Transcript
0:00What if you could build your own AI assistant?
0:02One that writes emails, books, meetings,
0:03handles customer questions,
0:05or even sells for you without being an AI expert.
0:08Now imagine being able to sell these AI agents to your clients
0:11for $5,000 or more. Just like the AI automation product that I sell
0:15that helps me generate more than $100,000 per month,
0:18all without writing a single line of code.
0:20Sound impossible? That's the power of AI agents.
0:23It's a lot easier than you think.
0:24Let me show you how. By the end of this video,
0:26you won't just have an AI assistant.
0:27You'll understand how AI agents work,
0:29when to use them and when not to use them,
0:31and how to customize or sell them as a service.
0:33This isn't just about building something cool.
0:35It's about unlocking a new skill set
0:37that will put you ahead of 99% of all the other people
0:40trying to figure out AI. So here's the plan.
0:42First, I'll demo an AI powered voice agent using Telegram.
0:45And then while we build it out from scratch,
0:47we'll talk about what AI agents actually are,
0:50why they matter, how they work,
0:52and when to use them. We'll build out your first agent,
0:54and along the way, I'll explain all of the logic that you need to know
0:57in order to imagine, build,
0:59and sell your own agents. So here's the agenda today.
1:02First, I'll do a demo of the AI powered voice assistant,
1:05and Then we'll get into the build.
1:06And while we do that build,
1:07we'll talk about why agents matter.
1:09And then we'll build out that entire Telegram AI voice agent.
1:12If you have ever felt overwhelmed by AI,
1:14feel like you're getting left behind,
1:16or if you've wasted hours trying to figure out where to start,
1:18this video is for you. Let's get started.
1:20Alright, so now
1:20let's go ahead and show you how this AI
1:22voice assistant actually works.
1:24And I'm building this out in Nathan
1:25largely for this AI agent that they have,
1:28which I'll explain in more detail in a bit.
1:30Now, I'm specifically using Telegram because I can send it voice notes,
1:33which is easy to do when you're on the go.
1:35I can also send a message and type that in as well,
1:38but I wanted the ability to send voice notes.
1:41Currently, I have the agent responding in text,
1:43but I could also have the agent respond in a voice note as well.
1:47So I'm here on my phone here.
1:48I'm gonna send an email to Steven.
1:50could you reach out to Steven Pope and see if he's available tonight,
1:54February 22nd, for a meeting to talk about YouTube
1:58and what we're gonna publish next week at 5 p m.
2:03So you can see there that we got the message back from the agent
2:06saying that they sent out the message.
2:08I can double check here in Gmail.
2:11Hey, Steven,
2:11hope this message finds you Well,
2:13are you available tonight
2:14to talk about the YouTube video that we're gonna publish next week?
2:17Looking forward to your response.
2:18So now let's just go ahead and assume he confirmed.
2:21I heard back from Steven. Can you go ahead and book that event tonight?
2:25So we got that message back,
2:26and now if we check out the calendar,
2:28we can see. I tried to book a meeting with myself tonight at 5:00
2:32to talk about the YouTube video,
2:34and you can see here on my phone
2:35that the agent was able to successfully book that.
2:37So this AI assistant only took a few minutes to build a consent emails.
2:41It can get all of your contacts and book events.
2:43Now, let's go ahead and build this out from scratch.
2:46And along the way,
2:47I'll explain the important things that you need to know
2:49about AI agents. When to use them,
2:51how they work, which ones to use.
2:53Everything that you need to know about AI agents.
2:56Now, this automation that we're reviewing today
2:58can be found in the No Code Architects community
3:01inside the classroom. When you click on this resource here,
3:03it'll download the template,
3:05and then from within 8 and N,
3:06you simply create a new workspace and import the file,
3:10and you'll have the entire automation set up just like that.
3:12But I'm also gonna build this entire automation from scratch.
3:14So let's get started. So I'm gonna start with the AI agent.
3:18Just to Talk a little bit about this agent and how it works
3:21and how to think about it.
3:23Because this agent is one of the primary reasons
3:25why I'm currently using N8N to develop some automations.
3:29They make it very straightforward to you add an AI agent
3:32that even make it so that you can chat with it
3:34without actually hooking up something like Telegram,
3:36which isn't necessarily hard.
3:37But when you want to chat with your agent just to test,
3:40it's quite nice to just be able to open up a chat and say hello.
3:43Now, we haven't configured this agent yet,
3:45so we're getting an error.
3:46But let's talk a little bit about what an agent really is.
3:49From my point of view, what an AI agent does
3:51is that it actually does some amount of thinking
3:54to determine what path to take in your automation.
3:58So you really only need an AI agent
3:59when you need your automation to think.
4:02Now, realistically,
4:03these agents really only have two functions.
4:06They monitor the chat, and then from that chat,
4:08which is just a typical conversation that you'd have with anybody,
4:11like a real assistant, it's trying to figure out, okay,
4:13what exactly does this person want to do?
4:16Like, do they want to send an email?
4:19And if the agent, through the conversation,
4:21is able to figure out that they want to send an email,
4:24it then has to figure out, okay,
4:26who is this email to? What is The subject
4:30and what is the contents of that email?
4:34So this agent is simply trying to figure out
4:36what is the tool we need to use right now.
4:40And then for that given tool,
4:42what are the different inputs that we need?
4:44Right, so we talked about what we need for email,
4:46but to book a calendar event,
4:50we'll need who is the event with and what is the date and time.
4:58And there are other types of agents as well,
5:01like Open AI's operator. This is an agent that actually runs on your PC,
5:05and then you can give it some simple instructions
5:07find and book the highest rated one day tour of Rome on Tripadvisor.
5:12And then it will actually take control of your PC
5:14and try to execute the entire task on its own.
5:17And so you can see here, there is a wide range of AI agents,
5:21those that understand specific tools and how to use those tools.
5:26And then there are other agents all the way on the other spectrum,
5:29where you give it a task
5:31and it tries to figure out everything on its own.
5:34And there really is a place for both of these agents.
5:38Really, the strength of the agent that we're gonna build today
5:41is that it's much more specific,
5:42the tools are more simple,
5:43and it's much more likely to succeed without any human intervention.
5:47These more advanced AI agents are trying to accomplish the entire task
5:52on Their own, which you can imagine is much more prone to error
5:55or they're like much more likely to make a mistake or get stuck.
5:59And then of course
6:00there's AI agents that are in between the more specific
6:04and the ones that are completely automated
6:08all on their own. And so again,
6:09the agent we're building today is definitely more specific.
6:12Where we give it specific tools
6:14and then we show it how to use those tools
6:16and that is really all that it can do.
6:18But the advantage to that is that it's much more simple
6:21and you're much more likely to get 100% success rate.
6:24Alright, so now let's explain a bit about how this AI agent works.
6:28There's a few different things that we need to hook up to this agent
6:31in order for it to work. We need to add a chat model.
6:35In this example, I'm just simply going to use open AI
6:38and I'll just go ahead with its pre built decisions here.
6:41You might actually go with GPT4O to start.
6:44This model tends to perform a bit better.
6:47And you can always downgrade it to a more cheaper model
6:50after you get it working.
6:51So the first thing that we added here was the chat model.
6:53And this is really the brain.
6:54This is what allows your agent to think on its own.
6:57If you think about a robot here
7:01where we have a head, a body and then the legs inside This head here,
7:06we definitely want to have a brain.
7:09Now, also,
7:10in order to make this agent work well,
7:13is we need a memory. As it's talking to us,
7:16it needs to be able to remember the conversations back and forth,
7:19not just remembering the last message that you sent.
7:22So in order to add memory,
7:24we're gonna come here,
7:25and we're simply going to use the window buffer memory.
7:29This is the easiest one to set up.
7:30We can go ahead and leave these details as is,
7:33and that should be everything that we need in order to test this.
7:36I'm gonna go ahead and type hello.
7:39So again, we just use this built in chat interface to start interacting.
7:43Before we've actually set up something like Telegram
7:46where we can chat with it in a typical interface that we're used to.
7:49But again, this is why I like N A N,
7:50because it gives you this chat interface
7:53so that you can go ahead and start chatting with your agent
7:55without having to set up anything else.
7:57So this is really great for testing.
7:58So I said, hello,
7:59and I said, hey,
8:00how can I assist you today?
8:01What are you able to help me with?
8:05So right now,
8:06it's just giving us some general information on how it could help.
8:09And it's really just using this chat GPT model right here
8:12to answer any of those questions.
8:14It doesn't have any additional tools to pull from In order to help us,
8:18it can really only help us as much as ChatGPT could hear.
8:22So if we were to log into ChatGPT,
8:23it would really be like conversing with Openai
8:26ChatGPT through the UI interface.
8:28So as you can imagine, that's not terribly helpful.
8:30So we're gonna add in our own tools as well.
8:33And you can also see here that the agent is also accessing its memory,
8:37so that as these messages come in,
8:39it's not forgetting the previous conversation.
8:42And the cool thing about these agents are,
8:43is that you can actually see what's actually happening in that agent.
8:47How is it making decisions?
8:48What are the previous conversations it's had?
8:52As it updates its memory and as it's chatting with ChatGPT
8:55and as it's using these tools,
8:57you can always come back here
8:59to debug how the agent is actually working
9:02and make it better. And so, again,
9:03before we move on, remember
9:04the primary objective of this agent
9:06is to converse with whomever it is
9:10via chat, and then to identify the right tool to solve the problem
9:16for the particular chat that is happening right now.
9:19And then once you identify what tool to use,
9:21whether it's to send an email,
9:23to get the contact list, or to book a meeting,
9:25it's to identify what information do we need to execute that tool.
9:28Let's have the conversation with the user
9:30to make sure we have that information,
9:32whatever it is. And then Once we have it,
9:35let's execute this tool to make it actually happen.
9:38Alright? So now
9:38let's go ahead and add the three tools that we need
9:41for this particular AI agent.
9:43And then once we're done, we'll come back
9:45and we'll replace this built in chat that we have in N8N,
9:49and we'll replace it with Telegram.
9:51And we'll actually respond back to Telegram.
9:53But until then,
9:54we'll just use the built in chat that N8N has to make things easier
9:57so we can focus on the tools.
9:59So first,
10:00let's go ahead and add a tool to actually look up our contacts.
10:03So for our contact database,
10:05I'm gonna go ahead and use Airtable.
10:07Now, I've gone ahead and added an airtable base.
10:10It's very simple. It's just a contacts list
10:12that keeps track of names and email addresses.
10:15For our AI agent, this could easily be hooked up to your CRM instead.
10:19So you'll want to create a connection to your Airtable base
10:22and then you just simply want to create a very simple table
10:26with name and email. That's all you need.
10:28And you'll be able to use a free account on Airtable for this example.
10:31Now in this case,
10:32we're using this to search the contact list for all of our contacts
10:35so we can get their email when we need to send them something.
10:38So for this, I'm gonna change the operation to search.
10:42Then we're gonna search a base type contact list.
10:45We'll Select that base. And then for the table,
10:48we'll select contact list as well.
10:50Now, I'm gonna leave the formula blank,
10:51cause we want to return all of the records in our contact list.
10:54And then I'm gonna go ahead and leave return all on.
10:56And then I should be able to come back here,
10:57I'll go ahead and save it.
10:58And then I should be able to ask the bot,
11:01please tell me who's in our contact list.
11:06And then from here, it should be able to figure that out.
11:09Now, one thing I do want to mention
11:11is that as you add these different tools,
11:13you want to give them a name that describes exactly what they do.
11:17And the reason why is because this agent is smart enough
11:21to actually look at the tools that we hook up
11:23and it's actually looking at the names that we provide
11:26in order to figure out which tool to use.
11:28So if this tool is to get our contact list,
11:32then name it, get contacts,
11:33and then just go ahead and hit rename.
11:35Because what's gonna happen here
11:36is that when the agent is actually trying to sort out
11:39what the user is asking for,
11:41it's gonna use all of the information,
11:43including the name of the tool,
11:45to figure out if that is the right tool.
11:47So as you add these, be as descriptive as possible.
11:50So you can see here that it went and used the tool
11:53and It returned the two contacts that are in my contact database.
11:57Steven Pope and Aaron Pope.
11:58So that's working just fine.
11:59Now let's go ahead and add the next tool.
12:01This time, I'll use Gmail.
12:03You'll want to create your own credentials
12:05so that you can connect up to your email.
12:07You can leave these as is.
12:08Set automatically. Resource message,
12:11operation send. Now here we have to actually send the email to someone,
12:15so N8N is making this very easy for us.
12:17In order to map this field to the agent,
12:20we just simply need to click on this little A I.
12:23Button here. And then what you're gonna see here
12:25is that this is now actually defined automatically by the model.
12:29So, again,
12:30remember what I said was, is that the main objective of this AI agent
12:33is one to figure out what the tool is to use right now.
12:38So it's gonna listen to that conversation,
12:40and it's gonna figure out, oh,
12:41they need to send an email.
12:42So it's going to look for that tool.
12:44And then by defining these specific fields here,
12:47we're telling the AI agent, hey,
12:49we need to fill in the subject.
12:51So can you monitor the conversation and extract out the email
12:55that we need to make this successful?
12:58So they make this as easy as possible?
12:59So we'll go ahead and do that for the subject as well,
13:01and then I'll go ahead and do that for the message.
13:04And What that's basically saying is
13:05we're gonna let the AI agent help figure out
13:08what these should actually be
13:10on our behalf. I'm gonna go ahead and add another option here,
13:13because I just know this is gonna be there.
13:15N A D N. Always appends this little message
13:17saying that the email was generated from N A D N.
13:20So I'm just gonna go ahead and turn that off.
13:22Now I'm gonna jump up here.
13:24I'm gonna rename this tool to send email.
13:27Remember, it's important to name the tool exactly what it does
13:32so that your agent is able to find
13:34that tool.
13:35Becomes a lot more important when you have a lot more tools here.
13:38So now we have get contacts,
13:39and now we have send email.
13:42So now one thing I am gonna do is I'm gonna come up to the agent here,
13:46and I'm just going to remind it that when it is sending emails,
13:49to make sure it looks up the email from the contacts first.
13:53So I'm going to come here to the agent,
13:54and then I'm going to add an option.
13:56We're going to add a system message.
13:57And I usually do try to keep this simple.
13:59So it starts off with, you are a helpful assistant.
14:01But I'm also going to add a directive here.
14:04Always use the Get contacts tool for finding an email address
14:13for the send email and book event tool.
14:19And then one other thing, I'm gonna Do.
14:21Because it's going to be booking events for me,
14:23and it needs to understand the date and time is.
14:26I'm going to tell the agent what time it is.
14:29Essentially, every time this agent runs,
14:32it's gonna look at this system message.
14:35And with N A N,
14:36we can actually define a dynamic message using the expressions here.
14:40If you're not familiar with this,
14:41you can have a fixed expression that will use this every single time.
14:45And you can also use the expressions here,
14:47which allows you to put in variables.
14:49So here we're gonna go ahead and say
14:52today's date is. Then I'm going to use the curly braces dollar sign now,
15:00which is going to get the current time.
15:02And then I'm simply going to dot format.
15:06And for the format, I'll use
15:09year, month,
15:11and day. And so you can see here we open this up right here.
15:15This allows you to expand it to see the full message.
15:17This is the actual system message
15:19that will be interpreted each time it's looking at a chat.
15:22You are a helpful assistant.
15:23Always use the get contacts tool for finding an email address
15:26for the send email or book event tool.
15:29And today's date is 2025 to 22.
15:33So that should be good. I'm gonna go ahead and save things here.
15:36So now we should be able to say,
15:38please send an email to Steven Pope about attending a meeting at 5 p m.
15:48Pacific on 2:22 about discussing the next YouTube video.
15:57See if he can make it. So we'll go ahead and send that message.
16:01So we're gonna see here that it already sent the message.
16:03I'm gonna jump over to my email and I'm just gonna take a look here.
16:07Looks like it did send that message.
16:09And now you're gonna notice here
16:10that it was able to send it to the correct email address,
16:13even though in this particular message,
16:16I said, send an email to Steven Pope,
16:18I didn't give the actual email address.
16:20But because previously in our chat,
16:22we had asked for all of the contacts,
16:24it was able to figure that out simply by looking at its memory.
16:28So it didn't actually have to go to the contacts to look it up again,
16:32because it had already done that.
16:33So again, I really wanna drive this home.
16:35What this agent is doing is that it's using its brain and it's memory
16:39and these tools to figure things out
16:42in order to produce a specific outcome.
16:45Now, if I had removed this memory,
16:47it wouldn't have had that,
16:48and it would have had to go to the context first,
16:51and then it would have had to send the email.
16:52Alright, so now that we have the get contacts and the send email,
16:55let's go ahead and add the next tool,
16:58which is the calendar booking.
17:00So add the calendar tool.
17:01Now you'll need to set up your credentials here.
17:03For tool description will set automatically resource event,
17:06operation create.
17:07And then we need to select which calendar.
17:09Go ahead and use S G P. Labs.
17:11Now, for the start time,
17:12it already has some values in here.
17:14Now, again,
17:15in this case, we're gonna go ahead and use the A I.
17:17Feature here, and we're gonna let the agent define the start time,
17:22because we're gonna talk about it in our conversation.
17:24So it should be able to figure out what that start time is.
17:26And we're gonna do the same for the end time as well.
17:29Now, the only thing that we do need to add here is the attendees.
17:33So let's go ahead and add the attendees.
17:35And then for this as well,
17:36we're gonna let the AI agent figure out the attendees for us
17:40based off of the conversation.
17:42And then I'm gonna come up here to the name of this tool,
17:45and we're gonna call it book event.
17:48And then we'll go ahead and rename that.
17:51So let's come back out.
17:52Let's go ahead and clean the actual workspace up just a bit.
17:56Let's go ahead and save. And then let's come back down here to our chat.
18:00So now we can say, Steven Pope said the time was good.
18:06Can you please book the YouTube discussion meeting with him for today,
18:152:22 at 5 p M.
18:19Pacific? So we'll send that into the agent.
18:22Looks like it already booked that event.
18:23Let's take a Quick, look at the calendar.
18:26We have the event here. It added Stephen Pope.
18:29But notice that we don't have a title or description.
18:32So let's come back to our workflow.
18:34Let's go into book event.
18:35Looks like we got to add a couple of other fields here.
18:37Let's add description, and let's add the summary.
18:42And again, we could type this in ourselves,
18:44but we're gonna go ahead and use this as well
18:45so that we'll let the agent figure that out,
18:47and then let's just take that message,
18:49copy and paste it again. Now,
18:52this is kind of interesting because it didn't actually send anything,
18:55because I think that it knows from its memory that it already did.
18:58So I'm just gonna type in a different command.
19:00Can you please schedule another meeting at 6 p m.
19:05On 2:22 to discuss the YouTube thumbnail with Steven Pope?
19:15So it looks like it's booking that event.
19:16Let's go over to the calendar.
19:17Now, you can see that we have the new one at six.
19:20Now we have the summary, we have the attendee,
19:23and we also have a description.
19:25So there you go. In just a few minutes,
19:26we created an entire AI agent with three different tools
19:29so that we can send emails,
19:31book events, and also scan all of our contacts.
19:33So now all we need to do is to hook up Telegram so that from our phone,
19:38we can send text messages directly into the agent.
19:40Or we can actually send in voice messages
19:43where we download the audio file,
19:45transcribe it, and then we map that into our agent
19:48and it's gonna do the work that we've already built out
19:51using these tools.
19:52And then we'll finally send a message back to Telegram.
19:55So now setting up that Telegram bot is super simple.
19:58Here I am in the web interface for Telegram.
20:01You're going to want to connect to a user called Botfather.
20:05And you're gonna Wanna look for that verification Mark
20:07so that you know you're talking with the real bot father for Telegram.
20:11Believe it or not, this is where you can start a conversation
20:14and create a new bot on Telegram.
20:16It's super simple. You're gonna say new bot,
20:19and then it's gonna ask you for a name.
20:21And I'll go with S G P. Assist.
20:25And I'll put YouTube. And then it wants a username for that bot
20:28and it gives you an example.
20:30We'll go with S G P, assist,
20:33underscore, YouTube.
20:35But it looks like I left out bot at the end.
20:37Has to be there. S G P, assist,
20:42YouTube, Bop.
20:43And then after that, it's gonna give you this access token.
20:46You just click on that, come back to N A N.
20:47And then we'll get rid of this built in chat for testing,
20:51add a new node,
20:53Telegram. We're gonna go with a trigger here on message.
20:56Then we're gonna come here,
20:58we're gonna add a New credential.
21:00We're just gonna drop that access token directly here.
21:03Save that. You should see a connection tested successfully.
21:06Then you can test the step.
21:07You can come back to Botfather.
21:10It's gonna give you a link to your assistant.
21:12You can go ahead and click on that.
21:13You can start that conversation.
21:14Even that in itself is going to send a message to N A N,
21:19which is the start conversation message here,
21:21and you can see it. So as long as you see that,
21:23you know it's working. So from here,
21:24we're gonna move this right there.
21:26Now, the first thing I like to do
21:27is to send a message right back to the user.
21:30Since the bot can take a few seconds to respond,
21:32I like to just send a message back.
21:34We'll send a text message,
21:35and you wanna pick the same credentials
21:37as you were using in the trigger.
21:39And then for the chat ID,
21:40you're gonna Wanna use the data from the initial start message.
21:43We can grab this here, drag that into our Telegram message.
21:47This is so that we can respond back to that same ID.
21:50And then for the text, I'm just gonna go ahead and put one second,
21:54please. Dot, dot, dot.
21:55And then we can test it. Come back to Telegram.
21:58We can see one second, please.
22:00You can go ahead and add a option to remove the N A N.
22:06Message that we saw right here.
22:08Test That again? One second, please.
22:10So now what we can do is we can go ahead
22:12and I'm just gonna move this over just a bit.
22:14We can add a switch
22:15so that we can have one path for when it is a text message like this,
22:20and another when it's a file.
22:22So then
22:22what we need to do is
22:23we need to move this back to the original incoming message.
22:26And then what we're gonna do is we're gonna take this text here,
22:28we're gonna drag it, and then basically
22:30what we're gonna do is we're gonna come here and we're gonna say,
22:32is string exists. If this text message exists,
22:37then we're gonna go down this main path here,
22:40and we can connect it directly into the agent.
22:42And now this agent should use the message coming in from the trigger.
22:47So we'll jump into the agent.
22:48And now, instead of saying source for prompt,
22:52we're going to change this to define below.
22:54And then we're going to pass in the text from the first trigger.
22:57We're going to come up here,
22:58go to the trigger, and we'll pass in the text
23:02that's going to automatically turn this into an expression
23:04so that it can pull in the value in the text.
23:07So from that, we should be able to come back.
23:09Let's go ahead and save it.
23:10Let's go ahead and test the workflow.
23:12Can you please give me all my contacts?
23:17So come back to Our bot here now looks like we have an error.
23:22Let's take a look at that.
23:23So what's happening here
23:24is that the message is actually for our memory here.
23:26It's basically saying, hey,
23:27we have no session ID. Originally,
23:30when you set up the window buffer memory,
23:32it has a default to look at the previous node
23:35for determining what that session ID is.
23:37And what this segment does
23:39is that it allows us to have separate conversations
23:41with different people and have a different memory for each person.
23:44So in this case here, we're gonna change this to defined below.
23:49And for the key, we're going to use the chat ID.
23:53And so what that tells the memory to do is, hey,
23:55use the chat ID for this memory,
23:58so that the memory that we store for this conversation
24:01is tied to this chat ID.
24:03So if other members or other people are chatting with this bot,
24:07they will have a different chat ID.
24:09And then, therefore,
24:10the memory that is associated with that chat will not be the same
24:13if the chat ID is different.
24:15So if I'm talking to the assistant,
24:16it's going to create one memory and one chat for me.
24:19And then if somebody else is chatting with the assistant,
24:22they will have their own chat
24:23and their own memory associated with the conversation.
24:26And then this message here just allows you to determine
24:29how many of those messages will it save and remember
24:32As it goes on. So the smaller this number is,
24:34the faster it will forget the conversation.
24:37And the bigger this is, the longer the conversation it can remember.
24:40And that's really gonna come down to context
24:43of the problem that you're trying to solve.
24:44In this case, five should be plenty.
24:46So let's come back here. Let's go ahead and save it.
24:49Let's test the workflow. Come back to our bot.
24:56Contacts.
24:57Come back to N A N. It's getting the contacts,
25:00and it's done. But the only issue is,
25:02is that it wasn't able to send that message back to us.
25:04So now we need to add another Telegram module here.
25:08Telegram. This time we're going to send a message.
25:12You'll want to set up the credentials for this new bot.
25:14For the chat ID, again,
25:16we're going to come back to the original trigger.
25:19We're gonna take that chat ID,
25:20we're gonna drop it right there.
25:21And then for the text, we're gonna go back to our agent itself,
25:25and we're gonna grab the output there,
25:27drag it into the text. And again,
25:29I'll come to the ad field and remove the N A N. Message.
25:33Come back. Save.
25:34Test the workflow. Come back to our assistant.
25:37Can you give me all my contacts?
25:39Paste that in. Come back to our agent.
25:42And now it send it back to us this time.
25:43And so there we have our contacts.
25:46So everything is working.
25:46The only thing that we can't do now is send a voice message.
25:49So let's fix that. I'm gonna come back to my agent here.
25:52I'm gonna test the workflow.
25:54And now again, from my phone,
25:55I'm gonna ask the same question.
25:57Please give me all of my contacts.
26:00Now, this time,
26:00the message is going to come in,
26:02but it's not going to go anywhere
26:03because there wasn't actually a text message.
26:06So let's take a look inside here.
26:08So you can see that there's no actual text message.
26:11So that's why I didn't progress,
26:12because this switch here is looking for a text message.
26:16In this case, we don't have that.
26:18We have a file. So
26:19now
26:20we need to create a new path that is going to transcribe that file.
26:23So I'm gonna come in here,
26:24we're going to add a route.
26:26This time, we're going to come back up here,
26:30we're going to go to the trigger.
26:31And now, this time,
26:31what we're looking for is a file ID.
26:34So I'm gonna drag this here into the other route,
26:37and then we're gonna come here,
26:38we're gonna say string exists.
26:40So in this first one, we're looking for a text message.
26:42For this one, we're looking for a file ID.
26:44We can now test this step,
26:46and we're gonna see it's now going to the second output.
26:48So now it's coming down this new path here.
26:50So now what we can do, Is we can download that file.
26:53Gonna go to Telegram. We're gonna get a file.
26:55And again, set up those credentials.
26:56With the same Telegram account.
26:59And now for the file ID, I'm gonna come back up here,
27:01go to the original trigger,
27:02and we're gonna grab this file ID.
27:04Gonna drop it right there.
27:06We want to download it. That's fine.
27:07Let's go ahead and test the step.
27:09Now we've got that file. Now all we need to do is transcribe it.
27:12I'm gonna come here, we're gonna go to whisper,
27:15which is from open AI. Grab that,
27:17transcribe a recording.
27:19Go ahead and set up your credentials for open AI.
27:21The operation is transcribe a recording.
27:23I think that's all we need.
27:25So we have that text message coming in,
27:27please give me all of my contacts.
27:29Which is correct. So now we just need to connect this up to the agent,
27:32clean this up a little bit.
27:33The formatting of the workflow.
27:36And now the only thing we need to do is to come into the agent here
27:39and update the message that we're sending the agent.
27:41Because if you remember, what we did before
27:42is we sent in the text message.
27:44But in this case, we're not going to have a text message coming in here,
27:49so this is gonna be empty.
27:50And that's also why it is red here.
27:53So instead, what we can do is we can Come into this expression,
27:55we can put two question marks.
27:57And what that's telling the system is, hey,
27:59I wanna use this if it exists,
28:01but if it does not, I wanna use something else.
28:03And the something else here is going to be up in our inputs here.
28:07We're going to go to the open AI module,
28:09where we can drag in the text from our transcription.
28:13And so what you can see here is I put dollar sign Jason,
28:15which means that it's referencing the previous module.
28:17This should work just fine,
28:18but sometimes I like to be a little bit more verbose,
28:20just like how we have here
28:22where it's actually addressing a specific previous module
28:25versus just the previous.
28:27So here we're giving a specific name for a previous module.
28:31So if we come here, we have Telegram Trigger.
28:33So this is actually accessing Telegram trigger.
28:35Whereas when it's just dollar sign JSON,
28:38this is shorthand for the previous module.
28:40And this normally works fine,
28:42but when that previous module changes,
28:44it breaks this. So I'm just gonna go ahead and be more verbose,
28:47and let's just go ahead and type this out.
28:48So I'll start with dollar sign.
28:50This is gonna pop up. We're gonna go with open AI.
28:52Dot item, dot Jason,
28:55dot text.
28:56And we can always just double confirm what it is if we go to open AI
28:59here. Text.
29:02And now it went green because we have the right value.
29:04So Now we should be able to come back to our agent.
29:08Let's go ahead and test it.
29:10And then from my phone again,
29:11give me all of my contacts.
29:13And there you can see it coming back on my phone.
29:15And then again, if we want,
29:16we can test the other tools as well using a voice message.
29:20Can you please send Steven Pope an email
29:22asking him if he can join us to talk over the new AI agent?
29:28I'd love to have a talk with him sometime this week.
29:30Just reach out and see if he's available to talk.
29:33So there is figuring everything out.
29:35And it sent the email.
29:36So here's the new email discussion on new AI agent.
29:39hope you're doing well.
29:40I love to have a conversation regarding the new AI agent
29:43sometime this week. Can you let me know your availability to talk?
29:46And there we go. Now,
29:47if you want access to today's template
29:49so that you can import that directly into N A N
29:52with everything already built out,
29:53as well as getting tech support directly from me
29:55and working with a bunch of other engaged members,
29:57make sure to jump into the no code architects community.
29:59There's a make an airtable course.
30:01All of my AI automation templates,
30:03including my content systems,
30:05social media content automation,
30:07and faceless video generators.
30:08And there's also a Business Clarity course
30:10if you're trying to build out An AI automation product.
30:13And we have calls almost every day
30:15where you can get personalized help from me.
30:16I hope to see you inside the community.
30:18Either way, I hope you enjoyed this video,
30:20and I'll see you on the next one.
Mind Map
Loading mind map…
Viral Breakdown View on GitHub →
Hook (first 3 seconds)
- Verbatim opening line: "What if you could build your own AI assistant? One that writes emails, books, meetings, handles customer questions, or even sells for you without being an AI expert."
- Hook pattern: Bold promise + rhetorical question + list of high-value outcomes
- Why it stops scroll: It targets a massive pain point (AI overwhelm/exclusion) with an irresistible "you can do this too" framing. The question format creates an open loop — the viewer needs to know the answer. The list of actions (writes emails, books meetings, sells) hits multiple desires at once.
Emotional Rhythm
- Beat 1: Curiosity — "What if you could build your own AI assistant?" Opens a possibility gap.
- Beat 2: Desire + Status — "Sell these AI agents to your clients for $5,000 or more." Triggers greed/ambition.
- Beat 3: Credibility — "$100,000 per month, all without writing a single line of code." Social proof + removes fear.
- Beat 4: Tension relief — "It's a lot easier than you think." Lowers barrier.
- Beat 5: Urgency + FOMO — "Put you ahead of 99% of all the other people trying to figure out AI." Status threat.
- Beat 6: Demo (satisfaction) — Live Telegram demo shows it working. Concrete proof.
- Beat 7: Educational payoff — "I'll explain all the logic." Viewer feels smart for staying.
- Climax moment: The live demo where the agent books the calendar event — "I heard back from Steven. Can you go ahead and book that event tonight?" → calendar shows it worked. This is the "magic" moment.
Keyword Density
| Keyword/Phrase | Count | Driver |
|---|---|---|
| AI agent(s) | 18 | Algorithmic reach (trending topic) + emotional (new, powerful) |
| Build/build out | 9 | Emotional (agency, empowerment) |
| Telegram | 5 | Algorithmic (specific tool, searchable) |
| Demo | 4 | Emotional (proof, trust) |
| Sell/selling | 4 | Emotional (greed, business value) |
| Without writing a single line of code | 2 | Emotional (relief, accessibility) |
| 99% of people | 1 | Emotional (status, FOMO) |
| $100,000 per month | 1 | Algorithmic (high-value number) + emotional (greed) |
| N8N | 5 | Algorithmic (brand search, tool-specific) |
| Think/thinking | 4 | Emotional (intellectual payoff, "you're smart") |
- Algorithmic drivers: "AI agent," "Telegram," "N8N," "sell" — these are high-search-volume terms that also signal authority.
- Emotional drivers: "Build," "demo," "99%," "without code" — these trigger agency, proof, status, and relief.
Why It Spreads
- "You can sell this for $5,000" — The video explicitly frames the skill as a revenue stream. Viewers share it with peers because it promises a path to income, not just knowledge. Line: "Now imagine being able to sell these AI agents to your clients for $5,000 or more."
- Live demo removes skepticism — The Telegram demo is concrete. Viewers see the agent actually send an email and book a calendar. This builds trust and makes the claim feel real. Line: "I can double check here in Gmail... the agent was able to successfully book that."
- "Put you ahead of 99% of people" — Status threat + FOMO is a powerful share trigger. People share to signal their own intelligence or to help friends who feel behind. Line: "Unlocking a new skill set that will put you ahead of 99% of all the other people trying to figure out AI."
- Step-by-step build + free template — The video offers a downloadable template ("No Code Architects community inside the classroom"). This reduces friction for viewers to try it themselves, increasing completion rate and shares. Line: "This automation can be found in the No Code Architects community inside the classroom... you'll have the entire automation set up just like that."
- Educational depth + clarity — The creator explains why agents work, not just how. This builds authority and makes the video bookmark-worthy. Viewers save it for later reference. Line: "What an AI agent does is that it actually does some amount of thinking to determine what path to take in your automation."
What You Can Steal
- Open with a "what if" question that lists 3–5 specific pain-solving outcomes. Don't just say "build an AI assistant." Say "one that writes emails, books meetings, handles customer questions, or even sells for you." Specificity = believability.
- Include a live, real-time demo early in the video. Show the tool working with a real task (e.g., sending an email, booking a meeting). This builds instant trust and proves the claim before the tutorial even starts.
- Give away a template or resource that lowers the barrier to action. Even if it's just a link to a template or a community, offering a "shortcut" increases the likelihood viewers will try it — and share it with others who want the same shortcut.
