Let's make the lives of our clients easier because that's the only way they're going to pay you more money and see whatever service you have more valuable. Essentially, we want them to vibe code your service. We want to receive a message, just a single message from any client that you have and get your service done for them without complicating delivery on you or you spending more time or hiring more people and so on. and while still keeping lifetime value up and keeping retention up and turned down and keeping more customers and referral and getting more customers out of this because they're going to love what you have and actually use it using the tools and everything that's available right now. That being said, let's jump into this video and let me show you the secrets behind how I'm doing it.
This is actually really fun to me because this by itself boosts and explodes the value that you provide to your customer to your clients by tenfold at least by making their life so much easier to use your software and that's why I call it vibe coding. So what we're going to cover today is point number one is I'm going to give you the exact framework I am using right now to make my service vibe codable if that's the word and very easy for my clients to use. And this is very much NASA inspired. I'm going to show you how.
Point number two is how I applied the this to N8N and the tech stack that I'm using right now to make sure it's production ready because we see all these videos on YouTube. They're all demos of things. It's nice to have things that are nice to have. And you see some videos that actually say that I've sold this to a client have more weight because they're practical and production ready.
And this is what we're going to talk about today. And point number three is we're going to demo and we're going to see how this is live in action when I'm going to send a WhatsApp message using my phone to one of the workflows that I have and see everything working together. That being said, there's one warning I need to make which is I want you to think in terms of a platform agnostic way because every single day we see these new AI tools coming in, new tech is coming in. But the same mindset you can approach any tech or any new tool that comes in whether it's AI today or towards something else.
This is how I'd approach it because this is how I did it in the past which is for space missions which are which is what I want to show you over here. So I've sent a couple of satellites into outer space. I had a comp space company. We secured funding.
We're in the news. This is entrepreneur middle east and so on. And also I was a speaker at many events speaking about space and the products we have and so on. And essentially this all came down to us repeating this process over and over and over again until we became masters in this process of building satellites.
And this is the exact process which is with slight modifications that I'm going to show you how I've applied this framework over here towards N810 building AI stuff AI employees AI machines and automations as well. So it has a couple of phases here. I'm going to show you the adaptation of this. Essentially it means that what's the concept we're going to do here.
This is proof of concept. what are the requirements of this? How what is the first design, the initial design of this? How are we going to test it? How are we going to make sure it works?
And how are we going to send it out to space? And this is the point that I recently hit right now, which is how can I make everything that's happening in AI automation and this has become one of my big goals or my my big missions for this year as well is how can I make everything that's happening with AI and automation because I've been part of a multi-million navigation guidance and control program. It was a 13 months program where we built control for satellites and it struck my mind that day that okay we use mat lab and similink. So I think this is flipped.
So these two are flipped anyways. But for M Simulink, it looks similar to how N8N looks over here. It's nodes. You connect those nodes and so on.
There are subtle differences where this is more for engineering applications. This is more for practical business applications. This can handle real time and essentially a a lot of other things that we can take into consideration. But what we did during that time is we built the control the full control of a satellite that's currently into outer space and is rotating around earth right above you using this tools over here which is connecting nodes looking the mathematical equations adding switches and so on.
So this is the exact same approach and it's kind of similar as well to how n we're doing it for business applications what we're doing right now. So why not tap into this multi-million dollar program that I was part of for 13 months and extract all the best practices we did there to send satellites into space to make sure we have production ready AI and automation tools for our businesses that because that can work really securely and we trust and because I've asked my clients the other day if an AI can send emails for you will you trust it completely or do you want to look do you want to be a human in the loop? loop. They told me that I want to see a draft first before sending it a couple of times before trusting it. So there's a trust problem over there as well.
So how can you make this a trustless thing where they fully trust it? So they have this capability and everything that I'm going to show you is based on surveys and everything that I've ran on my clients themselves just so I see where their head is at and how do how much they trust AI and automation and what we can build out. So, one of the things that was very clear to me is that they don't trust AI that much and they would like to see the feedback or see the results of AI before using it in their actual business use cases. That's point number one.
We can get rid of that following this approach because a satellite once you send it into outer space, it needs to operate properly because you just can't go up there and fix it. So, the same approach would be good for the current situation we are at which is a trust issue that we have. And point number two, I'm going to show you a survey what they wanted the most. So now let's begin this process.
So point the phase over here is mission concept review. And that's what we're going to start with, which is the mission concept review. And in this phase over here, and this is phase one of the framework that I'm showing you that I use as well is I want to know the client needs and my experience working with the client. So what the culture is at?
Do they use Instagram or WhatsApp? How do they believe things work? what is the misconceptions they have and so on and I asked the clients what do you need? So this was a question that I sent to my clients. Let me just show you very quickly.
It was on the community that I sent it. I said I want you to all answer the below which is imagine that you could hire an employee or person to take away the tasks in your day-to-day through a single WhatsApp message because that's me understanding they relate more to WhatsApp than Telegram and the cultural thing and so on. For what little task would you hire? The cultural part is that they want to be bosses over other people like have employees.
I wanted to related that to that as much as pos possible and you can see over here I said think of it as a task level and so on. So they gave me a huge list like I have more than 50 use cases right now or task that we can make an AI do and I had to rate each task what how is how much this is the priority will this work out or not should we do it is it worth it if there is an ROI to it or not will it save us money and so on so this is just an example of five of them which essentially I put them in a table saying that this is the client request my clients told me I want to prepare my LinkedIn content I want them to reply to my WhatsApp messages and so on the basic human task is over here for example LinkedIn content writing and scheduling post weekly that's a human task that needs to be done the machine idea would be a content creation agent the time spent doing this is two to four hours and the employee cost is around $600 to $900 per month I know a person who told me that I'm paying $700 per month one of my clients for a ghostriter to write four emails for me every single month so that's $700 saved every single month and you can write more emails of course. So this is just an idea. So you see where this all came out and this is forming the mission concept review forming the thesis over here that we're saying okay how can we make sure this is useful uh what's the mission what what are we creating over here and why are we doing this essentially and how can we vibe code all this or make it very easy for them to vibe code and use our service and actually achieve the results they want to achieve while saving them time and the money that they would have spent.
So given the niche that we are in and the needs of the business owners that I've asked over here, they want a simple system over here which is they want to build AI machines that take human input in text, voice and image and output complete business tasks in their world to save time and add profit. So that's essentially what we came out to in terms of the mission concept. What's the mission over here? What's what are we doing?
This is what we're doing over here. If we want to put this a bit more visually, we want to do this. We want them to have an input to a certain machine as a WhatsApp message and then output a completed business task while having the trust and all taken care of in the middle by having a bunch of employees. As you can see over here, this is a master slave system.
So this is essentially what they told me they want. And the core objectives if we break this down a bit more is they want to save time and repetitive and routine tasks because if you have a job, if you're a business owner and so on, you don't have time to do things. if you they want to reduce reliance on more staff. So they want to keep hiring staff because that's decreases the margins and essentially they make less profits and they rely on shady agencies for external services and they don't trust and that's a headache by itself. Who's a good person?
Do you have any referrals and so on? So this is point number two. Point number three, they want to automate critical tasks. So they reduce human error because even though the task is very simple to do like write an email there will be typos at some point like within 10,00 emails there's going to be typos but if we use AI we're going to reduce the human error so typos not may not be that much of a critical error but there are two tasks that have critical errors and then we look at keep where human is needed human in loop essentially some task for example let's say Steve Jobs or someone very important messaged you or sent you an email, you don't want AI to reply to that email, right?
So this is where human look could be important where look at the priority of this, look who's sending and so on. So these are other things that we can keep in consideration. So very important point here, we're just setting highlevel requirements and basically what's the problem definition over here. We're defining a problem.
What are we solving? Why are we building this AI machine? What are the objectives of it? And by just understanding these we can subcategorize each of this which we're going to do in the next step and say okay how can we do this at a business level at a marketing department level at a sales level and so on.
So this is very important I added it last minute but this is very very very much important. It is the AI threshold. So for every single business owner, they think I want to add AI within my business, but there's a threshold that would make them add AI or automations, let's say, within their business. Everyone wants it, but not everyone will do it unless they hit this threshold.
What do I mean by that? For example, some people if they see the ROI, let's say business owners in particular, if they see an ROI of a certain task being taken over by by an AI or automation and it's above, let's say $10,000, I'm going to save $10,000 every single month. If I do this, that hits a critical number in my head. I want to do that because my margins will be better and so on, and I'm going to keep more customers.
They're going to be happy and so on. So, that's a threshold for one business owner, which is $10,000 per month. Another business owner, maybe $10,000 per month. It's nothing for me.
I I'm fine spending that. It's not hurting me that much. So, it might be $100,000 per month to save a save time or save money on me on me or save even employees on me to get a certain task done. So, everyone has a threshold, which is the AI threshold. whereas the number that would make me actually use this machines to replace the tasks.
So everyone has different thresholds and that's very very important to take in consideration as you go and design this thing. So once you have this all cleared out the core objectives what the threshold is just like some blind sites and so on and the basic question over here and backed by real data then you can come and say okay now let's break this down even more and look at the system requirement level I've talked about this in the past but essentially I I see businesses right now of course there are HR departments finance departments and so on but in general I see businesses in three departments marketing department sales department and operations department and that's how I run my business as well. I look at these three levels and these are the three levels we're going to look into. And point number two is this is the world on machine business model that I am using as well.
Essentially you're at the core of the world over here which is your personality, your vibe and so on. You build a world around you which is your business. The value you provide to the world and then every customer or every individual or every person that comes within your world starts as a stranger then becomes a lead then becomes a customer then becomes a client then becomes a case study and goes back use those case studies to show them to more strangers to get more leads customer clients and so on. And this changes over here the transition between the stages for every individual happens by machines that you are using.
For example, here's the marketing system. Here's the conversion system. And here's another machine that's doing another type of conversion system. And here's a delivery system where you deliver the value to the client to get the result.
So this is essentially the basis that forms the subsystem requirements we're going to talk about right now. So the format I followed which is very space kind of but you can do other formats. It's fine but this is something I'm used to is essentially the system or the machine or the subsystem shall a very important word do a certain task and measured by a certain way if possible. This makes it super super clear in the future when we look back into this we say okay this machine actually achieved the requirement that we wanted it to have and this was the number that it achieved so it's a tick for example this can be your to-do list or checklist when you're checking whether this works or not then I broke this down to multiple levels so we had the master machine over here as you remember you can see the system over here as you I just showed you the master machine then we had a bunch of slaves which is the employees let's say in this case.
So the master machine at master machine level the machine shall accept which is a task input via WhatsApp. I didn't say telegram. We're not going to optimize it for anything else. We're just narrowing down the problem so much right now so we can actually deliver and solve for it.
So the machine shall accept input via WhatsApp text, voice or image. The machine shall interpret users intent accurately with a 90% success rate. So this is taking into consideration the hallucinations, misunderstanding and everything that comes with the AI and the machine shall reduce manual time spent on daily task business tasks by 70% within 60 days of deployment. So once you take this machine deploy it within your business, it should reduce the manual time spent by 70% at least using business task.
So I'm not going to go through all of them, but this essentially is how I set it up at the machine level. I have a big spreadsheet of it. I'm just showing you part of it just you have an idea how we've set this up. Then at the subsystem level as I mentioned we have marketing, sales and operations.
So for marketing I'm just going to read out an example of each marketing subsystem shall generate visuals, post thumbnails, basic design templates when required using predefined brand styles. This is one. It shall maintain a 30-day content calendar and so on. So this is just marketing subsystem requirements just to make sure we've achieved the mission of this task which is also making sure we have achieved this task over here.
This is kind of the boring part. When we used to do space missions, everyone used to say, "Oh, now we need to get into the system requirement stage where we're going to look at the requirements, fix it and redo it, go in a review and so on and so on." But for me, this is really fun because we're designing here. We're like we're putting that designer's hat on and we're designing the machine for us and by knowing this and everything we're designing over here, we know the end product that we're going to build will achieve what we wanted to do and what we want to achieve within the business and so on. So the sales system similar to this is going to monitor WhatsApp and email channels.
I could have added LinkedIn. If you have very specific channels use for your business, you can add it over here. operations. This is subsystem shall summarize podcast episodes because that's was one of the requests of uh one of the clients. I want three per day with focus on key insights, applications and potential client story content.
So this you can see videos on on on YouTube as well where people are doing it. But where does it fit within the big picture? This is over here. I'm showing you I'm showing it to you.
It's it's not a demo of one workflow, but more of I have a full business that I'm running and I need machines to take over certain things to save me time and money. And these are the one of the requirements as a demo that people show that may not be production level ready or integrate within the full system. But this is one of the requirements that I have that I want to actually use. So this is the subsystem requirements.
So these two parts over here we've just completed which is we set the mission that we want to achieve. One what are the subsystem requirements that we want to have for each part over here within our mission to actually achieve this mission objective. So this is exactly how I'm designing AI systems as well. I'm going to show you the further parts but this is exactly how I'm designing the machines the AI systems that I'm building the automations.
I want it to make sense. I don't want to just automate for the sake of automated because I'm a business owner first and then I come and leverage technology like AI and automation and all these things that are happening within the world and the tools as well to actually complement and achieve this requirements. If for example tomorrow a tool comes out and it does not help me achieve certain certain tasks over here for example one of these over here and it takes let's say more time or it does it with a less accuracy then they may not fit within my business. I don't need that tool.
I don't need that AI or technology within my business because I have something that's working. So this is a nice baseline to have. And before we move to the designing the system, I'm going to show you exactly how I've done it for my clients. But I want to just give you a hack over here because one of my clients told me the other day, okay, great.
I understand that part now. I understand the my business use cases as a business owner and how can I use AI and these tools within my business. But now when I come to think about it at a design level, I get lost. I overcomplicate things.
And maybe you're facing that as well. And I used to face that as well at the beginning where you may over complicate the design by saying okay instead of uh there's a saying let me tell you the saying is easier instead of grabbing your ear from the right over here you go all the way around your head and grab it like two times like you know like you do this so they over complicate the process and this method over here when I introduced it to them they really really loved it to sort everything out. So let me tell you what you can do. So if you have an ID in mind, you want to do this for for your business right now.
You want to do a certain thing for your business. What you can do is go on chat GPT or any LM you use and just click on the button that's over here most probably somewhere. It's going to be somewhere over here which is the record button and send it or mind dump what you want to do by reading out exactly what you've written in the past over here in these two past steps. So what I did and this is the voice note that I sent to it.
I'm going to show you the results of it. I sent to it this I want you to create a simple one-pager PRD for me put to put together my messy mind and the ideas and everything that I have in place right now in my head here's some context I want to build an AI machine that I was reading essentially the mission concept that we just wrote and the system requirements these are some requirements I pasted them at the bottom as well so sometimes I want to have humans in loop sometimes not and uh I don't understand this just organize my mind for me and get it done for me and just I want I want to How can I design this? I want to use it for design process. This is what I told it.
And then it gave me out something. I'm not going to read this out, but I'm going to show you what kind of output you can get. So, it gives you a mission statement. This is your mission statement.
This is what you want to do. This is your objectives. Amazing. This is your inputs.
You want to have WhatsApp uh based text, voice notes, and images. You want outputs being such as posts, replies, reports, schedules, and so on. Your core core business system flow. I I told it how my business flows.
Strangers to lead to customer to client to case studies. back to attract new strangers. And the AM machines move people through the cycles that I just mentioned to you as well. What's a system architecture from a big level is that you have a master machine that oversees everything and has connections to correct subsystems and the subsystems which are the slave machines that are handling individual tasks. Amazing.
Now I'm having a better idea of what I'm doing because to be honest with you, once you complete this, your brain is in creative mode and it's going to be really messy as a business owner thinking, okay, now how can I design this? what did I just talk about and you forget a bunch of things that you've written over there. This just summarizes everything really nicely for you. And then human in the loop points tools. You could use N8N for example.
You can use open AI or you can use custom LM or you can use cloud anthropic or whatever. And then you can use for the memory state for example superbase posgress air table whatever you you're more comfortable with. Then what's up cloud API and then optional you can use zap year and make so this essentially tells you the tools you need. Then how does it flow for building the system?
It gives you that machine flow. How would the machine work? So step number one, it receives a WhatsApp message. Step number two, it goes to this agent or the master machine that classifies your intent from that message and it then detects the task type whether it's content creation, whether it's marketing related, whether it's salesreated, whatever it is, and it just orchestrates that task within the system itself.
And I'm not going to go through all of this. It's over here. But essentially what I'm trying to say is that you're making or taking the load of your clients up front. So you make it very convenient and easy for them to use your your service or whatever you have.
Essentially all they need to do is tell it or tell this thing or AI or whatever you have in place the interface and tell it okay I want to get this task done and it does all this by itself and get the task done and notes. This just gives you the notes and gives you the end of the PRD. This is just a nice nice hack you can use for your business if you're a business owner and just you're lost about everything. And the final point is not the final point but let's say in the design phase there's a CDR that comes in which is critical design review and then testing it whether it shows works or not as validation and also then flight readiness and there's many other steps over here but I want to get into N8N and stop with the the screen over here but let me just explain to you this part the tagging system I use for everything and this was very very important when we were doing space missions is that you're going to have many folders you're going to have many files and you're going to have many workflows You're going to have some archived workflow.
There is many versions of certain workflows. So, how can we keep track of all of this? I'm going to show you how I'm doing it and I call it the world which is a nice way and I was lucky to get with this naming but WLD naming system and tagging. This is one way I do it and I organize everything in terms of building everything out.
So, this is the outline or the agenda or the table of contents or the way of organizing everything. The naming convention I use is I am enter the world name and then I put the level they are at and then the role and the workflow name. And the tags I put to them is world ro level and department. And let me show you an example.
So these are three levels. I'm going to discuss the levels in a bit which are the hierarchy within an AI organization. So I have this organization chart which is an AI organization chart similar to how humans have it for traditional organizational charts but I'm going to show you mine. And it's changing with time.
It's developing with time by the way based on the requirements I have based on what clients are telling me based on finding easier ways for them to actually implement this framework within their business as well. So my ultimate goal as well let me just make this point my the ultimate goal is that I don't want the business owners to depend on other people as well in terms of building their AI systems and automation systems and so on. I want them to have this mindset or side of the mindset that they can handle the task by themselves. So let's look at this part over here.
This is an example. So as I mentioned world name Cola the world is Cola. The level is level two level one or level zero. The role I have three roles.
The orchestrator which is the top role the master role. Then employee which is the agent role. And then the task which is at the task level. What's the task of this employee?
So the role is over here. Then the workflow name essentially this is a task name. I put as task name. This is the agent name.
It's either CRM agent, LinkedIn agent, whatever agent. And then orchestrator which is the master over here which I called it adami you can call it whatever you want. And the tags you can see as you can see over here is department marketing department marketing. This has no department because it's the master.
Then you have ro roll ro is a task ros orchestrator ros employee worlda world where the world and level level. Essentially this is good for me in case I want to filter them out. I want to just see for example employees only. I want to see I want to see orchestrators only and so on and different versions of them and so on.
So this is how I organize everything and it comes down to this hierarchy or this organizational chart over here. As we mentioned we have orchestrator which is the master over here and under the orchestrator it has a bunch of employees in different departments. For example an email agent uh CRM agent in different departments and LinkedIn agent these are the employees. These are the specific agents.
It can have multiple of these and then you have a task level for each agent. It has different tasks based on the requirements of the customers or clients or what does the business owner want it to have. I don't over complicate it with things that they have not asked for. It's just what have you asked for and what can take off the business use cases to reduce complexity and make this to work.
And you can see how this is very much similar to how we did it in space systems as well. It's u the naming the structure and of course by time I'm going to develop this even more as I go back into the documents that we had and how we built them out and how we named them and how essentially the best practices we had within these flowcharts and how did we present them to others and so on. So this is just an evolution but this is what we're starting with right now. And then putting this into practice, this is one which is it achieves the WhatsApp message or receive the WhatsApp message then classifies it over here.
Either it's a text or a voice needs to subscribe, transcribe it or it's an image or it's a text and then you see it go to a chief AI orchestrator which then defines okay what is the intent and this is all coming back to PRD saying what is the intent of the task that I just received through WhatsApp the message that I received. Then it saves it within a memory because that was one of the system requirements. And then it diverts it to either LinkedIn agent, CRM agent, calendar or email agent. And you can have a lot of agents over here as you go on and build the system.
And within each of these, I'm going to show you what's within each of them in terms of the functions and so on. And then it just outputs the result over here. And then the second part, as you can see, let's say we go into LinkedIn agent. We have it receiving the text that came over here and has been classified and sent to the LinkedIn employee.
And the LinkedIn employee says, "Okay, amazing. You gave me this task. Let me see what type of task this is. Is it me creating a post?
Is it me commenting on other people post? Is it me scraping for certain data on LinkedIn? What should I do?" Then it looks at the message. You tell it to create a post and say, "Okay, amazing.
I'm going to go to send it to this task over here." And this task in itself goes into this workflow which is the task creator or the person that's going to do the task and output you with the post or the business completed output task. Then this is sent back to this person and this is sent back to this person over here and this is sent back to the you which is essentially the message you send. This is the full system of how everything works. We talked about it in terms of the MCR, PDR, SR, SRR, PDR, then M, MCR, then FMRR, and then whatever comes next in terms on launch.
But now, let me show you how this looks like in a real life demo. I just explained this workflow to you and now I'm going to show you a live example of it. Essentially, how it works and how it routes the tasks. It's going to get a message from me.
Here's WhatsApp. WhatsApp is open. Admi, I don't know if you see it, but it's going to get a message from me. It's going to classify it either it's a voice note message or text message or image.
It's going to transcribe and do what's necessary in order to feed it this to this agent. Then assign the employee and the task to it and then come back and give me the results. So now for example, let's say I'm at the marketing. I'm a business owner.
I'm at a marketing event and I just met someone and I want to add this person in my CRM for later. So someone else sends him a message or even the AI agent by itself sends it a message later on or email later on. it was glad I was glad meeting you and so on. So now let's test this out and add that person in a CRM and simulate as if I'm at this networking event and so on. So first I'm going to click test workflow.
So it starts listening to for WhatsApp messages. But if when you're going to go into production, you can keep this active and it's going to work. But now let's test it and I'm going to send it a voice note right now. I just met Edam in a marketing event.
His number is 05011112223 and he told me that he's very much interested in dogs and cats for some reason and he's interested in also starting a business with me and partnering up in Cola and maybe funding us for certain activities and so on. Can you add this person in the CRM so I can contact later on? So I send this voice note and message. It should trigger and get it.
It send it to us audio message. correct transcribed it came here routing the task to the CRM agent as you can see it didn't use the calendar email and so on then it came back and let's see what's the output over here the output is name adam his phone number so let me show you if I can show you the output over here the output over here is this Adam has been successfully added with the following information and this is what I got on WhatsApp as well this is Adam he's been successfully added this is his phone number his name and his interests are dogged and cat and potential partnership is interested in starting a business and funding activities for Ciola. If you want to add anything else, let me know. So this is just adding it in a CRM and in the CRM itself, it added in air table. I have my CRM in air table and it added there as a slot.
So later on what I can do is I can send it another message over here telling it, hey can you send an email to everyone I met today in this networking event and tell them a personalized email. Just draft the emails for me and I'm going to approve it for you and then send those emails to them and it's going to do the same exact process. So nobody needs to see this. Business owners do not need to see what you're seeing right now on the screen.
What they're going to see is just this a WhatsApp message. They're going to keep on sending WhatsApp messages and getting tasks done. So this is just a short demo of showing you how everything works and you can expand this as much as you want to the task level that you want and ensure that it works for you every single scenario. Couple of things I want you to now start thinking about all your business use cases and what can you make this AI machine do?
What is causing you headaches? What is causing you a lot of time and money and you can't trust anyone anyone with? So those are tasks that immediately go at the top of the high priority list of making a machine do it since they're routinely and repetitive and so on because everyone talks that will AI take our jobs and no and so on. The thing is AI won't take people jobs it will take tasks and tasks that as you saw over here are repetitive they're routine tasks they can be done by the machine themselves.
So just start thinking about this process and start implementing the same exact framework I showed you because this is just the tip of the iceberg. what's possible with this is like I I just can say the sky is the limit of what possible what's possible with this because this is something I'm going all in on and it's how I'm doing it for my business and how you can use it for your business in order to get prompts or messages from your clients after they join and start working with you and they get the results immediately whether it's ads whatever they want if you're a marketing agency let's say if you're a coach or consultant whether it's worksheets whether it's whatever it is like you you're you just need to get creative and start using machines in your world and making them work for