Make Automation and ChatGPT With OpenAI Beginner's Guide To AI Automation
Let's build with Zapier and AI (100+ videos)📅 2023-08-23
Transcript
[00:00] welcome back to work here for AI we do daily artificial intelligence videos for your personal and business life in today's video we're going to be checking out the platform of make and doing a full comprehensive tutorial everything you need to know about doing make with openai and the platform itself this has been a highly requested tutorial so make sure to give a like right off the bat that shows me that we want more tutorials on this platform but for now let's go ahead and jump in and see what we can do with make how we can understand Integrations when it comes to open AI with chant gbt and much much more from my experience on this platform
[00:30] I believe one of the biggest factors and differences between this and other automation platforms like zapier pipedream is the cost associated with this platform but also it is a little bit more complex when it comes to doing certain functions due to the fact of limited immigration that can be found natively on the platform so right off the bat this is a lot more cost effective than zapier so if you're looking for a more cost effective solution you should probably use make but when using make understand that there is going to be less natively integrated API less natively integrated
[01:03] software so you have to require a lot of web hooks when using this platform if you want more information on this as you see here we can come down here with um this nice little diagram here giving all the information that is required and as you see what I said before only a thousand apps are natively integrated to this platform comparative to 5000 apps on zapier this isn't a big issue if you know how to integrate the other apps just using the API documentation if you don't though that is where zapier becomes a little bit more effective when
[01:33] dealing with more native integration alright so I went ahead and jumped over to our dashboard here if you don't have a make account you can go ahead and start for free we have our link down there disclaimer it is an affiliate link so you can go ahead and use it if you want to support here at webcaf AI but for now let's go ahead and jump through this entire platform it's dashboard and everything you need to understand we're going to show you how to integrate open AI onto the platform and we're going to go ahead and do a flow together or in this context they call them scenarios so right off the bat what you'll see when you go go to your dashboard here is going to be your operations up here so
[02:03] essentially how much either data is being transferred or operations in use as I said before operations in this context is like task on zapier it's their way of finding whether you're overusing in a plan so the way in operation would uh essentially be controlled is that when you do a flow it would be x amount of operations associated with the amount of tasks you operated or automated in that flow the next tab here we have is teams so this is where you'll be able to add multiple teams I think this is a pretty powerful feature do the fact for zapier I believe
[02:34] the minimum for teams is like 130 a month so you get access to teams right off the bat on a free platform once I know this is currently the free plan so I want to make that obvious as well so we are obviously getting access to limited amount of features but I think that'd be a great use case if you plan on just trying the platform out for free as well as you see here for users it has us here at webcam AI and as I said before in order to get multiple users on an account you need to do a teams account zapier which is much more cost much more costly compared to here subscriptions this takes you to your plans essentially
[03:06] so if you want to raise your plans here this is when you do it and everything else here is not really relevant for what we want to do here so let's go ahead and check out everything to the left here and understand what is going on over here so teams is just a more in-depth view of the teams tab that we saw earlier scenarios this is going to be essentially your workflows that you're working out here and what's great here is that we have the ability to categorize workflows by adding a folder here so I can just do test name this would be a great way to categorize if you have clients or there are specific
[03:38] operations you want in a specific folder such as you know maybe a marketing folder maybe a customer support folder and so on so you want to organize your flows a little bit more here you can do active scenarios in active scenarios and then Concepts I'm assuming in this context is more just like you messing around seeing what you can do with the API and stuff like that coming over here we have templates so this is kind of showing example workflows that exist within the community and essentially you can you know get ideas from these as you probably already know a lot of these
[04:09] workflows are typically pretty simple but they're a great resource if you were not sure on how to work a certain app as you see here we can search app names up here um you know such as Shopify and then it would show all Associated templates that it currently has and that it's readily available into your project so for example if I want to do this one I can click here and I can say start guided setup and then create new scenario from template and then obviously once I have the scenario here I would have to essentially give access and connect my
[04:40] accounts api-wise which we'll get into but first let's just go ahead and just walk through the side section here so the next section here is connections this is going to show all of your natively integrated connections so right now I have printful but essentially this is where you would have you know your Google Docs set up your Gmail setup this is essentially where we can access the apis on those apps because this is their way of saying to the API documentation hey they're authorized you can you know send an email create a draft and do this stuff a lot easier than for every single
[05:11] one of the requests would be a web hook a post a Git it kind of circumnavigates that and essentially allows us to access the API directly with authorization our next section here is web hooks this is going to be a big deal specifically this platform more than zapier through the fact that this platform isn't as integrated across the board here so you just get really comfortable with using web hooks web essentially is a you know a link they either can send data to or get data from and this allows you to do
[05:41] certain functions and different softwares all these automation companies are really is just using um essentially user-friendly Python and user-friendly calls between software like this made the UI look really good and it allows us to natively integrate it so we don't have to always call and authorize any type of function for here though this is gonna be important area to know because the fact that you will be using web hooks a lot if you decide to use make as your main automation software
[06:11] this essentially is where you'll create public and private keys and you can kind of store the data here that's pretty cool you can also add your device to your make account I thought this was pretty cool because I'm pretty sure you can't do this in zapier if you wanted to but for now you can do that with make you have data storage but as you'll see there are limitations to the data storage so I believe with a free plan you can only store up to one megabyte or one megabit of data currently so keep that in mind and then we have data structures this is a little bit more
[06:42] advanced this has to do more with you know Json XML CSV so as you see here it's most commonly used for sterilizing or parsing data formats such as Json XML CSV so essentially you receive data from a web hook uh and the data is unorganized or not how you want it structured for the preceding steps this is where you can set up a data structure now what I want to say here though is with chat gbt and openai you're able to do a lot of that anyways using the AI so you know searching through data and stuff like
[07:12] that so this has become not obsolete but it you know Works in its own function then we got resource Hub so you know essentially an area where you can ask questions get more information about certain Integrations and so on now if you go to my apps this is where you can build out custom apps here as you see here so this is a little bit more complex I don't really see anyone using this unless you know a client request building out a custom map and as you see we got examples of possible apps you can create slash connect to
[07:44] so now we got notifications but as you see here we have one called Zone US1 you might be asking yourself okay why is it called that essentially what make allows you to do that is more cost effective than I've seen in zapier is you can create multiple organizations so for example zoneus want to get notification specific to that organization I will say that is probably one of the bigger selling points other than the costs associated with zapier is how much um openness it gives to having a lot of
[08:15] people on your platform so if you had like a team of five people using a platform like this you'd be able to do it at very cheap and be still very organized all right enough of me giving you an overview of the platform let's go ahead and walk through a scenario together first though let's go ahead and connect chat gbt and open AI into this platform I also got to connect some other apps into this platform natively so we can start messing around and building out a flow together in understanding the intricacies when it comes to building out a scenario on make I'm gonna go ahead and come up here to
[08:46] create new scenario here and let's go ahead and start connecting some of our API connections here so to begin we're just going to go ahead and click this plus button here I'm going to come down to search applications I'm gonna put in open AI and then they have it as Chad gbt Whisperer let's see if they have a native one I don't believe they do have a native okay no so it's all under one uh connection here um let's just say from now we're just gonna say generate image right even though that doesn't make sense as a trigger we're gonna first just let's just create a connection with uh make here so I'm gonna say create connection
[09:16] as as you see here we need an APA API key and then you don't need a organization ID from my understanding here so let me go ahead and show you how to get your API key so I'm going to go ahead and Link this down in the description down below but essentially you're going to want to go to open AI developer side and we're going to go ahead and sign up here so when you click sign up you're going to be able to essentially use one of these as a login or just enter your email address so first create a open AI account alright so I logged into our
[09:46] openai account here at webcafe AI the next thing you want to do once you've created your account here is we're going to go ahead and go to few API Keys now I'm going to go ahead and create a secret key here together don't worry I'm gonna delete it after the tutorial so don't try to take it and expect it to work so I'm gonna go ahead and just call this one make create a secret key here once we have our secret keyboard go ahead and copy it there we go API key copied let's go back to make here and let's go to create connection put our API key in and hit save
[10:18] foreign perfect so as you see here we got our connection here and as you saw earlier we could rename it so if you want to rename it all you gotta do is come to connections and we could just say make key or API key and then this can be help of organization so if you have a ton of connections maybe for clients you can you know add like a little bit of a square bracket here and be like a client a so they're running their automations in your back end you know you can organize uh your
[10:49] connections like that a little bit I will say it seems like the connections found on zapier is a little bit more organized better than the once here this one seems a little jumbly but from here though we can go back to our scenario and we've successfully connected open AI to uh make here so let's go ahead and show what different functions here I'm going to go ahead and connect one other uh software here I'm gonna do Gmail so I'm gonna go ahead add a module here and let's go ahead and do Search application we're gonna do Gmail perfect and you know I'm just gonna do
[11:22] random action right now just so I can create the connection I'm going to create the connection so it looks like to create a connection to Gmail all we gotta do is sign in with Google as you see here we're going to do our courses account it's going to ask us to say allow we're going to allow it and let's see if it works perfect it works as you see some connections are easier than others obviously Google is huge so they're going to make your API connections very simple such as logging in um relative to open AI but all open AI is is a key that is actually pretty important too as well because of the fact that since we have Associated a key
[11:52] with make we now will know the amount of usage and costs associated with that key so that really helps you when you scale and maybe need to make multiple keys for XYZ reason such as maybe I make a key specific for XYZ clients you understand the cast associated with that individual client but for now we have the two apps we're going to go ahead and use today so I can kind of show you all the functionalities of make when it comes to creating a scenario example we're going to do today essentially is we went ahead and took our main company account here we sent it to our courses account here
[12:22] and you know as a pseudo email so we can have data to play around with but we're just gonna assume that we got an email from a potential lead that says interested in a animation for a restaurant hey wondering what services you offer for customer service and best regards Corbin Brown I guess I'm talking to myself but here is the email we're going to use as data so I'm going to show everything is possible on this make platform so let's go ahead and proceed here and we're going to go ahead and change our trigger it is not going to be open AI we're going to go ahead and change this to a Gmail I'm going to say watch emails triggers when a new email is received to
[12:53] be processed according to specific criteria and then as you see here it's going to give us a bunch of different options here and before we jump in any further here I want to showcase a couple of things you should know about the uh underlying UI that you see first off you'll notice that I mean when I had took the initiative to rename it I'm going to rename this to email response lead so that's where you rename the underlying scenario come down here this is how you turn it on or off come down here this is essentially the schedule setting so what this means in
[13:24] this context is that every 15 minutes it's going to check uh and listen for a new email coming in let's go ahead and proceed to the controls save is Big here for some reason I'm not too sure why but if you use zapier or other platforms their save settings are more cloud-based so it's like working on Google Docs where it automatically saves for make though you gotta hit that save button so if you're working on a long complex project make sure you hit that save button if you don't and your browser crashes your computer crashes
[13:55] you're gonna lose all your work keep that in mind now most these scenario settings aren't too important I would say the probably the most important one though for a client would be data confidential you'd want to enable this in the context that the data that has been passed around you wouldn't really want make to log any of that data this may cause issues with debugging but would be important in the context of confidentiality one thing I want to point out as well is when you do a Automation and use open Ai and use openai's API key they have stated as a company that they don't track the
[14:25] messaging and data that's been passed through the API key so to clarify a bit on that if you're using regular chat gbt UI like just a chat bot like the stuff that came out in November they do track every single message that's found on the chat gbt front end but if you use the open AI API key they have stated that they don't track the data this is on their play this is to incentivize large companies to use the API key to ensure that their data is secure so the next feature we have here is notes I can go ahead and right click this say add a no and then you know maybe you want to tell
[14:56] myself uh this is for potential leads this is a great tool to use in the context that you have multiple people in your team and multiple people working on a potential flow here and that's going to make it more obvious for them to maybe not mess with a certain incoming parameter or outcoming out parameter if you get messed up here and you're going through the UI here and it gets a little bit crazy here when we add more blocks you can hit this button here to Auto align reset the frame on your underlying computer and moving from here we have
[15:27] flow control tools and text parser so these are kind of think of these of if you use zappy before these are like the buy zapier apps such as formatter web parser and stuff like that this is kind of where they centralize it they call them you know flow control tools and text parser so this is kind of where they're in they're natively built apps that they have on their back end and you can use in their flows we're gonna go ahead and use some of those today so we can kind of show you how those work and then finally we kind of have our favorite apps that we like to use such as you know open Ai and Gmail so the
[15:57] flow we're gonna be building out today or scenario in this context essentially is we're going to be looking at all emails coming into our Gmail box here we're going to identify with a filter and gbt sentiment analysis on whether it is a lead or misc type email so whether essentially is this a potential lead or is this just a random email for maybe a customer service situation and then from there we're going to go ahead and create a response contextually to that email coming in and then we're going to set it up as a draft to be sent to that
[16:28] potential lead that didn't make a lot of sense don't worry we're going to walk through it together let's go ahead and start off with our first step here which essentially is going to be identifying uh what type of emails we want to be looking for now in theory I could set up a Gmail filter here and then look for a specific syntax so in that context maybe with every single one of your leads it comes from a form or it comes in a way that within that lead email there's a specific piece of data that can be found on every single one of those emails so for example when just say we had a
[16:58] webflow form where we had a Shopify form and within the subject line on every single lead email or within one of the form variables it essentially had you know client lead or it had somewhere in the footer like you know something like schedule a call like these type of terms that maybe could only be found in incoming emails you could use that as a syntax but let's assume that we don't have something like that and it's really just random where the subject line is in the body could be
[17:30] different on every single lead so we're going to go ahead and use AI in order to solve this so we're going to do a simple filter here and we're going to do all emails we do all unread emails because we're assuming we're getting it in now I will say this would be operationally heavy cost wise in the sense of a lot of operation tasks would be used so you know it'd be in your best interest to use a Gmail filter but I think it'd be really cool to show using AI in this sense to Cipher for you so we're going to do only unread emails and essentially we're going to do maximum results of one and as you see here some have the option
[18:02] to show advanced settings so if you're looking for very specific stuff here such as search phrase and all this different information that's kind of where you input that stuff but for now we're gonna go ahead and hit OK we're going to grab a folder folder in this context would be all the information here so for us we can just put inbox and then I'm gonna go ahead and make sure that I put this as unread because it wouldn't show up as a data point if I click that so I'm going to do that and then I'm going to hit OK and we're gonna do all emos
[18:32] or actually let's go ahead and just select it but in theory you would want to do from now on and then when you turn on and then keep proceeding right so then here is the email we're gonna deal with this is going to be the data point we're dealing with here so perfect we got our email okay and then let's go ahead and set up our first uh situation here so let's learn how to do a filter and we're gonna learn a bunch of stuff when it comes to chat gbt so we're gonna go ahead and do open AI here we're going to go ahead and choose create completion and we're going to go to do a uh here we go create
[19:03] prompt completion let's go to move this over here so I get too crazy here and then the model you have the choice to choose here now what's great is uh for this kind of context we use 3.5 one thing I want to point out is that I don't have access to gbt4 yet four specifically make it seems like so keep that in mind for us we're going to use obviously Tech DaVinci 003 and then we're going to go ahead and do some prompt structure here now what you'll notice is we do have a little bit of information and a little bit of ability
[19:33] when it comes to make but what is the one big thing we're missing here if you're familiar with my channel you already know what I'm gonna say it's a Memory key and that's huge a memory key is going to what essentially allows us to ensure consistent outputs so just off the fact that we don't have a memory key is already you know you're hurting yourself a little bit but for now there's still stuff we can do with chat gbt and this we're going to say based on this email do semicolon actually we're going to do
[20:04] is we're actually set this up as the filter or the sentiment and then we're going to go ahead and provide the email data here so we're going to say subject line I'm going to do semicolon parentheses I'm going to do body let's make one parentheses I'm going to provide the underlying data here so This Is How They format data here and if you're familiar with zapier obviously it doesn't look as nice UI wise let's go ahead and just get all the information we need here so we're going to get the subject we're going to go ahead and get the text content and there we go we're going to
[20:36] say generate a yes we do semicolon Yes or No semicolon no on whether this is a potential lead interested in our services now one thing we got to add here is a little bit of context the context we just received
[21:06] an email we are and AI automation agency I'm gonna go ahead and hit OK here and then let's go and test this action to test the action here which is another thing that's kind of different from zapier is we're going to right click the underlying module here we're going to say run this module only okay so I was having some slight issues and then I realized I wasn't on the right method so you got to choose the create a chat completion method gbt models this is
[21:38] their way of essentially and as you see here this is a better explanation of it but this is their way of essentially grabbing either gbt 3.5 or gbt through four as you see here if I deselect the map here I can choose to be 3.5 and 4 and you have the choice here so I'm gonna go ahead and hit map here we're going to continue using 3.5 and then essentially from there though we would choose the role of assistant this is essentially going to allow us to talk to it and it's going to respond in a helpful manner then we provided the information here so we just received an email we are on a
[22:09] automation agency based on this email provide the context of the subject and the body yes or no on whether it's a potential lead for an interested service let me hit OK here and let's go ahead and run this module as you see here we're going to provide specific data for the subject and text as we are using or we are just running that specific module so I'm going to come over here provide the subject I'm going to go ahead and provide the body here and then let's go ahead and hit okay another run as is here we can click here and this is
[22:40] the results so what we can do here is you can see your original message by clicking this and then if you come over to Output if I hit choices one it's going to go ahead and tell us the underlying message here which is going to be yes so because of the fact that that email just received was a potential lead it has identified that that email is in effect do in fact it is a yes for being a potential lead if that makes sense so from here we're gonna go ahead and set up a filter to do so we're going to go ahead and set up what they call here is a router and
[23:10] essentially a router is their way of identifying two different paths in a flow here and what we can do here essentially is we can go ahead and set up a Gmail for each one of these flows but for now let's go ahead and set up the if else of whether it is a yes or no so we're going to do setup filter I'm going to say if client equals yes and we're going to go ahead and do do condition and then you see how it says text is equal to so we're going to do condition and we're going to find the output here
[23:42] so we're going to go to choices message content and then we're going to say equal to yes and then we'll hit OK and then the other one is just going to be if client equals no I'm going to go ahead and do our condition here of the same thing so essentially go to our choices our message and then in theory if that email was not a client email we would have got an output from chat gbt saying no hey no here so essentially this sets it up where we have two different choices in theory
[24:13] if we have a client that says no we can set up a whole different flow for what occurs after that and then if they say yes we can go ahead and set up that draft email so we have Gmail here I'm going to create a draft we're going to go ahead and choose all mail we're going to add a recipient to that recipient is going to be the underlying email we received it from so we're gonna go ahead and go back to the watch Emos here I'm going to do the sender one so it's gonna be web Rista at webcafecommerce.com and then from here this is where we create the subject line and the content for the underlying email
[24:44] so actually before we even create that we do need to create essentially a gbt block that's going to create that subject line and that email I want to do is want to do an open AI block here and we're going to do creating completion and then from here we're going to go ahead and make sure we do gbt models and let's go ahead and do a gbt4 I'm going to add a message and do the role of assistance and then we're gonna go ahead and say based on this email and then we're going to go ahead and provide the relative context again
[25:16] subject body and I'm going to go ahead and do that real quick so I'm going to add the subject line here parentheses and I'm going to do the body of the email here and in this context we can keep it as plain text that should be fine and sufficient for what we're trying to do here I'm gonna come over here and real quickly grab the beginning of the context here and one thing I could have done in theory was clone this to make my life easier but I just want to show you step by step what we're doing here so we
[25:48] are an automation agency generates a body of an email in response to this potential lead and then essentially we can do here is I could give a ton of data a ton of information for this Gmail but I just want to give you a very simple flow here so I'm just going to say um respond to potential potential leads with a notice
[26:19] they will receive an onboarding PDF and a quick answer to any questions okay and then we're going to go ahead and add a parameter here we're going to say uh parameter Max of four sentences outputs in HTML I have no clue how good
[26:51] this is going to look we're going to go ahead and just hit OK here and let's go ahead and run this one thing I messed up on is that two two subjects here we gotta make sure to do the text content here and then we're gonna hit OK we'll run this module only we're going to provide the data and hit OK here I'll let this flow proceed and let's see what this and output will be for the underlying email I'll be very interested I think I need to restructure the formatting a little with the
[27:21] underlying body due to the fact that I didn't say no text before after so let's go see what it comes up with honestly perfect um as you see here it has dear Corbin which was the footer of that original email here we got we are excited to hear our animation AI Automation Services for customer service in the restaurant industry you are shortly going to receive an onboarding outlining the details of our service rest assured that any questions you will have will quickly be addressed kind regards your name now let me show you a little trick here if there's any time there is data that is found within the output but then it gives you know a variable Point instead
[27:53] of like the actual real data what you can do here is you can get go ahead and add a um you know simply this essentially do your name or my name actually my name semicolon parentheses and then in this context let's just say my name is Tim uh Adams and then from here and go ahead and retest this and there you go we got the same formatted email here but now it's taken to an account or underlying name Tim Adams let's go and proceed from here and
[28:25] build out that draft email I'm gonna go ahead and do a Gmail block here and create a draft we're going to choose a folder of all mail I'm going to make sure we choose our reception recipient here which is going to be the sender's email address here and then we're going to go ahead and choose our subject we're going to just do a fixed text subject we'll do your automation expert is here and then for the underlying content as you see you can use HTML tags and we have prompted our chat gbt to do HTML
[28:55] tags so let's go ahead and do that now one thing I want to point out is that as you'll notice when you're grabbing data it can kind of get confusing when you don't rename it so to do that you can go ahead and rename here we can put uh body uh email so we call upon that data it's a lot simpler a lot easier for us to understand we can add attachments if we wanted to but we don't need to we're gonna go ahead and hit continue or OK here and then for the purposes of this tutorial though I'm gonna actually send it to my email here or the courses email rather than the main email here and then
[29:26] for now let's go ahead and see what this will look like so I'm going to hit OK and let's go ahead and run this module so rather than running it from the end like that I'm gonna go ahead and just show from the beginning so let me go ahead and re-set this up real quick we have this ad on red as that's the only emails that are being looked for answers for I'm gonna come back to the beginning of this we're gonna go ahead and say run once actually this won't work I think I have to go ahead and choose where it starts we're going to select our data point of our email here
[29:59] just like that there we're gonna hit OK and then let's go ahead and hit run runs so as you see you can see it live how it's communicating with the underlying scenario here and then as it you saw it essentially deemed it as yes this is a client email and then we should have a draft set up in our back end here as you see once a flow incurs we can go ahead and check the outputs here and see what it does so this was the inputs here come over here to choices and message this is the message that was sent
[30:31] and it went ahead and sent it to our Gmail as a draft so this is a very cool little flow here showing a animation on make when it comes to handling cinnamon analysis and setting up an HTML body email now as you see here since we've ran it a couple times in our testing it does count that as operations and data transfer so we have currently 18 operations based off that expenditure and around 20 kilobytes of data transfer to create a scenario like this it's easy to clone move to folder and then essentially anyone within your team
[31:01] would be able to access that same scenario and help you work on it as you see here once you go to that scenario it comes up with all this information here we go to history we can see past runs the associated operations with that run and the associated data transfer with that run and then if there was any incomplete executions so that just about does it we did a complete overview of the made platform when it comes to open Ai jbt and just Automation in general with this platform make sure to like the video if you felt like you've learned something make sure to
[31:31] subscribe if you want daily artificial content we're doing a ton of stuff here at webcafe AI when it comes to a automation so make sure to stay tuned with us if you want to learn more about Ai and Automation and how to leverage that for business check out the playlist I did this video as we dive into further and more complex tutorials when it comes to zapier's platform and all 5000 apps that can be found there but without further Ado I'll see you in the next video thanks for tuning in and yes surprise I'm an AI Avatar make sure to explore
[32:03] more here at web Cafe where we demystify AI for your personal and business life until next time