um Gemini, they're now taking over image generation, right? So like CH GPT could be trying to figure out its new direction.

>> Which leads into this question here from Dwayne about thoughts on open source models and sovereign installations.

>> Uh it's a good question. I don't think I don't think there I think there's going to be a world where both open and closed uh are around for people to use. Um obviously open source has its benefits for cost and then you have your more specialized models where you have the best people in the world working to optimize and refine them and build features um to make them better. So, but in saying that, you know, almost all of the open source models are very close, if not on par with with some of the the best closed source. So, yeah, the I I think there's a world where both are used in different ways.

Um uh yeah, I I try not to go too far into like I'm not a I'm not an AI engineer. I try not to think about this stuff too much. The thing that I think about most of the time is how can I build workflows that help me or the people on my team work faster? And if that's closed source at the time, then I'll use closed source. If it's open source at the time, then I'll use open source. And um at our level, so you know, anything like small businesses, most small businesses are probably never going to uh have to worry too much about the differences in cost. I mean, I say that loosely. What what I'm getting at is for me to think about. Now, I just showed you we used $60 in um in credit in token costs to do a lot of our content that we needed to do for this summit. I could have maybe optimized the prompts to use a closed uh sorry a open- source model and the cost might be $10, but for me to save that $50, I would probably have to spend hours and hours of my time optimizing the prompts and maybe feeding in extra data and all that type of stuff. It just makes no

We've lost your sound, Mitch. something.

>> Yes, you're back.

>> I think the mic's out. One sec.

>> Okay,

>> he's back, guys. He's back.

>> What about now?

>> Yep. [laughter]

>> No idea what happened there because I didn't unplug anything. That was very weird. I think my iPhone took over just randomly. Uh so what anyway what I was saying is like I don't know where you lost me but for me to save $50 I would have to spend hours and hours which make no sense at all for the time involvement um based on you know if whatever my hourly cost is. So that's when I look at problems. If if the ai usage was $20,000 and I could potentially save $10,000 then 100% I'm going to be looking to open source. But right now it just doesn't uh make any any sense to us. We will have uh open source models on laptops that are coming out and local models on on laptops. Perplexity is bringing out a computer. So all of that is going to happen. Um so yeah, I just I I worry about the problems that I have when they're a problem, not not not beforehand.

>> Speaking of problems, can we also discuss um data privacy and regulation compliance? What are your thoughts there?

>> uh for Yeah. So, okay. So, yeah, I mean, fair enough. If you're doing some crazy stuff that requires all of that, um uh for sure. I I think you know the the businesses that most people build uh the obviously like customer data and stuff, you don't want to plug in anything any information from customers, which we we don't like it. We have some stuff that lives in Air Table but we don't give any of the AI models access to those components.

Um uh besides yeah customer information I mean how much of I of your IP unless you're building you maybe some sort of healthcare company or a rocket company or uh like those types of ones like the AI probably already has all of the data that you think is private to you anyway. So, [laughter] uh, again, like I don't see it as a problem. I'm not something that I think about.

Um, but 100% when it comes to your personal stuff, your personal information, your customer information, any of that type of stuff, 100% keep that away. But, um, if it's just like your workflows and stuff, I don't know. I I think the AI already has the the same knowledge if not more knowledge on what you do as a company which is a lot of people don't like to hear that but that's just my opinion.

>> Yeah, I love that. Thank you. Is there anything else you want to say before we start wrapping up?

I just kind of keep reiterating that um the the best thing to do is just start using these tools and and just getting a an understanding. One of the one of my favorite quotes of all time is a drop of experience is worth a bucketload of knowledge. So like you'll learn so much more by just using these tools and playing around and making mistakes. So yeah, just get in and and start applying some stuff.

>> Yeah, I love that. Thank you. And one thing as well I will say that I keep hearing from Mitch that um I think is really good to reiterate is that he is so he's he's so outcome focused. He's not caught up on the little details. It's almost like when he's thinking about building systems and designing systems, it's almost like he zooms himself out so that he's not getting caught up on, oh, should I try this system or should I try this or should I try this? He's like, okay, this is my outcome. What's going to get me there? So also just as you guys are building these things and playing with these things, keep in the back of your mind where you're actually trying to go so that you don't get caught up on little things. You know, I work with a lot of people who they'll get caught up on a detail that's not actually going to lead them to the result they want, but they're so caught up on that detail they might lose a week, you know, of of making money or getting clients or doing what they're doing. So as you are playing with all of this sort of stuff, just keep thinking, what would Mitch do? And Mitch would be thinking, how do I optimize to get the result I want? That's what Mitch would be thinking. So, [laughter] I don't know if that helps as well. But just something to think about there as well because that's also a key thing that you're really um bringing up a lot as well, which is super cool.

>> Yeah. Interesting. I appreciate that. I It is an important point. I when when I first got into this, I used to follow all of the AI news and there was always something new and you always feel like you're left behind. I don't follow AI news now. like it doesn't really matter that much if there's something new that is out. I just focus on strategy like what what are people building and what am I building, what am I trying to achieve and how can I build the foundational uh pillars behind that because then you can just easily plug in all of the newest models, all of the newest tools. It's the strategy that is the most important, not not the tools and the the latest announcements.

>> Yep. I love that. And now we've got our own hashtag. What would Mitch do? So, [laughter] we should be like tagging everything. WWMD. I love this. So, yeah, thank you guys. Thank you all for being here. Um, John has said, "Thank you for all the info and advice." Well, we will be here for the next few days going deep with you guys. So, bring your questions and we'll dive in deeper on the things that you guys actually want to know, you know.

Um, Dwayne has said, "Love that Mitch isn't obsessed with uh locking customers in. It's good business practice." Yeah, there's a level of integrity there in the way Mitch builds, which is really nice. So, yeah, I love that. I love that you guys are seeing that as well. So, I love it.

>> I appreciate that. If you if you're good at what you do, then people will continue working with you. You don't need to do uh lock them into anything. So,

>> yeah.

>> Amazing. Well, do you have any final things you want to talk about, Mitch, or should we love them and leave them for today?

>> No, we already did the final thing, remember? I have no more final final things. Let's go. Let's be let's let's finish off today and uh and then we'll be back again tomorrow with some more exciting things. And I just going to reiterate one last point. I do have a final final thing. Sorry. [laughter] If you are somebody who is someone that passively watches, which I am that person a lot as well. Just one comment this week at some point. Just send us one comment, one question. Can be in the chat here or it can be via email. Reply to an email. Just one thing. That would be great. Thank you.

>> Amazing. And we'll see you guys for content creation tomorrow. Yay.

>> Thanks for Thanks everybody. Bye.