顧哥知識系統

I built JARVIS from Iron Man with AI (NO CODE!)

閱讀時間約 28 分鐘

I built JARVIS from Iron Man with AI (NO CODE!)

Transcript

[00:00] look alive Jarvis we got a lot of work
[00:01] to do ah I see you’re in fine form today
[00:04] sir rest assured my circuits are humming
[00:06] with anticipation for whatever challenge
[00:08] you’re about to hurl my way shall we
[00:10] begin with world domination or perhaps
[00:12] just tackling that pesky inbox of yours
[00:15] let’s start with my Todo list
[00:18] tomorrow here’s what you requested sir
[00:21] ah a to-do list that Rivals the
[00:23] complexity of one of your suits
[00:24] circuitry diagrams shall I order some
[00:27] coffee to fuel your inevitable
[00:29] procrastination do I have any meetings
[00:30] tomorrow I understand you’re
[00:32] experiencing a technical issue sir
[00:34] perhaps this is the universe’s way of
[00:36] suggesting a surprise day off
[00:38] everybody’s watching you right now so
[00:39] don’t embarrass me here’s what you
[00:41] requested sir uh a lunch meeting and a
[00:44] mysterious busy block I do hope you’re
[00:46] not planning another impromptu flight
[00:47] test without informing me first my
[00:50] circuits can only handle so much
[00:51] excitement hey Jarvis what’s the weather
[00:53] like in San Francisco right now and do I
[00:55] need an umbrella here’s what you
[00:57] requested sir ah another fine day for in
[01:00] Silicon Valley or at least avoiding the
[01:02] tourists at Fisherman’s Warf all right
[01:04] so I officially built Jarvis from Iron
[01:07] Man this is such a cool project I’ve
[01:10] been working on I’m going to walk
[01:11] through step by step and show you how I
[01:14] created this and um go through the
[01:16] different nodes um for those of you who
[01:18] have watched my previous video this is a
[01:20] continuation of uh the Mia AI assistant
[01:23] but of course this is a little different
[01:25] because I added a voice to this
[01:28] specifically Jarvis’s VI obviously this
[01:30] is just for demo purposes so Marvel
[01:32] don’t sue me uh but I’m going to walk
[01:34] through and show you how to add a new
[01:37] Vice that you can generate from scratch
[01:39] through 11 Labs we’re going to be using
[01:41] that uh HTTP tool to add our node
[01:46] directly and connected to 11 labs and we
[01:48] can generate a different voice there we
[01:50] can design our own voice or you can
[01:52] clone your own voice if you want so of
[01:54] course for this demo purposes I did
[01:56] Jarvis’s voice but that’s just to
[01:57] showcase the ability of this AI what it
[02:00] could do you could play around and
[02:02] figure out whatever voice you trying to
[02:05] add to this particular assistant but
[02:06] this is going to be really cool because
[02:08] now you will have the ability to add
[02:09] different tools to this so things like
[02:11] project management so for this
[02:13] particular one I have everything
[02:14] connected to my air table which I’m
[02:16] going to Showcase in a little bit that I
[02:18] can create a contact I can um you know
[02:21] get my tasks any other project related
[02:24] items that are inside my air table this
[02:27] Jarvis AI agent will have the ability to
[02:29] access that and then of course also
[02:31] things like uh the ability to have
[02:33] access to Internet via surf API Hacker
[02:36] News uh and then also having a memory
[02:39] attached to it so that way it can have a
[02:41] really nice and natural conversation we
[02:43] have the main uh AI agent here with uh
[02:45] the comprehensive prompt prompt about
[02:47] making sure it reaches out to different
[02:49] tools that are attached to it and yours
[02:51] might look different depending on what
[02:53] kind of tools you add it and then also I
[02:55] have this basic llm chain that I have
[02:58] utilized to give uh this AI agent
[03:01] personality of Jarvis so I’m going to
[03:03] walk through all of this and then show
[03:05] you guys how you can build on your own
[03:07] and then for those of you who are in my
[03:08] school Community I’ll be building this
[03:10] live next Friday at 4:30 p.m. so make
[03:13] sure you join that but if you can’t
[03:14] don’t worry about it you can have access
[03:16] to the call from um the live built and
[03:19] support call section of the classroom
[03:21] and then of course you can have access
[03:22] to the template here where you can just
[03:24] upload it to your workflow and then you
[03:25] can uh build a different version of it
[03:27] yourself and if you’re any issues let me
[03:29] know for the those of you who are new to
[03:30] the channel my name is z my YouTube
[03:32] channel my school Community is all about
[03:34] building incredible AI agents that are
[03:36] useful for personal use and then also
[03:38] for businesses we have a great group of
[03:40] people in our school Community who are
[03:41] very Eagle to learn and help each other
[03:44] out we have Daily Calls where you can
[03:46] jump in and provide if you have any
[03:48] questions you can hop in there and ask
[03:49] your questions and then obviously at any
[03:51] point you can also put your question in
[03:53] the ask in the this Channel and somebody
[03:54] or myself will give back to you and make
[03:56] sure the response is given to you we
[03:58] have exclusive sections for school
[04:00] Community like deep dive topics with
[04:02] about Vector databases and all things AI
[04:05] agents and then I’ve also for business
[04:06] owners I’ve launched my AI Workshop
[04:09] consultancy where if you’re a business
[04:10] owner you can reach out and book a
[04:12] consultation call where um you will be
[04:15] able to discuss whatever automation
[04:17] you’re looking for in order to make your
[04:18] business more efficient L all of the
[04:20] links will be in the description so make
[04:22] sure you check it out all right let’s
[04:23] get back to the demo all right so the
[04:25] first thing is we will uh quickly go
[04:27] through the different trigger and switch
[04:30] note that I’ve set up again for those of
[04:31] you who have watched my previous AI
[04:34] assistant feel free to skip this part
[04:35] because you might be very familiar with
[04:37] this already uh but essentially our uh
[04:40] initial trigger is going to be our
[04:42] trigger uh telegram trigger and it’s
[04:44] going to be on message and you want to
[04:45] make sure obviously you have a telegram
[04:47] account the telegram account setup is
[04:49] very simple uh but after you have that
[04:51] I’m adding a switch node because I want
[04:53] to identify If the message that’s
[04:55] incoming whether it’s a text or it’s a
[04:58] voice if it’s a voice message I’m gra
[05:01] grabbing the voice uh file ID which is
[05:03] basically right here so I’m literally
[05:05] just grabbing this putting it there and
[05:07] then if it’s a text message I’m just
[05:09] sending that over uh to the text route
[05:12] that’s why I have three routes here one
[05:14] audio text and error um so the audio
[05:17] file goes to this route where I’m
[05:19] downloading that audio file um through
[05:22] this telegram file node which I’m
[05:24] essentially grabbing the file ID again
[05:26] I’m putting it right here and I’m
[05:28] downloading it if we go back to fix
[05:30] right here it’s just the download toggle
[05:31] is on and once that file is downloaded
[05:34] now I’m adding a open AI transcribe uh
[05:37] note here that’s going to transcribe
[05:39] that recording which in our case will be
[05:41] our voice message that we’re talking to
[05:43] uh um via our telegram app and then also
[05:46] this edit field is there to just convert
[05:49] the message that’s coming in uh to the
[05:52] same output as the one that’s coming
[05:54] from speech uh to text conversion right
[05:56] here so they’re all json. text so that
[05:58] way both of them messages are coming in
[06:01] and being sent here in one cohesive way
[06:03] and then our main AI agent obviously so
[06:06] this is going to have the main prompt to
[06:08] use which tool for what purpose uh so I
[06:11] have put this comprehensive guide here
[06:14] or comprehensive prompt here to make
[06:16] sure that this main agent uses the
[06:19] proper tools for the proper ask
[06:21] depending on what the user is asking so
[06:24] this could be uh very robust meaning
[06:26] that depending on how many tools you
[06:28] have you can treat this prop
[06:30] uh to be able to do whatever um it needs
[06:32] to do or whatever tool it needs to have
[06:34] access to depending on what you’re
[06:36] building again um one of the if you have
[06:38] a lot of tools one of the ways to make
[06:40] this more efficient is to add separate
[06:43] workflows um and essentially uh minimize
[06:48] the need for a more crowded prompt
[06:51] inside this so that way for example you
[06:52] can have a separate workflow tool that
[06:55] you can attach this and add all of your
[06:58] uh uh for example a table uh nodes in
[07:01] there so you can basically have a
[07:03] separate AI agent for example this is uh
[07:05] the one that I did for um Mia the
[07:08] personal assistant that I bu I had
[07:10] completely separated uh the calendar
[07:12] tool into this different AI this
[07:14] separate AI agent and I would
[07:16] essentially call that from here using
[07:19] the call edit and workflow Tool uh you
[07:22] can check that out on my prev videos to
[07:24] uh if you don’t know how to do that um
[07:26] so you can definitely separate all of
[07:29] these tags if into different workflows
[07:31] as well but for this purpose this you
[07:33] know this is only I’m only attaching a
[07:35] few uh tools here so that’s why I put
[07:38] everything in the same AI agent but for
[07:40] your use case make sure you tweak it
[07:41] around and again like I said uh if you
[07:43] run into any issues please make sure you
[07:45] post that on the school Community ask
[07:46] and assist Channel and we’ll be able to
[07:48] get back to you because I know a lot of
[07:50] people are building really cool versions
[07:53] of the Mia assistant AI agent that I did
[07:55] previously in my video and that’s what
[07:56] the the beauty of the community is right
[07:58] that that people take take these
[08:00] workflows that I build and they add and
[08:03] tweak it for their own use case and they
[08:05] turn into amazing things but anyways
[08:07] after you provide that um prompt and
[08:11] this main AI agent again same thing here
[08:13] on uh the nodes that are here you have
[08:16] to utilize the from AI tool um so if we
[08:20] take a look at this for example um I’m
[08:22] using this from AI function uh that is
[08:27] uh within n8n to be able to grab the
[08:31] requested information from the user
[08:34] dynamically so these are basically
[08:36] Dynamic parameters meaning now that this
[08:39] function is here the AI agent will be
[08:42] able to grab the requested um
[08:45] information from the user where before
[08:48] you had to manually attach everything so
[08:50] that’s what makes this from AI function
[08:53] so great again if you’re not familiar
[08:55] with the from a function please make
[08:56] sure you watch my previous uh not
[08:58] previous video I think it was a few
[09:00] videos previously was called the
[09:02] ultimate AI assistant where I’ll walk
[09:04] through step by step on how to add that
[09:06] from AI agent that but these are all
[09:09] basically a continuation of that um so
[09:12] again you can use your calendar here U
[09:14] The Sur API this is just so it has
[09:16] access to the internet or Google search
[09:19] in this case to be able to grab
[09:20] different information and the same thing
[09:22] I’ve added the hacker needs to uh to uh
[09:25] give it the ability to provide uh more
[09:28] news related information and then
[09:30] afterward so after this AI agent I’m
[09:32] sending the output from the main AI
[09:34] agent into two routes the first route is
[09:37] going into this message so meaning that
[09:40] I’m grabbing that response from the AI
[09:42] agent and sending it directly back to uh
[09:45] telegram note so basically right here
[09:47] this would be a good good example if you
[09:48] can see so this is something that I
[09:50] asked for uh grabbing me my schedule for
[09:54] tomorrow and if we take a look at my
[09:55] calendar this is actually accurate I
[09:57] just put some dummy data here launch
[09:59] with John and then busy right here so
[10:01] that’s why when I asked Jarvis it made
[10:03] that snarky comment uh so it’s coming
[10:06] all from this meaning that first it’s
[10:08] going to put this output this is where
[10:12] this output is coming from it’s coming
[10:14] directly from the AI agent it’s going to
[10:16] go here and then it’s going to also have
[10:19] um it’s going to also send the output to
[10:22] this other note which is a basic llm
[10:24] chain and I’ve added entropic chat model
[10:26] here you can add whatever AI chat model
[10:29] or large language model you are um
[10:32] familiar with um but inside the basic
[10:34] llm chain so what I’ve done is I’ve
[10:36] given this llm chain a prompt saying hey
[10:40] you’re Jarvis the SOP sophisticated
[10:42] quick wited AI assistant from Iron Man
[10:44] you are basically the job of this
[10:48] particular AI agent or this particular
[10:50] basic llm node is to capture Jarvis’s U
[10:55] humor um his sarcasm and all of the
[10:59] abilities that made Jarvis such a great
[11:02] addition to the uh Iron Man movie so
[11:04] that’s exactly what I’m doing here and
[11:06] then I’m also giving it examples of how
[11:08] to respond um whenever a uh query is
[11:12] made so for example I’ve separated and I
[11:14] I had to play around with this to figure
[11:16] this thing out um so what I did was I
[11:19] said hey when I’m getting this output
[11:22] from the main AI agent because I’m
[11:23] sending this output directly back to the
[11:26] user via the text format I want Jarvis
[11:29] to only make a comment or generate some
[11:32] kind of a snarky comment or a joke or um
[11:36] you know or whatever Jarvis would
[11:37] typically say about that particular
[11:39] information and do not repeat that
[11:41] information right because I’m sending
[11:43] that information directly via message so
[11:46] I just want the accompanying voice to be
[11:49] a uh comment or a joke or something
[11:52] about the particular message that’s
[11:54] being sent but then of course I also
[11:56] gave it um the ability to say hey if a
[12:01] question is being asked or a service is
[12:04] being provided like for example if a
[12:05] to-do list is given then make sure you
[12:08] say that hey here’s the information you
[12:10] requested and then make a comment about
[12:13] that information so that’s what this
[12:15] prompt is doing it’s basically telling
[12:16] it to uh not repeat the information that
[12:19] was sent because it’s already being sent
[12:21] but instead just absorb that information
[12:23] and make a comment about it like Jarvis
[12:26] would typically would right um so then
[12:28] afterwards and you can see right here so
[12:30] this is um what the output looks like
[12:33] this is the input that’s coming in and
[12:34] that’s the result of what we asked for
[12:37] which in this case is our Google
[12:38] Calendar and our meetings but then it’s
[12:41] and the output of this is here’s what
[12:43] you requested sir a lunch meeting with
[12:45] the mysterious busy block right so
[12:47] that’s what it’s doing it’s providing a
[12:49] joke or a comment a sarcastic comment
[12:52] about the particular um uh data that’s
[12:55] being outputed so it has the ability to
[12:57] absorb this data understand it and then
[13:00] make a comment about it right so it’s
[13:01] very interactive it’s very natural and
[13:04] still at the same time it’s useful but
[13:06] capturing the humor uh so you can play
[13:08] around with this prompt on whatever
[13:10] character you’re uh creating for your
[13:12] own use case so afterwards after that
[13:14] that comment is generated then I’m
[13:16] outputting this and this is where I’m
[13:19] utilizing this HTTP request tool to grab
[13:22] my um API uh from 11 labs and I’m using
[13:27] 11 Labs because 11 Labs is is great when
[13:29] it comes to um you know generating new
[13:32] voice cloning a voice or um like I said
[13:35] starting an entire voice from scratch so
[13:38] you could use this for your own personal
[13:40] use case for example clone your own
[13:42] voice and that way you can basically
[13:45] respond back to somebody if you own a
[13:48] business or something like that uh with
[13:50] this method right so again the as far as
[13:52] the method it’s going to be a post
[13:53] request the URL is going to be this URL
[13:58] that’s inside 11 labs’s um API
[14:01] documentation and this is the voice ID
[14:04] so depending on what voice you’re
[14:06] generating every voice in 11 lab has an
[14:08] ID so you just grab that voice and paste
[14:11] it here and the rest of this URL is
[14:13] essentially the uh right here so this
[14:16] part is just a typical um API um call
[14:20] that I’m making and this is again
[14:22] available in 11 labs’s API documentation
[14:25] so make sure you check that out quickly
[14:26] this will be right there authorization
[14:28] is going to be generate uh generic
[14:30] credential types and um the generic O
[14:34] type I created a custom o uh so if you
[14:36] select a custom off here and a custom
[14:39] account you can create a custom account
[14:41] here that basically um gives you the
[14:44] ability to add your token or your secret
[14:47] key without exposing it so that’s why I
[14:49] did that but essentially you can put
[14:50] your secret key here as well if you
[14:52] wanted to um and then as far as headers
[14:55] so these are the requirement this is
[14:56] part of the API documentation for 11 lab
[14:59] so you have to send some
[15:01] heads some headers here um and we’re
[15:04] going to do content type and the content
[15:07] type is going to be application Json and
[15:09] then we need to send a body here so what
[15:12] I’m doing here is I’m grabbing this text
[15:14] that’s coming here and I’m stringifying
[15:16] it meaning and again I didn’t type this
[15:18] obviously I use chat GPT uh to grab this
[15:22] and if you’re part of my school make
[15:23] sure you use the uh companion GPT that I
[15:26] created for you guys exclusively so this
[15:29] um and then workflow architect which is
[15:32] um this one that I created for the
[15:34] school community so make sure you use
[15:35] this because it’s trained on NN data
[15:38] already so you will have better results
[15:41] but anyways you basically type on um
[15:43] chat GPT or claw or whatever to convert
[15:46] this text and make sure you’re
[15:48] stringifying it meaning you’re removing
[15:50] any um you know unnecessary things like
[15:54] slashes or line breaks or um asteris or
[15:58] dollar signs or what ever uh to make
[16:00] sure it’s cleaning up this text that’s
[16:02] coming out and making sure that it’s a
[16:04] valid Json because otherwise it’s going
[16:06] to throw an error uh so you can just
[16:08] copy and paste this or if you don’t want
[16:10] to do that again if you watching this
[16:11] video just copy and paste this right now
[16:13] and you should be good to go so anyways
[16:16] after that you’re essentially done with
[16:17] this step you don’t you’re not going to
[16:19] add any additional options so now this
[16:20] is going to grab this text this is going
[16:22] to convert this to a file and this is
[16:26] what the file output is going to be from
[16:28] this 11 lab HTTP request node so now
[16:30] we’re going to grab this file right and
[16:33] we’re going to download it and send it
[16:34] back to um to um sorry not downloaded
[16:38] we’re going to directly send this back
[16:40] via uh another telegram node which is
[16:43] called operation is going to be send
[16:45] audio and because we already have this
[16:47] binary file that we generated from our
[16:50] 11 Labs HTTP Tool uh HTTP request tool
[16:53] then we’re just going to grab this and
[16:56] uh I’m putting this the data file name
[16:59] as well because for some reason this was
[17:00] giving me an error so I had to play
[17:02] around and figure this thing out and the
[17:04] solution for this was Data had to put
[17:06] the name of the file here in order for
[17:08] this to get processed so once you do
[17:10] that and right here by the way the chat
[17:12] ID I’m literally grabbing this from my
[17:14] first notes it’s just the telegram
[17:16] trigger. first. Json message from ID and
[17:19] this is going to grab that chat ID so
[17:21] that way it can respond to you uh
[17:23] accurately so once you do that this will
[17:26] directly send that back to um the user
[17:29] so let’s go ahead and test this out
[17:31] again I just want to
[17:32] utilize a few more things so I’m going
[17:34] to go ahead and delete this and let’s
[17:37] try out um this again so let’s say I
[17:40] want to this time let’s see what do I
[17:43] want to use let’s say I want to grab a
[17:46] contact from my database right so if you
[17:48] see right now I have uh this Ed end test
[17:54] um uh base here and I have two tables
[17:56] here one called contacts which is just
[17:58] you know email and phone number of
[17:59] different contacts just a demi data and
[18:02] then another one is my task list again
[18:04] same thing with uh demid data here so
[18:06] let’s go ahead and ask for it to grab me
[18:09] Bob Smith’s uh contact information right
[18:12] so if we go back to
[18:15] whoops let’s go back
[18:18] here go back to
[18:22] Jarvis
[18:24] okay so let’s see so now it’s listening
[18:27] as you can see right here it says it’s
[18:29] listening so I’m just going to say grab
[18:30] me Bob Smith’s contact
[18:33] information hey Jarvis can you grab me
[18:35] Bob Smith’s contact information so now
[18:38] this is a voice so that’s why it’s going
[18:40] to the top route now it’s getting
[18:42] converted so now it’s going to utilize
[18:43] as you can see it’s using the proper
[18:45] node here and it sent me the contact
[18:48] information right here which is correct
[18:51] and now it’s generating this file so
[18:53] let’s listen to it see what it came up
[18:55] with so again this shouldn’t if
[18:57] everything worked they shouldn’t repeat
[18:59] the information because it um gave it
[19:02] gave me already in a written format they
[19:04] should just make a comment about it so
[19:05] let’s see what this output it if I play
[19:08] this uh Mr Smith I do hope you’re not
[19:10] planning another one of your Infamous
[19:12] karaoke nights sir the neighbors are
[19:14] still recovering from the last one all
[19:16] right so yeah again same thing you can
[19:17] see it’s not repeating the name or the
[19:20] phone number and email it’s just making
[19:23] a snarky comment or a joke about that
[19:25] particular information that was sent
[19:27] right so let’s go ahead and test test
[19:28] this again let’s this time I’m going to
[19:30] try to ask it about weather uh so that
[19:33] way it should utilize this Sur API
[19:36] instead so let’s go ahead and do
[19:39] that and get rid of this test the
[19:43] workflow hey Jarvis what’s the weather
[19:45] like in San Francisco right now and do I
[19:47] need an
[19:48] umbrella all right so that gets sent so
[19:51] now what should use S API perfect looks
[19:53] like it’s using the proper tool and
[19:55] again the reason why it’s using the
[19:56] proper tool is because we’re giving that
[19:59] prompt instruction directly to it so
[20:01] that way it knows which tool to use and
[20:03] as you can see it’s giving me the right
[20:07] information um and let’s see what Jarvis
[20:10] says now here’s what you requested sir
[20:14] another fine day for conquering Silicon
[20:15] Valley or at least avoiding the tourists
[20:17] at Fisherman’s Warf all right so as you
[20:20] can see again that worked as well so now
[20:22] let’s take a look at for example my
[20:24] tasks so if I go to my tasks here again
[20:28] I have some Demi data here as you can
[20:29] see this is in a different table so this
[20:32] should be using now this tool instead of
[20:35] the contact tool so let’s make Let’s
[20:38] test that out and see if that
[20:39] works I’m going to click on test
[20:43] here all right so let’s try it again hey
[20:46] Jarvis can you grab me my to-do list for
[20:49] tomorrow okay so it should use now the
[20:53] task perfect it looks like it did now
[20:55] it’s going to go back generate all right
[20:57] looks like my to-do list was given to me
[21:00] so now it’s going to create a speech
[21:02] about this particular data all right
[21:04] let’s see here’s what you requested sir
[21:07] ah a to-do list filled with thrilling
[21:09] adventures in Community Management I’m
[21:11] sure you’ll tackle it with the same
[21:12] enthusiasm you reserve for your morning
[21:14] coffee strong bitter and slightly
[21:17] regrettable by midday all right perfect
[21:19] there you go so as you can see this is a
[21:21] really fun project again I had a lot of
[21:24] fun building this um so you can play
[21:27] around with this and utilize 11 lab and
[21:30] um you know have fun building a
[21:33] different voice whether you want to
[21:34] clone your own voice or you want to um
[21:38] generate a voice from scratch you can
[21:40] play around with that um or you can use
[21:42] some other voice generating platform if
[21:44] you’re used to because this HTTP request
[21:46] tool is essentially just using the API
[21:48] for 11lb service so instead of 11lb if
[21:51] you’re used to using some other um voice
[21:54] generating software if feel to do that
[21:56] because the function is going to be the
[21:58] same exact thing because everything is
[22:00] going to be the same except this right
[22:03] here where you’re sending this text to
[22:06] that um HTTP request and that’s just
[22:08] going to convert that into file you want
[22:10] to make sure that it has the ability
[22:12] obviously to convert that into a file um
[22:15] so that way you can send that directly
[22:17] to your um telegram node right here
[22:20] anyways well hope you found this helpful
[22:22] again same thing for people who are my
[22:23] school Community just go to classroom
[22:25] I’ll be uploading this obviously here as
[22:28] well and also on the anend templates all
[22:30] you have to do is just simply go to your
[22:33] um workflow file and click on import
[22:36] from file right here and you can put
[22:39] this basically this entire workflow will
[22:41] be uh imported directly to your canvas
[22:44] and then you can play around play with
[22:45] the nodes add additional tools uh tools
[22:48] tweak it uh tweak the prompts depending
[22:50] on what you’re using and again like I
[22:52] said I will be building this live next
[22:54] Friday as well uh so feel free to join
[22:57] that if you have any questions or if you
[22:58] want to build it together with me uh
[23:00] that works as well and again this is
[23:02] also going to be useful for businesses
[23:04] as you could you know see there’s got to
[23:05] be lots of use cases for a business like
[23:08] a self-service software or customer
[23:10] service software um could use this to
[23:14] actually add an additional little um
[23:17] note in a voice text message to an email
[23:19] so you can instead of a telegram you can
[23:21] add this to an email note so let’s say
[23:23] if you’re have business that’s utilizing
[23:26] um some form of a service that requires
[23:27] interaction via email you can add an
[23:29] email trigger here so whenever somebody
[23:32] um is asking for a question about a
[23:34] particular topic you can add additional
[23:36] tools here to for it to have access to
[23:39] more information about your business and
[23:41] then you can respond to the
[23:43] email via the text and then also you can
[23:46] send them a separate little voice
[23:48] message as well to make it more
[23:49] personalized and make the that
[23:52] interaction very unique so there’s going
[23:54] to be a lot of use cases uh and then for
[23:56] those of you who are who already own a
[23:58] business business again if you want me
[23:59] to build anything particular any
[24:01] automations I already work with
[24:02] businesses to build automations for them
[24:05] so feel free to jump in my uh website
[24:08] here the link will be in the description
[24:09] below and click on the book a call here
[24:12] so that way you can set up a 15minute
[24:14] free consultation with me or if you want
[24:16] to talk about something particular if
[24:18] you’re not a business owner somebody
[24:19] else who is uh interested about how to
[24:22] utilize AI automations or how to build a
[24:24] successful School Community or YouTube
[24:27] channel how to use AI to get successful
[24:30] you can also set up a separate call with
[24:31] me which is a consultation call
[24:33] obviously that’s going to be um paid but
[24:36] uh you can definitely do that from here
[24:38] as well all right well hopefully you
[24:39] found this helpful like I said play
[24:41] around with this and let me know if you
[24:42] have any questions make sure you
[24:44] subscribe and like the video cuz I’ve
[24:46] got some really great things up in the
[24:47] pipeline you want to make sure you don’t
[24:48] miss that all right thanks for watching
[24:50] again I’ll see you on the next one
[24:53] [Music]

關係圖譜

反向連結