Pipedream and ChatGPT With OpenAI Beginner's Guide To AI Automation
Let's build with Zapier and AI (100+ videos)📅 2023-08-25
Transcript
[00:00] welcome back to webcafe AI we do daily artificial intelligence videos for your personal and business life if you clicked on this video that means you want to learn comprehensively everything you need to know about the pipe dream platform when it comes to automation artificial intelligence and everything that's encompassed therefore we're going to provide you value in today's video we're going to give you a full-blown tutorial here with everything you need to do and learn when it comes to using pipe dream so make sure to give a like for the value but provide here let us know if you want more content when it comes to pipe dream also I hope you like
[00:30] our new setting here where we have uh we kind of structured our beginning of our videos different and you'll see later throughout this video how we structured it so without further Ado let's go ahead and jump in here we're going to go ahead and explore everything we need to learn about the pipe dream platform starting with its model and pricing so if we come to the PIP dream website here as you see this is another automation software like make and zapier and essentially this allows us to connect you know a ton of different apis as you see here with this software we can connect over a thousand
[01:01] different apps here that natively integrates and then a lot of it has to do with doing custom calls with apis one thing you'll notice between softwares like pipe dream and make relative to zapier is a huge cost difference but on top of that it requires a little bit more coding knowledge and a little bit more understanding of how to do API calls which isn't a bad thing learning how to do that kind of custom work can help you tremendously for any type of thing you're trying to automate in your business or for a client's business so right off the bat is free to use pip dream what you're going to see in
[01:32] today's tutorial is a free version we're going to go ahead and Link it down below and you'll be able to create your own account here let's go to pricing here as this is going to be one of the biggest indicators of whether you want to do a platform like this and as you see here I'm going to switch to monthly here it goes from 029 $149 and 749 depending on your use case what's great is that a lot of these platforms start with $ Z so we can really get an understanding intuitively whether we're comfortable with the UI and everything that the platform provides knowing this look at one of the bigger indicators here here when it comes to this platform relative to other ones and that is they put a cap
[02:04] on the amount of active workflows and connected accounts this could become a pretty big issue if you're scaling or you know growing as a company due to the fact that you know with the free plan we already get access to five active workflows and three connected accounts that should be fine but look at this this is a big indicator here as you see here if we go up to basic $29 we get uh access to 20 active workflows and only 10 connected accounts in order to increase from there we would have to go Advanced which means we can then get unlimited workflows and then get unlimited connected accounts but as you
[02:36] see here we deal with a pretty big price tag of $149 a month regardless as you see here with the free plan we get access to web hooks pre-build actions and the ability to write node Python and Bash so a lot of the fundamental stuff we can do with the free plan which is always nice so let's go ahead and jump into this platform and learn everything we need to know once sign out to sign up all you need to do is provide a email with Google or your GitHub account or just simply provide any type of email provider here and sign up for free so when you launch into your pipe dream
[03:06] dashboard we got a lot of stuff we're going to go over here today we're going to go over this entire sidebar here everything you need to know about it then we're going to go and actually build a workflow together we're going to understand the intricacies of how their UI works and then on top of that we're going to finish off this tutorial with us building out an AI automation live so you can understand how to one connect your API key to this platform but also just intuitively uh understand how it communicates when dealing with open Ai and API calls so right off the bat a new feature that they added here is going to
[03:36] be projects so essentially how this works is that you'll create a project and then you'll have workflows associated with that project so that'll make more sense as we get going here workflows is kind of you know where we do automations I believe this is called scenarios and make and then obviously zaps for zap year coming over here we got sources so this is where you're going to connect your underlying apps to your account but as you noticed before you you have a limitation so you can only do three new or three connected accounts when it comes to a free uh account or free uh account with Pip
[04:09] Dream from here we have uh the accounts that we've connected uh when it says sources though this is more specifically for the action so I want to clarify that real quick and then come down here we got data stores dat stores is a way for you to either itemize specific data you want to use in flows you can also put in code here it's a bunch of different stuff you can do with data stores then we got event history so this is going to show workflows that and you know obviously events that happen within those workflows and its past you can select a date range here you can filter
[04:39] by the status whether it was an error or success and you can search by the underlying workflows name the next tab here we got is settings so settings essentially is where we're going to be able to you know change a lot of stuff in the back end here so you can change your workspace URL for us which is to webcafe AI you can change the email that you want to receive notifications on and then as you see here you can even receive air notifications on slack but it seems like they want to do that over a advanced plan but to be honest with you you could probably set that up natively with one of their workflows if you wanted to and just access their API
[05:10] from here though you have different options for sign in your workspace ID whether you want to delete your workspace and then we got a bunch of different options up here is where you're going to be able to invite members so if you're with a team you can invite multiple members here and I think that's one of the biggest competitive features that I found relative to zapier is zapier kind of gatekeep the ability to add multiple multiple people to your zampier account so that could possitively run into issues um from here though this is where you're going to be able to add uh people to your account here everything else seems pretty standard here as you see here we got our
[05:40] billing and usage our plan and then the underlying uh cost Associated every single day with you know the different workflows that are currently incurring so the next tab here is the explore tab this is where natively integrated apps show up here so see you can kind of search through here and see if they have a specific app so do they have Shopify let's find out they do have Shopify do they have um you know see they got telegram there they got GitHub there uh do they have you know curious I think they do have Twitter for now but they did deprecate recently so they may not
[06:12] have Twitter in the future uh side note Twitter you're going to have to do API Custom Calls now because it seems like they're changing their platform and the way you access their backend but this is where you be able to explore apps that are natively integrated into pipe dream here obviously we can do custom API calls to integrate any type of software if they allow us to access they're back in the next tab here we got is community so with this they have a pretty good form here you know think like stack Overflow you post like uh you know you have a question and people help you out on the sidebar here so you know this is pretty helpful if you're new to
[06:42] something or want to do a certain feature that may be not readily available and you can also search through this they have a ton of different information when it comes to documentation University support and so on what I would say in most cases go with the community as you're going to be able to get very on point answers if you need to actually contact them you know this is where support comes into uh grounds here they even suggest you go to the community right uh from here though essentially you can even create a support ticket if you want to and then finally they got careers and essentially if you want to apply to pip dream there
[07:13] you go now if you come down here and you go to my account this is where you're going to be able to get access to your API key with the software here you going be able to change different stuff about the applications and the front end and then you know one experimental feature was SMS notification which is not Rel relevant to us right now so let's go ahead and jump over to sources actually so let's go and jump over to workflow here now one thing we should do before we create a project here is I want to point out one other thing is that you can create a new workspace so you want to associate workflows with you know maybe a client or a specific part of
[07:45] your business you can obviously create more workspaces here but for now we're going to go ahead and use our web Cafe AI workspace here we're going to go ahead and create a new project and let's begin we're going to do uh test project and you even have the option if you choose to do so to configure it with GitHub sync this is pretty cool here this isn't really a native feature found in make or zapier so keep that in mind if you're more uh you know use GitHub a lot and you want to access a lot of stuff within github's backend this could be very helpful I'm going to go ahead and create the project here so as you see here we have our
[08:15] project and with each project we have the ability to essentially create our workflows and if you run into any issues you can see here we can give access to the pipe dream support staff once we create our project here we can create folders so let's go ahead and start by creating a folder we're going to just say test run I'm going to go ahead and say create folder this is pretty helpful honestly uh I kind of like this organization or or how they organize their UI when it comes to workflows I find myself in zapier you know they have folders and and that kind of structuring but it it does get a little overwhelming
[08:46] I would prefer there to be multiple workspaces like this context or at least easy UI like this um from here though let's go ahead go into the new folder here test run and let's create a workflow so once you create workflow we're going to be able to create a name here I'm just going to do AI animation and then we're going to be able to uh identify so there's a lot of stuff here that isn't available in other automation platforms so we're able to adjust the timeout so this could actually be pretty useful in the context of AI due to the fact that some of the executions and flows take a little bit more time than
[09:18] uh typical right so when it's doing a lot of output for an AI prompt you may want to do a higher uh timeout here so in this context you know something like you know a minute or 104 seconds remember memory isn't too relevant for what you're trying to do in most cases you keep that at 256 and then we have all these different options here for this underlying workflow so do you want to send error notifications I would assume so automatically retrying errors I typically like doing this for AI animations but we would need to upgrade so keep that in mind uh limit
[09:48] concurrency limit execution rate uh eliminate cold starts most of other stuff is not too relevant for us right now so we're going to go ahead and create the workflow so once you come into a workflow here it's kind of similar to zier's UI we got a trigger here we got the deploy button here then we have a bunch of other information relevant to this now they have the ability to share links so if you want to share this with an individual that maybe isn't a member within your team yet you got to expand code and config expand results and you know you got the settings of the underlying workflow here
[10:19] you can come back and readjust stuff that maybe you wanted to uh you know want to come back there and you know readjust some settings that you wanted to change over time you got deployments and inspector so when you add a trigger you're going to get this kind of menu here as you see this is kind of similar to zapier uh with popular uh different triggers here and then essentially you'll be able to search their database of what they natively integrate on if they don't natively integrate with it that's when you're going to use an HTTP or web poer request and then under that context go to the underlying software's documentation and proceed from there so
[10:51] before we proceed any further let's go ahead and jump back over to accounts here and let's connect an open AI account so we can start accessing atificial int intelligence with our flows here so we're going to say create our connect new app let's do open Ai and just for my sake let's make sure they don't have a native chit gbt and they do not okay so we're going to do open AI CH gbt and essentially let's go ahead and grab our API key to do so you're going to Simply type into Google open AI developer and then you're just going to sign up with your email completely free this is a usage based uh
[11:21] software so therefore there is no flat reoccurring fee it's purely based off the amount of usage of AI you use if you're more if you want to learn more about that you can just type in open a pricing uh we're talking about uh you know fractions of pennies when it comes to open AI usage uh from here though once you're an account what we're going to do is we're going to jump over to our main admin panel here and we're going to uh go to view API keys so we go to API Keys here we're going to go to create secret key here and I'm going just call it pipe dream don't try to take it I'm going to delete this after the video so
[11:52] if you want to go ahead if you want to copy it but it's not going to work so I'm going to go ahead and just copy our API key here obviously with it goes without saying make sure no one gets access to your API key as you don't want you know your charges to go out to Wazoo here I'm going to go ahead and paste it in here and we can give it a nickname so we're going just give it uh AI for fun and then we going to hit save there we go so we got our open AI connected here I have a Google drive connected here as well let's go ahead and connect one more app here we're
[12:22] going to go ahead and just connect YouTube here so I'm going to hit Youtube now when connecting a Google backend it'll simply just ask you to sign in I'm going hit continue here and we should be working here perfect so we connected our YouTube we've connected open Ai and we connect to Google Drive let's go ahead and jump back over to that project and build out a flow real quick so for this flow we're going to go ahead and just do a YouTube as trigger here we're going to go ahead and do the trigger of new video by username here we're going to select our account here and then pulling
[12:53] interval essentially is how often does the API check if this incurred so for this context we're doing every 15 minutes the username here um I believe it's asking for this format I may be wrong so let's go ahead and see what this comes out with create Source it seems like the better option for this trigger would be just going by Channel ID we're going to go ahead and create this Source here and one thing that's important is if you want to do anything else when it comes to like actually manipulating YouTube then you would actually make sure you know connect to your account that gives you the access
[13:23] to the API that allows you to push that forward as you see here we went ahead and created this event here also notice you can add more triggers here so maybe you want to flow which is actually pretty interesting maybe you want to flow where maybe there's a new YouTube Channel video maybe there's a new tweet you want to kind of Encompass it all within one workflow for multiple different things that could cause the trigger to go off for our context though we're going to go ahead and just go with simply just a new YouTube video by Channel and we go ahead and test this event if it is successful as you see
[13:55] here we'll get a green little box and then it'll provide us a bunch of different different data here now as you'll notice like always um the data is you know going to be formatted like this which isn't too bad so you can go ahead and find specifically what you're looking for here uh from here though let's go ahead and proceed so we're going to go ahead and say continue one thing I want to point out though is that event was a test event so in the context that you can't actually trigger the event we can still provide it with test data here I'm going to go ahead and change that to an actual real piece of data from our channel here at webcafe Ai and we're going to go ahead and proceed with this data let's go ahead and see
[14:25] how we can Leverage The Open AI block here so we're going go ahead and type in open AI choose open AI here and then we can do all these different options here so as you see here their way of structuring open AI in this context is having all within one specific API call so we can do chat summarize text classify items and categories translate text create trans transcription create embeddings create completion for R context you would most likely always choose chat from here we're able to choose
[14:56] between 3.5 and all the different models here so we got 3.5 uh gbt 4 gbt 3.5 uh turbo 6 I'm going to go ahead and up the model to gbt 4 and as you see here this is where we're going to put in our underlying message all right so let's go ahead and test this out by making a user message here we're going to say generate a new YouTube title use this as reference simply call parentheses here let's go ahead and find that title right here now one thing I want to point out
[15:27] right off the bat is as you see here this isn't the actual underlying dictation intext that you see in the title it uses what would be seen in an API call and actually just gives you the straight up you know data from that API call so this is like front facing rather than like a nice little UI that you know zapier would show us where it show the actual text of the title rather than the API call uh from here we're going to say uh make it more SEO effective just as an example and let's go ahead and see how they generate this output and here we go so as you see here we have our output
[15:58] put here a couple things you want to note to find the response you're going to go to original message with assistant response click down here and then this is the response for that underlying input the output in this context um and then as you come down here this is pretty important so this is the amount of tokens used for that underlying um you know API request so this is important in the context of building up flows and really understanding the amount of costs associated with each type of run that you do so from here though we have our generated output
[16:28] using pip dream and open AI now just to take this one step further so we can show you how to build upon open AI outputs I'm going do another chat here I'm going to go ahead and say based on this and I want to point this out because you know there is certain tools that are maybe functional within pipe dream that can do this automatically such as you know like a zap your filter so or a zapier formatter they have their own version of formatter but I want to point this out because you can use AI to format text still and you don't necessarily have to use a native integration found in the back end of
[16:59] pipe dream so we can say based okay based on this YouTube title do semic parentheses and then I'm going to put the output found from that previous step here now one thing I want to point out here is the actual UI to find um previous outputs is kind of jumbled here I mean I have to scroll through this entire thing to find what I'm looking for if I was going to build out a very complex flow this would really get annoying after a while so from here
[17:29] though I can find what we're looking for here and then essentially put in generate again but without quotation marks me put a parentheses here because this may not be able to understand specifically what I'm doing here I'm actually going to up this to gbt 4 so we don't run into any issues here and go to test this flow all right let's go ahead and see if it worked correctly here and it did okay so perfectly so we got the exact same title but as you saw pre previously I come back over here this had quotation
[18:01] marks but now we can format them out uh this can come very useful so essentially if you get a lot of outputs and you want to deal with data and restructuring data you can always use a CH DBT block to do that now so you don't have to necessarily do any type of code or python or some type of formatter that's not using AI I mean this is a very simple and fast way to format data if you run into issues like that now just to show you the functionality of this platform I'm going to go ahead and add a Google drive here and then essentially just do a simple action of upload file now one thing to point out is when you build out these different flows and
[18:31] actions it isn't inherently going to give you um essentially everything that you would want right off the bat what I mean by that essentially is I could test this action and you know nothing's going to occur so you got to go ahead and make sure you choose the different option fields that you want specific to what you're trying to achieve so for this context maybe I just want to give the drive that we're dealing with and then the fire URL and then that is sufficient enough for what I'm trying to achieve maybe in some context you want to actually add a name to it or you want to do an upload type and so on so from from here though I can go ahead and grab the file URL to do so I'm simply going to go to
[19:03] that YouTube trigger I got the jpeg here from that underlying you know thumbnail from YouTube I can goe and change the name here to uh test up and then from here I provided the URL provided a new name provided where I want it to be dropped I could have had a folder here I'm going test that action and as you see it has successfully upload that file to my drive here now a couple of things I want to show you real quick is essentially let's say you're working on a flow and especially with Chad gbt a lot of times you Proctor it and the prompt may be you
[19:33] know 80% same but 20% different for the context that you want to you know do another step within that flow in order to expediate the process of handling you know multiple blocks that are similar we can come over to these three ellipses and we can simply duplicate the step which is nice so everything that was stored here is stored here now another thing we can do here is we copy it and then paste it later we could expand section we can move the step up and down and then as you see here we can edit the underling step just simply clicking that or just clicking the actual step itself one thing that is really useful is you
[20:04] can go ahead and rename the step here so for this context we're going to say uh new YouTube title so this is pretty useful in the context that you know when you're building out complex flows you can call upon exactly what each uh functionality of the block is rather than just saying chat one chat 2 upload file uh you need more context you can rename the block and once it is live you'll be able to see it here you'll see the test run you'll see the apps associated with it the file size and so on this is really really solid stuff here from pipe dream I will say that overall though it really depends on your use case and what you're trying to
[20:35] achieve when it comes to automation for what type of platform you're trying to choose I hope this tutorial was helpful in trying to show you how to leverage pipe dream when it comes to Automation in general and integrating AI with it to learn more about leveraging artificial intelligence for your personal and business life make sure to subscribe here as we do daily artificial intelligence content if you feel like you learned something make sure to like the video it's completely free and it shows us here at webcafe AI that you prefer more content that is structured like this furthermore we're going to provide you with a playlist at the end of this video that dives into other Ai
[21:06] and automation tools with full breath comprehensive tutorials so make sure to check out that playlist there we go over chat bot we go over other automation platforms and so on so you really learn how to start implementing this into your business backend 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 more here at web Cafe where we demystify AI for your personal and business life until next
[21:39] time