Build a Complete AI SaaS with zero code (Lovable + N8N)

0:00 today we're going to show you how to 0:01 build a full AI SAS Marketplace from 0:04 scratch with zero coding experience 0:07 we're going to use lovable for the 0:08 front-end website na automations in the 0:11 back and even allow to process payments 0:15 using stripe and Link all of the data 0:17 together so it works as a full 0:19 functioning SAS using superbase in the 0:21 database if you're someone like me with 0:24 pretty much zero coding experience then 0:26 this must sound really exciting how many 0:29 times have you had an idea for something 0:32 but you gave up on it simply due to a 0:35 lack of technical skills no idea where 0:37 to start or not enough money to hire 0:40 developers but what I'm about to share 0:42 with you in this video might change all 0:45 of that for you and by the end of this 0:48 you'll be able to turn your idea into a 0:50 full AI SAS that you can monetize and 0:54 sell to others and by the end of it you 0:56 might just have built a fully working 0:58 MVP version 1:00 of your business idea 20 times faster 1:03 than you could before at a very low cost 1:06 we'll show you some examples of what 1:08 people are already building using this 1:10 exact technology and how they're 1:12 monetizing it to thousands of users 1:15 we'll give a simple overview of how the 1:17 whole thing works interacts with each 1:19 other and each tool and we'll give a 1:22 step-by-step guide on how you can turn 1:24 your ideas into a full app that you can 1:27 monetize so we're talking payment 1:30 data website backend the whole lot 1:33 you'll be able to build by the end of 1:35 this but a quick disclaimer as you guys 1:38 know I try my best to give you all the 1:40 information that you need to know 1:42 without the hype just practical 1:44 Solutions so would you be able to build 1:47 a million dollar SAS using this probably 1:49 not but that's why I recommend watching 1:51 this video Until the End so you can see 1:54 what is possible and what's not possible 1:57 and exactly how to capitalize on this 1:59 opportunity 2:00 especially if you're like me and you 2:03 don't have an insane stack of technical 2:05 skills so let's jump straight in with 2:07 some examples of what you could build by 2:10 the end of this so we've got agent AI 2:13 first it calls itself a professional 2:15 Network for AI agents but it's a a web 2:18 app where users can add their own AI 2:21 agents you click into them so for 2:23 example this company research agent and 2:25 you can pay to use that AI agent so on 2:29 the back end that would just be an N 2:31 automation for example and on the front 2:33 end the user is able to click through 2:36 and put in some inputs and receive an 2:38 output from our agent there this website 2:41 alone supposedly has above 500,000 users 2:45 and you can see this agent alone has 2:47 above 25,000 tasks completed for that 2:50 specifically we've then got websites 2:53 like relevance AI which is a very 2:56 similar concept it build teams of AI 2:58 agents that deliver human quality work 3:01 so if we go to the agents here we've got 3:03 all sorts of different agents for 3:05 example CRM enrichment where if you sign 3:09 up you can effectively input a prompt 3:11 input some inputs link it to your data 3:14 and run that exactly in the same way 3:17 we've then got more specific 3:19 marketplaces that are suited towards for 3:22 example voice agents so we've got retail 3:24 ai.com supercharge your call operations 3:27 with voice AI so again it's a simp SAR 3:30 product where they've got various 3:32 specific voice agents that you can use 3:35 like receptionists appointment Setters 3:37 lead qualification that's all on the 3:39 front end and on the back end they're 3:40 just running an automation that takes 3:42 our inputs and gives back to the front 3:45 end our outputs we're going to build 3:47 something exactly like this with two 3:49 clear examples so that you can go away 3:52 and build an AI SAS Marketplace for 3:55 yourself or for other businesses so 3:58 let's look at how we're going to 4:00 implement that and this is all with zero 4:03 coding so we'll run through step by step 4:06 exactly what we need to do but first 4:08 let's look at how the web apps going to 4:10 interact with the different components 4:12 and what those components are so we've 4:14 got a front end here so the front end is 4:17 just what the website looks like and the 4:19 user sees on the front end so if we go 4:22 back to agent AI as an example this is 4:24 the frontend website for agent Ai and 4:28 every time we go into a specific agent 4:31 we've got another page in this URL 4:34 that's part of our front end so we're 4:36 going to build out our front end with 4:38 lovable and lovable is your superhuman 4:42 full stack engineer so let's have a look 4:44 at lovable. deev and it's all about 4:46 creating an app from an idea in as 4:49 little time as possible with zero code 4:52 so lovable we can prompt like we do with 4:54 an llm or chat gbt to create a full 4:58 stack web app for us where we can 5:01 connect to all of our Integrations in 5:03 the back end and we can start for free 5:06 with lore and then when we hit the 5:08 capacity limits we can upgrade to $20 a 5:11 month so for replacing a four stack 5:14 development team $20 a month $50 a month 5:17 is very very cheap and it performs very 5:20 well as you'll see so we're going to 5:22 have lovable on the front end and every 5:25 page we click through on our website is 5:27 going to be built by lovable but it's 5:29 also going to set up those Integrations 5:31 with what we need on our back end to 5:33 actually make this happen so we've got 5:36 the front end in lovable now let's look 5:38 at the back end so Three core features 5:41 will make up our back end and this will 5:43 depend on what you want to build for a 5:47 full AI SAS Marketplace we're going to 5:49 need three things we're going to need 5:52 the AI automations running in the back 5:54 end and for this we're going to use n8n 5:57 we're going to need a database 6:00 that stores all of our user dataa so 6:02 that users can log in and we can track 6:05 their usage of the different tools and 6:07 therefore charge them for it and make 6:10 sure they're not exceeding their limits 6:12 or or credits and finally we're going to 6:15 need some way for them to pay so we're 6:17 going to integrate stripe with that and 6:19 all of those are going to interact 6:21 together so when a user clicks a button 6:24 on the front end to log in it's going to 6:27 either force them to register through a 6:29 sign sign up route which will put their 6:32 name and their details in our database 6:34 in super base or if they're already 6:37 logged in and they're taking an action 6:38 with a specific automation then we want 6:41 the database to update as the action 6:44 happens and similarly when they first 6:46 sign up for the first time they're going 6:48 to need to pay and that's going to be 6:50 through a stripe link but it's also 6:52 going to be registered in our database 6:53 which is the middle linking to 6:55 everything and then finally we need the 6:57 automations to actually run so we're 6:59 going to use na for that so let's have a 7:01 quick look at each of these now before 7:03 we delve into the specific automations 7:06 so we've got na. here which is an AI 7:10 workflow automation software for 7:12 technical teams it effectively allows us 7:15 to Plug and Play drag and drop and 7:18 connect to all of these different 7:20 softwares that we use like slack Google 7:23 Mail our database thousands of different 7:26 software but also connect to AI agents 7:29 as well if you're not familiar with n 7:32 then I'll link to my full 4-Hour course 7:35 which covers everything you need to 7:37 start building in n and everything I 7:40 learned over eight months of building 7:42 over a 100 automations in N so that's 7:44 where we're going to have our 7:45 automations now superbase is effectively 7:49 a database so when you're using 7:51 something like air table that's also 7:53 considered a back-end database but 7:55 superbase is much more scalable for web 7:58 apps so you can see here building a 8:00 weekend scale to Millions this is where 8:03 we'll store all of our user data 8:06 including which users we've got signed 8:09 up but also their usage on different 8:11 tools and you can see as well here super 8:13 base is fantastic because you can get 8:15 started for free and then finally let's 8:18 have a quick look at stripe stripe 8:20 allows us to create quick easy payment 8:22 links and Connect into different 8:24 softwares so stripe is really good for 8:26 things like International payments so 8:28 we'll use stripe 8:29 for everything payments related let's 8:32 take a quick look at the automations and 8:34 I just wanted to Quick touch on how 8:37 these are going to work and the 8:39 technical infrastructure behind that but 8:41 it's actually really very simple when 8:42 you understand the concepts so they're 8:44 all going to run by web hooks that are 8:46 activated from our front end so when we 8:49 click a button it's going to send the 8:50 data to n and Na's going to receive that 8:55 that is effectively what the web Hook is 8:56 doing it's receiving the data from the 8:59 user clicking a button on our website 9:01 we're going to process the automations 9:03 and then we're going to send data back 9:04 to display to the user also using a 9:08 return web hook there so that our front 9:10 end can receive all of the information 9:13 pushed back by n So today we're going to 9:16 look at two specific examples that we 9:18 can emulate from 9:20 agent. and 9:22 relevance. actual examples so first 9:25 we've got a sales lead qualifier which 9:28 if we open it up directly on agent. a it 9:31 scores and qualifies leads based on 9:33 scoring criteria so a fairly generic 9:36 overview there but it costs one credit 9:37 per execution and 4,000 people have 9:40 completed tasks here if we go through 9:42 the process it effectively asks us for 9:45 an email our criteria and then returns 9:49 this scoring criteria we're going to 9:51 replicate this on our own and what we're 9:54 going to input are a LinkedIn URL 9:57 criteria to score against and it's going 9:59 to Output the same score against 10:01 criterias with details the second one 10:04 which is arguably a little bit more 10:06 complex because we have to send a file 10:09 which is why I chose this as the second 10:11 example is extracting data from a PDF 10:14 and summarizing so if we go into 10:16 relevance ai. here we've got a PDF 10:19 Studio where we can effectively upload a 10:22 PDF file in the interface it's going to 10:25 send it to our back end in N we're going 10:28 to receive that file 10:30 extract information summarize that 10:32 information with an AI agent or llm 10:36 chain and then return that information 10:38 back to the user again so we'll be 10:40 dealing with more complex uploads there 10:43 and we might need some more complex 10:45 tools in the back end like like an OCR 10:48 tool to help read images off the PDF for 10:51 example But ultimately we're going to 10:53 build out those two use cases today step 10:55 by step so you can see how to do it but 10:57 this tutorial will give you the 11:00 capability to build out any other AI or 11:03 non- aai tools for yourself by showing 11:06 you two clear different examples of how 11:10 we can use this and how we can set this 11:12 up relatively quickly and maintain and 11:16 edit it super easily so now that we 11:18 understand our infrastructure there 11:20 let's go down to the pathway that we're 11:22 going to take to get there so that you 11:24 can see what's coming up first we're 11:26 going to figure out how to prompt 11:28 lovable to build the app that we want 11:31 we're then going to build out the 11:32 automations in N we're going to set up 11:35 our superbase database for login and for 11:38 payments and prompts lovable again to 11:41 change a few things there to make sure 11:43 that works we're going to have to set up 11:45 the infrastructure on stripe directly 11:47 for our payments and then we're going to 11:50 iterate and test with one of those 11:52 automations and then we'll move on to 11:54 the second automation following exactly 11:56 the same Pro so the bulk of the work at 11:57 the start is going to be done in lovable 12:00 and I want to create a prompt that can 12:03 help us create the full AI SAS 12:07 Marketplace that's really clear so I 12:10 looked up a few tips on prompting for 12:12 lovable to improve the way in which I 12:15 prompt and I found this uh lovable 12:18 prompting Bible which you can search for 12:20 as well and all I've done is gone to my 12:23 favorite AI tool perplexity which is a 12:26 web search engine powered by AI models 12:29 in the back end so powered by llm and 12:33 I've written out this prompt to give us 12:35 a starting point of the prompts we're 12:38 going to enter into lovable so it's just 12:40 a starting point it's not going to be 12:41 perfect we're going to need to refine it 12:44 but I definitely recommend doing this as 12:45 well so we've got I want a prompt 12:48 optimize for lovable use the lovable 12:50 prompting Bible it needs to be about 12:52 developing a web app with a technical 12:55 Focus clean homepage so that I can have 12:58 a similar look and feel 12:59 it needs to have a superbase connection 13:02 once users are logged in they can see a 13:04 page of the different automations that 13:07 will be searchable so similar to what we 13:09 saw on our agent. a page you'd have 13:13 various automations but also a homepage 13:15 as well to get them to sign up and once 13:17 they're logged in you see those 13:18 different automations I've then talked a 13:21 little bit about the functionality you 13:23 really need to be specific with the 13:25 prompt here and what functionality you 13:27 want from it so a few things that came 13:30 to my head we need a place to log in 13:32 which is stored in superbase so we're 13:34 just giving it a heads up that we're 13:35 going to need to connect superbase you 13:37 need as a user to use credits to use an 13:40 AI tool the third one is each AI tool 13:44 has some sort of input Fields so we'll 13:46 Define those later but now let's just 13:48 say they have input Fields And once the 13:51 tool is used first thing is credit 13:53 should be deducted so it needs to 13:55 interact with our back end here to sign 13:58 up there'll to pay $20 a month which 14:01 will be connected through stripe so 14:03 again laying the foundations that we're 14:05 going to have an integration through 14:06 stripe and once the tool is used it 14:09 should send the input data to the web 14:11 hook address and receive back the output 14:13 so we're telling it it's going to send 14:15 some data on the click of a button and 14:17 also receive back data from a web hook 14:20 at the end of it so it's a very short 14:23 concise prompt but I've laid out clearly 14:25 the functionality and the Integrations 14:28 between the different Services there so 14:30 that lovable can go and set up a really 14:33 solid starting point for us to go in 14:36 test modify Etc and it's come out with 14:40 this really clear guidelines so it's a 14:43 full prompt optimize for lovable and by 14:45 the way if you want access to these 14:47 prompts rather than creating them 14:48 yourselves head to 14:51 school.com scrapes come into the 14:54 classroom go to the plug andplay 14:56 templates for your business and you're 14:58 going to go go to publish recently which 15:00 is all of the templates published in the 15:03 last week and we're going to go to AIS 15:05 SAS Marketplace and you've got all of 15:08 the prompts down here and then all of 15:10 the N files to work with as well to make 15:13 it super quick and easy so let's head 15:15 back to the prompt in perplexity we're 15:17 going to copy that and we'll just stick 15:19 with this first prompt first we're going 15:21 to take that to lovable we're going to 15:23 make sure that we uh sign up I'm sure if 15:26 we click enter here it's going to tell 15:28 us to sign up in or sign up so sign up 15:30 for free sign into your account and then 15:33 once you're signed into your free 15:34 account you're just going to copy paste 15:36 that into the the chat box there hit 15:39 enter and it's going to start building 15:41 out our app so the thing that lovable is 15:44 going to do for us is start building out 15:48 the framework in code without us 15:50 touching any code the framework for a 15:53 full web app so normally if you were to 15:57 build this you'd need 15:59 a full stack engineer or you need to be 16:01 a full stack engineer yourself to be 16:03 able to write out all this code but 16:04 lovable starts by just understanding 16:08 your core requirements and turning that 16:10 as you can see into code so it will 16:12 start writing all the code blocks and 16:14 you don't need to worry about all of 16:15 this this is just telling you that it's 16:18 starting to work on your idea it's 16:20 probably going to take five or so 16:22 minutes to fully write out the code but 16:24 it's effectively powered by an llm in 16:27 the background that specifically 16:29 tailored to creating website designs but 16:33 not only the front end of a website 16:35 which are quite easy to already make 16:37 using something like webflow or framer 16:39 but it will start to integrate the 16:41 backend functionality for us and you can 16:43 see for example it's starting to create 16:45 the superbase connection here so we'll 16:48 let that all load up and whilst that is 16:50 loading up let's go to superbase and 16:52 start setting up a project there so 16:54 you're going to sign up to superbase and 16:57 then you're going to start your project 16:59 if you've got no projects already you're 17:00 going to be in this dashboard page once 17:03 You' signed up for a free account and 17:05 you're going to click new project choose 17:07 the organization that it should sit in 17:09 and we're going to call it something 17:11 like lovable demo you're going to then 17:14 save a database password and choose a 17:17 region but just leave it as the default 17:19 I would and that will start creating a 17:21 new project in superbase we're going to 17:24 later need to take the secret keys from 17:27 our project in superbase and give that 17:29 to lovable so that it understands what 17:32 database to connect with and whilst 17:34 that's still setting up we can go up in 17:36 lovable dodev to the superbase 17:39 connection here we're going to click 17:41 connect to superbase and what that's 17:43 going to do if you signed up for 17:45 superbase and you're logged into your 17:47 account is allow you to authorize API 17:50 access for lovable so we're going to 17:52 give it access to the correct 17:54 organization and we're going to 17:55 authorize that and at this point you're 17:57 not going to be able to connect to any 17:59 of the tables because actually if we go 18:01 into view tables we've not set up any of 18:04 our database tables yet it's just an 18:07 completely empty database in superbase 18:10 so we're going to cancel that come back 18:12 let's see how our project's getting on 18:15 we've had our first error so lovable is 18:19 really good in some aspects and then can 18:22 error out quite often so all we're going 18:25 to do is just try to fix and it's then 18:28 going to push the the error back to lore 18:31 in order for them to start fixing the 18:33 error and if this doesn't work we're 18:35 just going to simplify our prompt and 18:36 work at it a little bit at a time give 18:39 it a simpler prompt to start with and 18:41 then add functionality as we go through 18:43 so let's go back to lovable now and it's 18:45 actually identified the error for us and 18:48 uh started a fix on that and actually 18:50 now output our first view on what our 18:53 Marketplace s would look like and it's 18:55 not bad for a first go we'd obviously 18:57 prompt this to improve the UI change the 19:00 marketing copy Etc but we've got unlock 19:02 the power of AI with a simple credit 19:05 system so not really selling the value 19:07 there but the description underneath is 19:09 not bad access a Marketplace of Premium 19:11 AI tools with a flexible pay as you go 19:14 credit system no commitments just a 19:16 powerful AI at your fingertips so it's 19:18 really understood the brief and created 19:21 this nice header on the homepage it's 19:23 then got a link to view all the tools 19:26 with a little uh how it works so 19:28 purchase credit select a tool get 19:30 results which is exactly as we described 19:32 it's got a foot down here again the 19:34 color scheme needs a bit of work but 19:37 it's done a really good job we've not 19:38 had to build anything or build out any 19:40 code we've just clicked fix a few times 19:43 and it's come here so let's first click 19:45 get started and you can see that 19:47 automatically it's thrown us to a new 19:49 page which has the endpoint register and 19:53 we would register our account here and 19:54 it's also potentially set up our Google 19:58 so up here we can click to go back into 19:59 the various different pages let's go 20:02 back to register and it's also set up 20:04 these other pages so dashboard tools 20:06 pricing billing let's see if we can 20:09 access any others right now okay this is 20:11 a really good example of where it's 20:13 implemented some security measures here 20:15 so we've tried to access the tools or 20:17 the dashboard and it's thrown us back to 20:20 the login page which is exactly what we 20:22 want if we've got a paid service we want 20:24 to be logged in or we want our users to 20:27 be logged in before they can access any 20:28 private Pages it's then given us uh 20:31 individual tools pages so this tells me 20:34 that in the back end we're going to have 20:36 multiple tools each with a unique page 20:39 but following the same structure we've 20:41 then got a pricing page let's have a 20:42 look at that okay and it's followed what 20:45 we said but it's actually given us a few 20:47 different options here let's see if the 20:49 sign up free button works and instead of 20:52 using the Google oo here let's see if we 20:54 can sign up using an email Okay so we've 20:57 hit an error here error signing up up 20:58 failed to fetch what we will do is let's 21:01 just come down here and I have a 21:04 suspicion that this is because we 21:06 haven't connected to our super base yet 21:08 so now the website is loaded let's go up 21:10 to the superbase connection in the top 21:12 right corner we found the lovable demo 21:14 project and all we're going to do is 21:16 connect to that project great so it's 21:19 connected follow the instructions in the 21:21 chat to complete the integration so 21:23 let's go back to the project and it's 21:25 given us some additional instructions to 21:27 follow to connect to super base so it's 21:30 encouraged us to just keep building and 21:32 giving us a few bits of information 21:33 around user accounts login Etc but the 21:37 one thing we're missing is we still have 21:39 no tables inside 21:42 our super based database so let's ask it 21:45 to do a couple of things let's ask it to 21:47 fix the login page and that's still not 21:51 working so it wasn't a super base issue 21:54 or it potentially was because there's no 21:56 tables there so let's tell it the errors 21:58 that we're faing and it will go into the 21:59 code and fix that for us so let's go 22:01 back to our school page where we have 22:03 the implementation plan in here and it 22:06 already set out a plan for what the 22:08 different tables should look like in 22:09 super base so if I copy and paste that 22:11 we'll go back to lovable the first one 22:14 we've already put in so login sign up is 22:16 not working either through Google oo or 22:18 normal signup and then I've put the name 22:20 of the error in and then I'll put we 22:23 have no tables in our superbase database 22:26 here's an example of what they could 22:29 look like add any other necessary 22:33 tables and we'll just paste in that 22:35 super base example set up there and 22:38 we're just going to let lovable run 22:39 through that and see if we come back 22:42 whether the login and sign up errors are 22:45 now fixed so this is where lovable gets 22:47 really powerful because they have the 22:48 inbuilt integration with superbase we 22:51 don't even need to go to superbase to 22:53 set up our tables if we were setting up 22:55 a project through superbase we'd have to 22:58 go to the SQL editor and this is an 23:00 alternative way to do it we'd copy and 23:02 paste in these table schemas which is 23:04 basically the different tables and how 23:07 they link to each other and the data 23:08 types and we'd have to paste those into 23:11 here we'd have to run the command but 23:13 because they've got this inbuilt 23:14 integration it gives us the different 23:16 tables that we suggested along with 23:19 access policies and you'll need to go 23:21 through these because they won't be 23:22 perfect and make sure that these access 23:25 policies are secure before you publish 23:27 don't treat this like a production ready 23:29 app you still need to make sure that 23:31 it's got the security policies test out 23:33 the website test out all the different 23:35 edge cases see if you can break into 23:38 your own website because there may be a 23:41 policy missing that means other people 23:43 can access certain pages they can't or 23:46 people can access uh information from 23:48 the database so you need to go through 23:49 and check that as well or use an AI tool 23:52 to help you check those policies and 23:55 understand what they mean but we've got 23:57 a few policies in here like users can 23:59 view their own transactions user can 24:01 insert their own transactions so all it 24:03 asks us to do is to apply changes so 24:06 that's going to create those tables in 24:09 super base for us so again we don't have 24:10 to touch a thing and soon we'll be able 24:13 to start building out the backend 24:14 automations that can link to this so 24:17 first let's just see if the generic sign 24:20 up is working before we do anything 24:23 around Google so let's go to create 24:26 account here and it's come up with a 24:29 cool little popup here account created 24:31 check your email to confirm your account 24:33 and if we go back to our super base 24:37 tables so we can go to our database on 24:39 the left hand side here you have a 24:41 visual view it's set up three tables for 24:44 us one is tools one is profiles so that 24:48 would be our users and another is 24:49 transactions so you can see that it's 24:51 understood and started building out what 24:54 our tables should look like and the 24:55 links between those tables and and if in 24:58 fact we go instead to table editor we 25:02 click into profiles you can see that 25:04 it's created a new user under my email 25:08 and it's given me 50 credits because it 25:10 said on the free plan you get 50 credits 25:12 so it's got all the information we need 25:14 to start connecting us up there it's 25:16 missing a stripe customer ID because it 25:18 hasn't forced us to go through a a pay 25:20 wall which we'll cover but you can see 25:22 that it is now working the front end 25:24 with the back end which is really 25:26 awesome and taking no time at all so 25:28 let's go back to lovable here and let's 25:32 sign in so signed up as well with a 25:35 different email address here whil the 25:37 camera was off and we're just going to 25:39 log in there and that will hopefully 25:42 take us through to a dashboard great so 25:44 what we've got here is a dashboard it's 25:47 come up with error fail to load tools 25:49 because we've not actually uh created 25:51 any tools here but it's actually 25:53 predetermined that we've got a search 25:55 bar here to search the different tools 25:57 it's from our initial requirements that 25:59 this is like a tools Marketplace we'll 26:01 be able to pick some tools spend some 26:04 credits on those tools so it's given us 26:06 50 credits for the free trial and we've 26:09 got different categories that it's 26:10 assigned here so two things we're going 26:12 to need to do here we are going to need 26:16 to one create some tools on the front 26:18 end but two be able to connect those 26:21 tools to our back end which we said 26:23 we're going to build in n and enable the 26:27 connection between the front end and 26:28 back end so let's start doing that now 26:30 so we're going to go into our workflows 26:33 in n8n and once we're inside n we're 26:36 just going to go to create workflow and 26:39 just to remind you the automations that 26:41 we're going to build here we're first 26:42 going to build a sales lead qualifier 26:45 where the inputs are a LinkedIn URL and 26:48 criteria that we're going to score 26:50 against and it's going to output for us 26:52 a score against certain categories that 26:55 we give it using Ai and the second one 26:58 is uploading a file extracting data from 27:01 that PDF and summarizing it using Ai and 27:05 sending that back to the front end so 27:07 within our workflow we'll just call 27:10 this lovable D sales lead qualifier and 27:15 all we're going to do is open up the tab 27:19 here of tools and we are going to click 27:21 on the web hook so we're going to 27:23 receive a web Hook from loable to this 27:27 address and let's first just test that 27:29 that actually can be set up for us so 27:33 let's take the production URL from here 27:36 and we're just going to make sure that 27:37 workflow is saved and we're going to ask 27:40 in lovable to create two tools and we're 27:43 just going to say create one tool a 27:45 sales lead qualifier with an input field 27:47 of LinkedIn URL and a second input field 27:50 is a description of qualification 27:53 criteria these values should be sent to 27:55 the following web hook address and what 27:57 I've done is taken the address from our 27:59 production URL copy paste this get 28:02 address but let's actually just change 28:04 it to a post address because we're 28:07 sending data let's leave all of this as 28:10 is but actually we can change this path 28:13 here to sales lead qualifier just to 28:17 make it a bit more clean let's take that 28:20 back to our lovable environment and 28:23 replace that web hook address and then 28:25 we've told it it should display a 28:27 descrip text box with the web hook 28:29 response when it returns so what we're 28:33 going to have here is the web hook 28:35 that's receiving the data from lovable 28:37 when I user inputs and clicks next we're 28:40 then going to take some action so for 28:42 now we're going to use the do nothing 28:45 operation just to test it works and we 28:48 will have a web hook respond to web hook 28:52 node which will send the data back to 28:56 our lovable site and what we're going to 28:59 do is say respond with all incoming 29:02 items back to The Lovable site and in 29:05 here we're going to create a node or a 29:09 set of nodes in N to qualify those sales 29:13 leads but right now we're just testing 29:15 that the connection works so let's ask 29:17 lovable to create that and it has the 29:20 web hook address in here so it's going 29:22 to go ahead and start making that and 29:25 then whilst it's doing that let's 29:27 duplicate at this flow and just to 29:31 remind you we were extracting data from 29:33 a PDF and summarizing it so let's just 29:36 call it extract from PDF let's duplicate 29:39 that flow and again we have the web hook 29:42 do nothing respond to web hook and we're 29:45 just going to change the path here to 29:48 extract from PDF we're going to copy the 29:52 production URL and we're going to just 29:55 have to make sure that if we're using 29:57 the production URL and not the test URL 29:59 that we've activated the workflow so 30:01 that they can actually receive the web 30:04 hook trigger here so again at this point 30:06 we're just testing we're not trying to 30:08 achieve anything in particular apart 30:10 from sending the data from the front end 30:13 received do nothing with it respond with 30:16 everything that it's received back just 30:18 to check it displays on the back end as 30:21 well so I've come back here and it's 30:22 told me that the build was unsuccessful 30:25 if you click into this it tells you what 30:27 was unsuccessful cannot find name 30:30 skeleton did you mean selection so there 30:32 was just some build error but what we 30:34 can do whenever we get build er error is 30:36 go up here on try to fix it it's also 30:39 giv us a warning that a file here is 30:42 getting too 30:44 lengthy but let's not worry about that 30:46 let's just ask it to rebuild and see if 30:48 it can get it done so I logged back in 30:50 and I noticed there were no tools 30:52 showing and it it's had the error failed 30:54 to load no tools again so I've told it 30:57 that and it said okay the issue existed 31:00 because in our database there was no 31:03 tools is featured column and it was 31:06 relying on that to exist to show the 31:09 different tools so it's given us the SQL 31:14 code to add some columns to the tools 31:17 table so that it can then apply those 31:19 changes so let's apply those changes and 31:22 we'll run it and we'll see what it comes 31:24 back okay so after a few iterations of 31:26 just saying the tool isn't showing at 31:28 all it's added some sample tools to our 31:32 tool table that are now showing in our 31:36 that we just accepted and told it to run 31:39 it and they are now showing in our 31:42 dashboard so we've got featured tool 31:44 section and an all tools section and you 31:47 can see it's added four tools but not 31:50 the tools that we asked it to add so 31:51 let's go back up and just reiterate this 31:55 prompt that we'd like it to create this 31:57 tool 31:58 and then hopefully we can move forward 32:00 with this tool and start testing it out 32:03 you can see that it's also assigned 32:04 different credit amounts to these 32:06 different tools and it's done all of the 32:09 user interface for us so we don't have 32:10 to worry about that and it's even 32:12 created descriptions put it in different 32:14 categories so this is really cool and 32:16 really quick this would take you a long 32:18 time to mock up by yourself and then 32:22 create the database in the back end and 32:24 all the data and the fake data whereas 32:26 this has done it super quickly for us 32:29 even though it's a bit back and forth 32:31 it's going to save you a lot of time if 32:32 you want to create a front-end website 32:34 that interacts with the back end so 32:37 let's create this tool and then start 32:39 testing if it interacts with our web hug 32:42 and one thing I will mention here is 32:44 whilst we were going through that 32:45 process it asked me to upgrade and said 32:47 I was nearing my daily limit so if 32:50 you're going back and forth trying to 32:51 create something complex like this you 32:53 might have to upgrade to the $20 per 32:56 month once but arguably that's well 32:59 worth it for the functionality you get 33:01 and the time you're saving we've now got 33:03 the tool appearing in the all tools and 33:07 it's given us a little description here 33:09 so let's click through and what we can 33:11 see here is we have a description 33:14 qualify sales leads based on LinkedIn 33:16 profile and custom criteria so very 33:18 appropriate it's got an about section 33:21 and we can see there's a few options 33:23 here it's a sales tool we can go back to 33:25 the dashboard tells us about the credits 33:28 per use uh gives us another description 33:31 and then it has a run tool makes it very 33:34 clear that we're going to use eight 33:35 credits to do so but obviously we 33:37 haven't set up the functionality to take 33:38 away the credit so I'll be interested to 33:40 see if it does that correctly it's got 33:43 the LinkedIn profile URL with an example 33:46 in here that we could insert and then 33:48 it's got the qualification criteria 33:51 which we enter a specific criteria to 33:54 qualify this lead so actually it's done 33:55 a really good job of setting up this 33:59 interface so that we can pass inputs in 34:01 let's actually try and put some inputs 34:03 in here let's add my LinkedIn profile 34:06 just as an example and see if our web 34:10 hook actually receives it and we'll see 34:12 this in the executions inside n so I've 34:15 added in my LinkedIn URL and just added 34:18 test qualification criteria let's click 34:21 analyze lead it said it's processing and 34:24 it said there is a problem problem 34:26 processing your request please try again 34:28 later let's go to n and it seems like 34:32 something has run so let me come back 34:35 inside here inside our executions we'll 34:38 be able to see if the data was received 34:40 and the issue was not that the data 34:42 wasn't received actually seems to have 34:44 received the data it's just that we had 34:46 respond immediately rather than use the 34:49 respon to web hook node so let me just 34:52 come in and change that to respond to 34:54 web hook node and we'll do the same on 34:57 the other tool as well whilst we're here 35:00 let's save that let's go back let's give 35:03 it a go again analyze lead and it's 35:06 processing and it's actually come back 35:08 with an output straight away so it's 35:10 actually received a web hook response so 35:13 let's go into the executions let's see 35:16 what's happened here and check that 35:18 we're actually receiving the data 35:20 correctly and that it's sending back 35:22 correctly so we've received the body 35:25 which contains the LinkedIn URL and 35:27 criteria so that's amazing it's got all 35:29 of these headers which we're not really 35:30 concerned about says that it's coming 35:32 from The Lovable project but effectively 35:35 we are sending and receiving the data 35:38 and also we have sent back the LinkedIn 35:42 URL and criteria so we've sent 35:44 everything back and you can see that 35:47 we've received the body back of the 35:49 LinkedIn URL and criteria so that's 35:51 awesome the web Hook is actually working 35:54 so now what we need to do is build out 35:56 the workflow to actually do something 35:58 for us like sales lead qualification so 36:00 whenever we want to use an app that we 36:02 know isn't inbuilt into n for example 36:06 here we are trying to scrape LinkedIn 36:08 profiles we're going to come to 36:09 something like appify that has a whole 36:12 list of user built nodes that we can 36:15 connect into or apis that we can connect 36:18 into an alternative to that is finding a 36:21 solution that's another external API 36:24 that scrapes LinkedIn profiles or does 36:26 the exact thing you need here if we just 36:28 got a list of lots of different options 36:31 that we can connect to with one simple 36:34 API call so we're just going to type in 36:36 LinkedIn profil that's what we're 36:37 looking for here and we're going to find 36:39 a LinkedIn profile scraper like this one 36:43 uh 2.7k users 4.8 out of five that seems 36:47 like a good bet you'll be able to try 36:51 this for free and with every plan on 36:53 appify you get a free plan up to $5 36:56 usage for free and you can see this one 36:58 charges $10 per thousand results so you 37:01 could get 500 results on your account 37:03 for free here but when you're creating a 37:06 paid SAS anyway you're going to have 37:08 paying customers so this might be easier 37:11 than going out and building your own 37:13 LinkedIn profile scraper so it's hit try 37:15 for free on this one and I'll show you 37:18 now how to set that up in na so I've 37:21 previously looked here and we've added 37:23 some profiles so we've got Bill Gates 37:25 here and my profile here and all we're 37:27 going to do is we're going to go back to 37:30 the lovable na backend sales lead 37:33 qualifier flow going to open up a HTTP 37:36 request which allows us to connect to 37:38 any API and we're just going to make 37:41 this a post request because we're going 37:42 to be sending data we're going to find 37:44 the url for this case we don't actually 37:47 need the authentication because I'll 37:48 show you you pass it in the URL we pass 37:51 our API key and then we're going to send 37:53 a body which are going to be the 37:56 profiles that we want to scrape for that 37:59 we're going to make sure that's Json and 38:01 we're going to go and get that from 38:02 appify now so inside the appify scraper 38:06 we go to Json here in the inputs and you 38:08 can see that it gives us the Json that 38:10 we need to pass back in our actor so 38:14 we're going to turn this into an 38:15 expression because we want to change it 38:17 to Dynamic inputs in a moment but we'll 38:19 test it on these two profiles first and 38:22 then we're going to go back and we're 38:24 going to hit Integrations use API endp 38:27 point 38:28 and what this is going to do is give us 38:30 a few different options to run this 38:32 endpoint but it's also going to append 38:34 our API token to the end if you haven't 38:37 set up an API token before go into 38:40 settings in API and Integrations create 38:43 a new token and that will be the token 38:45 that's created you probably have one 38:47 already in your profile so let's go back 38:50 to the actor use API endpoints and we 38:54 are going to want to run the actra 38:56 synchronously and get data set items so 38:58 basically we want it to return the data 39:01 and don't uh so we don't have to keep 39:03 asking if the data is ready we want it 39:05 just run and return that data so we've 39:08 copied that here we're going to paste it 39:11 into our actor here inside the URL and I 39:15 will delete this token afterwards 39:17 because you can see it just appeared in 39:19 here as well and that means we don't 39:21 need authentication because it's already 39:23 built into the request so let's just 39:26 test that and see if it Returns the 39:28 information from these two profiles 39:30 successfully and because we're doing two 39:32 profiles and it was $10 per th000 this 39:35 is going to cost 2 cents and at this 39:38 point we're just testing if it works and 39:40 you can see it's returned all of this 39:42 different M information if we go to the 39:44 schema you can see we've got the 39:46 connections the number of followers job 39:48 titles job descriptions so a lot that we 39:51 could use to qualify this person if it 39:55 was a sales lead qualifier tool so we've 39:58 set that up in our appify actor so what 40:00 we need to do now is say okay from the 40:03 data we received from the web hook scan 40:06 that information and do something with 40:07 it so let's just start adding some notes 40:11 and if you followed my videos before I 40:13 always add notes so that when I reflect 40:15 back on workflows I understand exactly 40:17 what's happening so we're going to have 40:18 scrape LinkedIn profiles in here and 40:21 then we're probably going to have 40:22 another step where we qualify that lead 40:26 against the criteria that's input so for 40:29 that we're going to need an llm node 40:32 which is effectively being able to chat 40:35 directly with n sorry with AI and we're 40:39 going to need to give it a model so 40:41 we'll give it open AI chat model and 40:44 we'll connect to open router so we can 40:47 access multiple different models through 40:49 there if you've not set this up before 40:52 check out the full course where we run 40:54 through all of this including what is an 40:56 AI agent how to use different llm 40:59 workflows and everything like that but 41:02 if you're familiar this should be fairly 41:04 straightforward so we're going to 41:05 receive the data we're going to scrape 41:07 the LinkedIn profiles and we're going to 41:09 qualify the leads just going to rename 41:11 this and again I like to name all of my 41:15 nodes so that they're really clear so 41:17 this is going to be qualify leads and 41:20 we're going to Define below the prompt 41:23 in which we ask it to qualify those 41:26 leads and tell it what it needs to 41:28 return in response so inside the qualify 41:31 leads I'm just going to write a or I've 41:33 just written a basic prompt you're a 41:36 sales lead qualif specializing in 41:38 LinkedIn base lead evaluation your task 41:41 is to assess each lead based on their 41:43 LinkedIn profile which is the input and 41:46 rank them against the following criteria 41:48 job title and role industry and Company 41:50 fit professional experience and skills 41:53 for each Criterion provide a score one 41:56 to five along with a brief justification 41:57 based on the provided on the available 42:00 information and then we will just give 42:02 it the inputs dynamically profile info 42:05 and we'll pass that in in a second and 42:08 then criteria to assess against and what 42:12 we need to do for that is pass those in 42:14 dynamically so we can just go to a 42:16 previous execution of this a successful 42:20 execution copy to editor and we'll be 42:23 able to see the content that we passed 42:25 in the LinkedIn URL and criteria so 42:30 inside here we are going to have the 42:32 criteria from our initial web hook let's 42:36 put that in here and that's the criteria 42:38 it's going to assess against and then 42:40 for profile info we've got a couple of 42:42 options here because we're going to 42:44 receive potentially multiple profiles in 42:47 a list so what we want to do first is 42:51 let's just test a single profile so 42:54 let's pass in the LinkedIn UR here 42:57 inside the body and you can see that 43:00 that's populated dynamically we just 43:02 drag and drop that 43:03 over and then if we just have one it's 43:06 going to be relatively easy because 43:08 we're passing through one here one here 43:10 one output here so let's test it with 43:12 one first and then we'll show you what 43:13 to do if you want to qualify multiple 43:16 leads at once so we need to make sure 43:19 that we're pulling in the profile info 43:21 here as well so let's say from the web 43:24 hook do item Dot 43:27 Json and this if you haven't done the 43:29 full course is just reaching into the 43:32 previous 43:34 inputs 43:35 body. LinkedIn URL and you can see now 43:38 if we open this up down here that it's 43:40 passing in the profile info that's 43:44 that's actually incorrect I need to pass 43:47 in the response of the appify actor so 43:51 let's just deactivate these and what we 43:53 can do is just run this so that it P 43:57 calls the data from the applify actor 44:00 and we've got an error here Json needs 44:01 to be a valid Json so let's remove the 44:04 speech marks from there and there's 44:07 still an error here so let's dig into 44:09 this ah it will be the comma here so 44:12 let's put the speech marks back you 44:15 should have a comma between the items 44:18 but never at the end of an item that is 44:23 never on the last item so now you can 44:25 see it's a valid J object that's going 44:28 to run and return the data hopefully 44:30 there we go we've returned the data and 44:33 now what we're passing into the qualify 44:35 leads will be any data we deem suitable 44:40 so ideally what we want to do is 44:41 aggregate all of this information and 44:43 pass it in as a single item so what we 44:46 can do is just go back into our prompt 44:49 and instead of doing this what we're 44:51 going to do is dollar Json to Json 44:55 string and that's just going to convert 44:57 everything we've received into one 45:00 parameter there are other ways you can 45:01 do this like the aggregate node or just 45:04 pull the information you need like the 45:06 set node but let's just run this on a 45:09 test locally before we interact with 45:12 lovable and you'll want to just pin the 45:15 data once it's run so that you can pin 45:17 it whilst you test so you're not being 45:19 charged every single time and that will 45:21 just run the same information and it's 45:23 returned uh multiple items here Simon is 45:27 the founder of automake as a Founder is 45:29 likely the key decision maker so it's 45:31 assessing against criteria but obviously 45:33 our criteria that we put in was terrible 45:35 it was just test criteria but it should 45:38 only be returning one item in the output 45:41 here so let's make that clear in the 45:44 output format and in fact what we're 45:47 going to do is a method where we can 45:50 just return without a specific output 45:52 format so let's delete these and let's 45:55 see this time it will return one output 45:58 format but for multiple leads we're 46:00 going to make a few amendments in a 46:02 moment okay so it's returned one output 46:05 format and let's actually test this with 46:08 a profile URL using the web hook and 46:12 that will actually respond to the web 46:13 hook then let's add multiple so let's go 46:16 back to lovable we will go into our tool 46:20 and we're in the dashboard and we'll go 46:22 back to the sales lead qualifier and if 46:23 we go to run tool we know that we can 46:25 put in our LinkedIn URL and our 46:27 qualification criteria and let's unpin 46:30 the data here unpin it in this test save 46:33 the workflow and all we're going to do 46:36 is put in the LinkedIn URL that we want 46:38 to scrape so we've put s uh looking for 46:42 co-founders who have marketing 46:43 experience to advertise our marketing 46:45 tool to so this is qualifying my profile 46:50 against that so what we're going to do 46:52 is analyze that and it's going to run 46:54 through and we'll see this in the 46:57 executions of the tool it will run 46:59 through here so it's going to scrape the 47:02 profile it's going to qualify my profile 47:04 based on the information it receives 47:06 against that lead and then it's going to 47:09 return that hopefully to lovable so 47:12 we'll go back to lovable yes it doesn't 47:14 look brilliant but we can format that 47:16 output but it's got all the content so 47:19 the output is displaying so it's 47:21 returning all the data as we saw before 47:23 and it's given different scores so 47:25 actually okay 47:27 professional experience and skills score 47:29 three Simon has a solid technical 47:31 background with significant experience 47:32 in operations and automation true but 47:35 his marketing experience is not 47:36 explicitly outlined in the skills or 47:38 endorsement on his profiles his past 47:41 roles have involved scaling delivery 47:42 operations and automating processes 47:44 which can translate to insights into 47:46 marketing and user engagement however 47:48 specific marketing skills appear to be 47:51 lacking making this aspect of his 47:53 profile somewhat less compelling for 47:55 advertising your marketing tool so 47:57 actually it's done a really good job of 47:59 kind of reading the profile analyzing it 48:02 and returning that data so you can see 48:04 that you could offer a tool like this to 48:06 a user but one thing you've probably 48:08 noticed is that the credits have not 48:11 gone down so something is not working in 48:14 terms of when a tool is used updating 48:17 that in the database which is really 48:19 important when you're trying to charge 48:22 users for a tool that's based on credits 48:25 or number of uses one quick thing we're 48:29 going to do is just make this a bit more 48:30 useful able to scrape multiple profiles 48:33 at once and return information based on 48:36 that and then we will deal with the 48:38 credits not updating and move on to our 48:42 second automation before we finally set 48:44 up the payment what other tools you 48:47 could sell with this kind of s so let's 48:49 say we have a comma separated list as 48:51 the input what we're going to do is use 48:53 the edit fields or set node and we can 48:57 just say Okay based on the input we're 49:01 going to need to grab the LinkedIn URLs 49:04 but perform manipulation on them so 49:07 let's call this URLs not URL and we're 49:10 going to say okay turn this into an 49:13 expression so it's going to be a 49:14 JavaScript expression and what we are 49:17 effectively saying here is if you have 49:19 multiple inputs under LinkedIn URL which 49:23 would come through our field then we're 49:26 going to separate by the commas and then 49:28 each URL so each item in that by comma 49:33 we are going to remove any white space 49:36 and then we're going to join it by these 49:39 quotation marks but also a new line 49:41 between it so it looks complex but 49:44 actually I've just used perplexity to 49:46 talk through the formatting that's in 49:48 the input and the formatting that is the 49:51 output and it's helped me come up with 49:53 this n expression that's going to work 49:56 the job for us so let's go with an 49:59 example of multiple and check that just 50:02 up until this point it's working 50:04 correctly so if we go back here and 50:07 let's put in a comma and let's put in a 50:10 second LinkedIn URL and let's just use 50:13 build Gates Again so we've got two URLs 50:16 here let's send that to be analyzed and 50:19 for some reason it's there was a problem 50:23 proing the request so let's go check if 50:25 the problem was in inside our workflow I 50:29 see so because we've disconnected this 50:31 respon web hook flow it's not worked 50:34 correctly so let's just connect this 50:35 back up and try it again and it should 50:39 now start processing that if we come 50:40 back to n we'll see in the execution 50:43 history what's happened here so you can 50:46 see that it's received multiple items 50:51 and it's actually split them exactly how 50:53 we need for our appify input so so 50:56 appify has then uh got the incorrect 50:59 input but if we just drag LinkedIn URLs 51:02 across here that should work correctly 51:04 so in appify actor let's go back to 51:07 executions and this is a great way to 51:09 test things by the way you can copy this 51:11 to editor and now we can see okay 51:14 instead of putting this in here let's 51:16 just dynamically pass in the LinkedIn 51:18 URLs and we can see that would populate 51:21 exactly like we need in our appify actor 51:24 now to process multiple of these so this 51:28 is going to pass multiple items to our 51:30 qualify lead which is then going to pass 51:33 multiple responses back to the web hook 51:36 so to pass it as one single response so 51:39 that we give a response based on all of 51:42 our inputs in fact let's go back and 51:45 test it and I'll show you exactly what 51:46 it does if we put input multiple let's 51:49 analyze lead which is running the tool 51:51 again and that's processing and you can 51:54 see in the executions that it is running 51:58 and actually that's worked better than I 52:00 thought it would I thought we would need 52:01 some additional method here to get the 52:04 response in here but it's actually given 52:06 us the response for both of those so 52:10 we've got that back in here now 52:11 obviously the format that it's returning 52:14 in is not ideal and we could do some 52:17 extra formatting steps but for this 52:19 demonstration it's about actually 52:21 passing that data back we could do some 52:23 extra formatting steps we've got plenty 52:26 of Community Resources about 52:28 reformatting data so we won't touch on 52:30 it here but basically we've got our 52:32 first SAS tool that you could sell to 52:35 other people which is our sales lead 52:37 qualifier and it didn't take any time at 52:39 all most of the time is in the setup and 52:42 waiting for lovable to run and then 52:45 iterating back and forth with that so 52:47 that's really cool let's go to our 52:50 second one here which is extracting data 52:53 from a PDF what we're going to do is set 52:55 up this one 52:56 exactly in the same way but touch on a 52:59 few things about file transfers and then 53:02 we will look into the credit system and 53:04 how to make sure that that updates so 53:06 that your users aren't abusing their API 53:09 keys or so that you can charge credits 53:12 for using certain tools and then let's 53:14 also look at applying a pricing gate so 53:17 that users have to pay to sign up and it 53:20 will confirm in the back end that 53:24 they're signed up and paying before it 53:26 them to use the tool so the key things 53:28 you need to make it into a proper SAS so 53:32 again let's go back to lovable and let's 53:34 just test if we receive the binary data 53:36 inside the web hook and also can send a 53:39 file back because we're going to need 53:40 both of those functionalities 53:42 potentially in the future so we'll go 53:44 back to the lovable prompting and 53:46 similar to when we told it exactly what 53:48 we needed for the first tool let's do 53:50 exactly the same for the second tool so 53:52 I've started prompting it here and I've 53:54 just decided to be fairly generic at 53:56 this point and not give it much 53:58 instructions so just a PDF extraction 54:00 tool and it's identified in lovable that 54:03 we want to create a PDF extraction tool 54:05 that can pull content from PDF files and 54:08 it said okay to do that let's add 54:10 certain things to the database so let's 54:13 approve those changes the one thing 54:15 we'll also need to do is enable a user 54:18 to upload a file so we might have to 54:20 prompt it that as well so let's see what 54:22 this does for us so it's already said 54:24 now that we've successfully added the 54:26 PDF extraction tool to the database I'll 54:28 create the PDF extraction tool 54:30 functionality so that's not what we want 54:32 to do here um but it we do need to allow 54:36 the upload of PDF files so let's see 54:38 what it makes and then uh modify it from 54:41 there and as you can probably already 54:42 tell I'm regretting being so vague in my 54:45 inputs so I should have been more 54:46 specific I need the user to be able to 54:49 upload the file in inputs not do the 54:51 extraction itself the file should be 54:53 sent as data to the following web hub 54:56 and I've given it the web Hook from our 54:59 test or sorry production URL and we will 55:04 update that there so that it can 55:06 actually send the binary data file or 55:09 the PDF to that web hook let's Let It 55:13 edit come back and test the tool out 55:16 whilst we're doing that we will add the 55:19 ability to download the file so we've 55:22 got the extract from file here and if 55:25 we're just doing basic 55:27 extraction we can actually just connect 55:29 up this to the web hook extract from PDF 55:32 the input would be called data and again 55:35 we're just going to add an llm chain 55:37 because all we're going to do is ask it 55:38 based on the inputs summarize the 55:41 content in the PDF so we'll put notes on 55:44 our workflow in a moment but for now 55:47 let's just connect the same as we did 55:49 before to open rout or open AI whatever 55:52 your preference is through there and in 55:55 the prompts let's 55:57 say to summarize the incoming PDF data 56:01 and obviously if you were building out 56:03 these tools you want to add in your own 56:05 prompts this is just a demonstration of 56:06 the capabilities of sending files to and 56:09 from our web page that we've built with 56:12 lovable and we'll reference the correct 56:15 data point in a moment once we 56:17 understand the data is being sent 56:19 through so we know this is going to fail 56:21 at first it says the PDF extraction tool 56:23 now uploads files directly to your 56:25 specified webhook H when a user selects 56:27 a PDF file and clicks the upload PDF 56:30 button it's sent to our web Hook and the 56:33 response from the web Hook is displayed 56:35 in the text area below great so let's go 56:37 into our tools in the dashboard and 56:40 we've got PDF extraction cost five 56:43 credits again given all the details here 56:46 let's go to run tool and it's given us 56:48 an upload PDF file option which is 56:52 awesome when we click it it opens our 56:54 most recent files and let's go to this 56:57 file in my documents AI Explorer road 56:59 map which is one of the files you get 57:02 when you join the school Community it's 57:04 like an onboarding guide so let's see if 57:06 it understands that file let's click 57:10 upload PDF and we know already it's 57:12 going to fail in N so we just want to 57:14 see what the error is there and also 57:17 what data is being passed through so a 57:19 lot of this is going back and forth just 57:21 testing there we go the errors come up 57:24 and great it's actually sent the PDF 57:26 file but it's called it file not data so 57:29 all we need to do is either tell it in 57:31 lovable to change it to data when it 57:33 sends or just call this file and that's 57:35 absolutely fine we'll extract from the 57:38 file now we will be able to tell if we 57:41 run it quickly again and it's often 57:43 better to just open this into a new tab 57:47 because this keeps refreshing in this 57:48 window so you can open it in a new tab 57:50 and refresh instead it's probably going 57:52 to save you a bit of time if we refresh 57:56 here we can see it's the correct tool so 57:59 there was an error there which is 58:01 probably coming from our n environment 58:03 again yes it was and that's because we 58:05 don't know what data is being passed 58:06 through yet so we can see that it's 58:09 passing through this text field so again 58:12 let's debug in editor which allows us to 58:14 see the previous executions get rid of 58:17 this and let's just pull through the 58:19 text all we want is the text and let's 58:22 take this to the top summarize the 58:24 incoming PDF data and we'll call this 58:27 summarize PDF as we always want to label 58:29 this up we'll add a few workflow notes 58:33 and again refer to the longer 4our 58:37 tutorial on N best practice for the 58:39 reasons in why I do this but when you 58:42 revisit workflows or create lots of 58:44 tools this is going to be super helpful 58:47 for you to reference what's actually 58:49 happening so we have summarized PDF 58:51 we'll have extract from file and it 58:53 should just upload text in response 58:56 so let's try it again upload the PDF 58:59 that's going to process let's see if it 59:01 comes back with a summarized response 59:04 and again the formatting won't be 59:05 perfect because it'll be the out direct 59:07 output of our llm node but here we go 59:11 we've got a uh a file that says 59:14 extracted content but it's also put it 59:16 in an output the AI Explorer road map is 59:19 a guide designed to help businesses 59:20 leverage Ai and automation effectively 59:24 it talks through all of the different 59:26 steps that are part of the guide so 59:28 we've now successfully passed the file 59:31 from a front end to the back end 59:34 summarized it using Ai and pass the 59:36 response back the final thing I just 59:38 want to touch on here is like what 59:39 happens if you want to pass the file 59:41 back so I'm going to put this should 59:43 also be able to receive an output file 59:46 back if one is passed and to remove the 59:49 extracted content section CU it already 59:50 had an output we're going to run that 59:53 that's going to update that for us so 59:55 now we've got two working completely 59:57 different AI SAS Tools in a very short 60:00 space of time you can see how quickly 60:02 you can mock this up if you come into 60:06 the school you can go into the classroom 60:08 and the plug andplay templates and what 60:11 you could do is use absolutely any of 60:13 these to create SAS tools for yourself 60:16 so we've got things like a meeting 60:18 summarizer so people could upload their 60:20 meeting notes you've got an email inbox 60:22 agent you've got scraping any website 60:25 data you've got the thousands of tools 60:27 on appify you've got a voice planning 60:30 agent so somebody could so somebody 60:32 could upload their voice and it could 60:34 plan something out for them you've got 60:36 competitor research so you could input 60:38 links you can see the opportunities with 60:40 all of these templates are endless when 60:43 you combine it with the fact that you 60:44 could then sell this as a product with 60:48 Dynamic inputs that would push outputs 60:50 to to your users so it's really an 60:53 amazing tool or an amazing tool set to 60:56 use together and I've created this with 60:59 no experience in very little time so 61:02 we've come back now to lovable and it's 61:04 saying the PDF extraction component now 61:06 uploads PDFs to the web hook and can 61:10 also receive them back so let's refresh 61:12 our page here and let's just go into our 61:17 workflow here and test that theory so 61:20 let's connect it directly to our respond 61:23 to web hook node let's just deactivate 61:26 those so the file should now be received 61:28 and sent back so let's see if that 61:30 actually is true so a lot of this is 61:33 like back and forth testing to check 61:34 things work but you can see we've built 61:36 a lot of different functionality not 61:37 just text back and forth but also any 61:40 files you wanted to upload really simple 61:43 to to do so let's do exactly the same 61:45 again let's upload the PDF see what the 61:47 response is interesting so it's now come 61:50 back and immediately it's downloaded the 61:53 process document so we have confirmed if 61:56 we look at the execution that actually 61:59 it did succeed and what it's done is 62:02 receed the file and send that file back 62:04 so we can send files back and forth as 62:06 well so let's just connect this all back 62:09 up and that is to working SAS tools with 62:14 very little time this whole time we've 62:17 been running on 50 credits which is a 62:19 big issue because if you're selling this 62:21 to others you want them to use their 62:23 credits up as they're doing this so we 62:25 need to to tell it so we need to tell 62:27 lovable to take credits away when 62:30 somebody's using it and then also we 62:32 need to implement a pay wall with 62:34 stripes so we're going to cover both of 62:35 those now so let's now talk about 62:37 setting up a credit system that works 62:40 and if I just remind you of our diagram 62:43 here we effectively have the front end 62:47 built so lovero is integrating with our 62:50 backend we've got two automations set up 62:53 and you can set up any automation like 62:56 this impr prompt lovable set up the N 62:59 Automation in the background and we've 63:01 got two automations already set up 63:03 inside n and like I mentioned at the 63:06 start you can Implement absolutely any 63:08 automations and have this as a tool on 63:12 your Marketplace or as a SAS separately 63:16 those are running now correctly we now 63:19 need to tackle these backend steps which 63:22 is making sure that when a tool is used 63:24 we log the usage for that user in 63:27 particular but also on a tool level and 63:30 then stripe we need to connect our 63:33 account so that we ensure that users 63:37 have paid to sign up to our service as 63:39 well so let's start with the database 63:41 and ensuring that when we're using 63:43 credits which are just arbitrarily 63:45 assigned at the moment but we could go 63:46 back in and change that they're actually 63:49 counting on the DAT on the database and 63:52 then once we've used a certain amount it 63:53 should then tell us that we need to up 63:55 upgrade to a paid plan so we're first 63:58 going to go back into the database 64:00 project we'll go to table editor and 64:03 we'll just have a quick look through the 64:04 tables here so we've got two different 64:06 profiles that have signed up so those 64:08 were the two IDs I signed up with we've 64:11 then got the tools so any tools that we 64:14 don't want to appear on our SAS we can 64:17 remove here because actually we've not 64:19 generated any of these tools and these 64:22 are just ideas it's put in for us but we 64:24 have made the sales lead qualifier in 64:25 the PDF extraction so let's delete those 64:28 four rows from the back end we'll 64:30 refresh here which is connected to our 64:32 database we'll go back to tools and 64:35 you'll see now that only two have 64:37 appeared here we might not want this 64:39 featured tools section but it's not a 64:41 problem for now and again we could 64:43 change the categories but we're not 64:45 worried about that for now so 64:46 transactions in here we have an empty 64:49 database so you can see that none of the 64:52 users have been added here so whenever a 64:55 button is clicked in our app like say we 64:57 use the sales lead qualifier it should 65:00 actually deduct those credits if it's 65:02 successful from our account so once we 65:05 receive a response or once we click the 65:07 button we should make sure that it's 65:09 deducting that from our account so we're 65:12 going to go back to lovable and just 65:14 tell it that that's not happening so I 65:16 said each time a tool is run it's not 65:18 deducting credits from the users's 65:19 account also the user that's registered 65:22 doesn't seem to have a transaction 65:23 registered in the table so I know this 65:26 from our transactions table we've got 65:28 all the data in here but actually none 65:30 of it is being registered so let's go 65:33 back and let's process that with lovable 65:37 and that's going to make updates to that 65:39 in the meantime let's head to stripe 65:42 where we're going to set up payments so 65:44 you're going to need to set up an 65:45 account in stripe and once you set up a 65:47 stripe account it will give you some 65:50 instructions on whether you want to 65:52 start looking at one-off payments or 65:53 recurring payments so here we're looking 65:55 at recurring payments so say pay $20 a 65:58 month get 100 credits Etc and whilst 66:02 we've been away it's fix the issues of 66:06 credit deduction and transaction 66:08 recording When tools are run so before 66:11 we set up the payments let's just test 66:13 the credits here so we've got 50 credits 66:16 here and if we go to the profiles we can 66:20 see that we've still got 50 credits on 66:23 our profile in the database so there's 66:26 two things that need to happen it needs 66:27 to update in the database but it also 66:29 needs to update on our front end to show 66:32 the user that they've used some credits 66:34 so we're going to refresh this here and 66:37 this says that it's going to use eight 66:38 credits per use and in our backend datab 66:41 base we can see inside the tools that 66:44 it's got a credit cost of eight in there 66:47 so that's great so we're going to run 66:49 the tool we're going to pass in the URL 66:51 we're just going to put some random 66:53 criteria here but because all we're 66:56 looking out for is that it's updating 66:57 our credits both in the database and on 66:59 the front end we will have to add logic 67:02 to this that says if the user has zero 67:05 credits they or not enough credits to 67:07 perform the action then they can't take 67:09 the action so we can see that this is 67:12 not updated up here let's go to the 67:15 database and refresh and again it's not 67:18 updated here has it recorded a 67:20 transaction yes it has so it's saying we 67:22 used eight credits and it gives us the 67:25 in and the output but actually our 67:27 profile S suggest that we've got 50 67:30 credits so the profiles has not been 67:32 updated and nor has the front end so 67:34 let's go back to lovable so you can see 67:36 how you pick apart these different 67:38 problems and solve them one by one with 67:40 lovable help so I said the profiles 67:41 table was not updated with the credit 67:44 usage and nor was the display of credits 67:46 on the front end I.E this part and I've 67:49 added here make sure a user cannot use 67:51 the tool I make them buy more credits or 67:54 put them through through the process to 67:56 buy more credits if they don't have 67:57 enough to perform an action that's a 68:00 critical step in our process because we 68:02 want customers to be able to perform 68:04 actions or guide them through the 68:07 process to buy more credits to do 68:08 actions so let's send that let's see 68:12 what lovable comes back with and test 68:14 that out again okay so coming back here 68:17 with a bit of persistence it's now seems 68:20 to be working you can see it's updating 68:21 my credits here I just contined to say 68:24 similar things 68:26 about the database tables not updating 68:28 eventually it got there let me show you 68:31 exactly what it looks like it's 68:32 currently got a different issue but 68:34 you'll see these credits up update it 68:37 will run it will deduct the credits 68:39 correct amount of credits the database 68:42 tables will update let's see that update 68:45 there you'll see in the transactions 68:47 it's added a new 68:49 transaction however it's not returned 68:53 our it's like refreshed the page too 68:55 quickly and not return the output so 68:56 we're going to need to fix that so we'll 68:59 just say it updated the credits 69:01 everything's working but I was awaiting 69:03 the output which never arrived basically 69:06 so we're stepping in the right direction 69:09 the other thing I want to mention here 69:10 is we have functionality now to deduct 69:13 credits but when you're building out a 69:16 credit based system or any sort of 69:18 recurring Revenue base system the 69:21 database tables need to have a fair bit 69:23 of functionality 69:25 if you come into the school I've 69:27 included a section on setting up an 69:31 effective credit management system 69:33 including the backend prompt here the 69:35 frontend implementation and edge cases 69:38 so you can come in here and grab those 69:40 prompts and that will help you speed up 69:42 the process and remove some of those 69:44 errors when we're going through and 69:47 trying to add this to the UI because it 69:49 will give more specific information to 69:52 lovable about certain edge cases and not 69:55 just deducting credits but thinking 69:58 holistically about how the credit 69:59 management system works okay it's 70:02 updated the states here so let's try and 70:05 run this again and let's analyze lead 70:08 again we're getting a similar problem so 70:10 it's not giving us our output H okay it 70:14 is giving us the output but it's 70:16 throwing us back to the about page so so 70:20 I'll just tell it to keep us on the Run 70:21 tool Tab and then it's going to make 70:23 those edits and then we're approaching 70:25 our zero on here so we can test that as 70:28 well and at that point we want to 70:30 implement our 70:32 stripe payment link which we spoke about 70:35 previously before we'll go and grab that 70:37 stripe payment link and integrate that 70:39 as well it should have updated that now 70:42 so let's go back to tools try the sales 70:44 lead qualifier run tool put in some 70:47 criteria analyze the lead and it keeps 70:50 us on the Run tool it's deducted our 70:52 credits and it's going to give us the 70:53 output so it's now performing exactly as 70:56 we expected let's start integrating our 70:59 payment links here so we're going to go 71:02 back to stripe we're going to grab the 71:05 link from our payment link section and 71:09 we're going to get lovable to set up the 71:11 payment 71:16 links so've added so I've added this 71:20 here so when a user signs up to a plan 71:22 or gets to zero credits gets to 71:25 insufficient credits get them to 71:27 purchase through this payment link and 71:28 I've linked the stripe link it's the $20 71:31 a month plan and should give them 100 71:33 credits in the database the back end 71:35 should all be linked up so we know which 71:38 user is paying and when it's paid so 71:41 let's see what it comes back with here 71:43 it might ask us to set up something on 71:45 the stripe side as I believe we have to 71:48 connect our stripe account to The 71:51 Lovable app here that there's a link 71:53 missing we need to a user to to pay but 71:56 then in our database we need to link 71:58 that user base back from stripe based on 72:01 their transaction to the user in our 72:04 database so I followed up with how does 72:06 this connect to my stripe account and 72:09 know which customers created a 72:12 purchase and it's given me a few extra 72:14 steps to consider we need to have a web 72:19 hook Handler to process stripe events so 72:21 when stripe sends the data that suggests 72:24 a purchase has been made on a certain 72:26 product it should be received back by 72:29 our SAS so that it knows to update the 72:32 database and it knows which user is 72:34 associated with that transaction and 72:37 then along with that is the customer 72:39 linking that I just mentioned so to 72:42 implement this properly we would need 72:43 your stripe secret key do not share your 72:45 stripe secret key with anyone and a 72:47 stripe web hook endpoint set up and then 72:51 Edge functions to handle the web Hooks 72:53 and update your database would you like 72:55 me to implement these stripe integration 72:57 components all we need to do is say yes 72:59 so it understands the steps we need to 73:01 take and we can just ask it to take 73:03 those steps for us and we're going to go 73:05 and grab the stripe secret key here and 73:10 alterate way it may ask you to do things 73:13 is to use the product ID which if you 73:15 click into the product you can just grab 73:18 from these areas and say okay this is 73:20 associated with the AI SAS 100 credits 73:23 product as well instead of of using a 73:25 payment link and it might ask us to do 73:27 that so we've come back here and it says 73:30 we need to create all of this stuff for 73:32 you first I'll need your stripe secret 73:35 key again I reiterate do not share this 73:37 with anybody but it links to the stripe 73:40 API key page here so once we've done 73:42 that it's told us to add the API key so 73:44 I've take gone to that page got my 73:46 secret key come back and just pasted it 73:50 in here we can submit that and now it's 73:51 going to save it in the back end 73:53 securely for us let's move on to the 73:55 next step it asks us to do so while that 73:58 processes just to be clear this is the 74:00 stage that we're at now so we've built a 74:03 front end and we've got the automations 74:05 working on the back end it's now 74:07 deducting Credits based on our usage 74:11 however when we run out of credits it's 74:13 going to force us to sign up and pay and 74:15 right now we are connecting stripe so 74:18 that stripe can interact with our 74:20 database which can interact with our 74:22 front end and everything is synced up so 74:24 that our customers are getting a good 74:26 experience and that also you can sell 74:29 these SAS automations or the process of 74:31 running them in the back end to other 74:33 people great so it's taken our stripe 74:36 secret key and it said it's created a 74:39 complete stripe integration with the 74:41 following features it creates stripe 74:43 checkout sessions for subscriptions and 74:46 onetime payments it handles stripe 74:48 events so whenever stripe receives an 74:51 event like a customer's paid it will 74:53 effectively handle that by receiving 74:55 them back it verifies a user 74:57 subscription status with stripe so 75:00 everything we need it to do with stripe 75:03 it's going to do for us and it also is 75:06 integrated with our existing stripe 75:08 customer ID field in our profiles table 75:12 if you remember in the super base table 75:14 and adds credits to user users accounts 75:17 when they make payments the UI updates 75:19 it's connected the pricing page to the 75:22 checkout flow and updated a billing p 75:25 page to use real product IDs so they 75:28 need a couple of things they need us to 75:29 set up a stripe web hook in the stripe 75:32 dashboard to point to our superas web 75:35 hook endpoint so basically we need to 75:38 connect our database so it updates with 75:41 stripe where we detect that a user has 75:45 made a payment and we need to get the 75:47 web hook signing secret from stripe and 75:50 add it in as a secret in the super base 75:52 with the name stripe web hook secret we 75:55 go through all of this and replace the 75:57 placeholder stripe IDs with your actual 75:59 product price IDs from your stripe 76:02 dashboard and it's actually linked us to 76:04 all of the relevant resources let's 76:07 let's first set up a stripe web hook in 76:09 our stripe dashboard that points to our 76:13 superbase web hook endpoint so we'll 76:16 click on the stripe web hook setup here 76:18 we're going to create an event 76:20 destination and then we need to find the 76:22 endpoint URL and then endpoint 76:25 is effectively a URL that this is going 76:28 to connect to and it's going to hear our 76:31 stripe events so every time customer 76:33 pays superbase is going to hear and 76:37 receive data from stripe and that will 76:40 update the database directly so we'll 76:42 open the edge functions on superbase and 76:46 it gives us these different web hook 76:49 endpoints so we're going to connect it 76:52 to this URL or this end point and I just 76:56 go back there and there's a little copy 76:59 tab there so I'm going to copy that I'm 77:01 going to take it back here and we're 77:03 going to connect it to here and we'll 77:05 just give it a description here of 77:08 superbase stripe we also need to select 77:11 which events it's going to listen to so 77:14 stripe has a load of different events 77:17 for different things that we could take 77:19 actions upon but we're actually just 77:21 looking for checkout session completed 77:25 so when a checkout session is completed 77:27 it has loads of sub events that we'd be 77:29 monitoring and actually just detecting 77:32 so whenever a checkout session is 77:33 completed it's going to send that event 77:36 and we'll just go down there to add 77:38 events add endpoint and that should now 77:41 be set up we'll go back to lovable and 77:44 we need to get the web hook signing 77:45 secret from stripe and add it in as a 77:48 secret in the super base with the name 77:50 stripe web hook secret so we're going to 77:53 go down here to Edge function 77:55 secrets and it's going to pop up with a 77:57 page with all of our secrets here and 78:01 they're redacted at the end here so 78:04 we're going to need to put in like it 78:05 said stripe web hook secrets and then we 78:10 need to get the signing secret from 78:12 stripe and add it in to that value so 78:16 you're going to come back into your 78:18 stripe web hook page where you've 78:20 created that here and you're going to 78:22 click reveal you're going to take that 78:24 back and copy that in in just a sec so 78:26 back in super base you're going to put 78:28 the stripe web hook secret and you've 78:29 copied there that is effectively 78:32 connecting super base to Stripes so we 78:34 hear those event and the secret is used 78:37 to validate that it is receiving them in 78:40 the correct places and somebody's not 78:42 intercepting those events and then 78:44 finally we need to replace the 78:45 placeholder stripe price IDs with the 78:47 actual product price IDs from your 78:50 stripe dashboard so let's go to our 78:52 stripe dashboard let's go to payment 78:55 links let's open this link here and 78:58 within the products we will have an 79:00 individual product ID and I'm just going 79:03 to copy that there and I believe the 79:06 price ID is the product ID here and then 79:08 it's saying once set up users can 79:10 subscribe or purchase credits and your 79:12 system will automatically track these 79:14 transactions and add credits to their 79:16 accounts so this should be all set up in 79:19 a second and we'll be able to go in and 79:21 test that okay so the product idea is 79:24 not not not the same as the price ID so 79:26 we'll go back into stripe go to products 79:29 go to the pricing section and it should 79:30 start with price and actually if we're 79:33 in the product here in the price you can 79:35 click on here and copy the price ID so 79:39 we'll copy that back into lovable and 79:42 everything should be set up ready to 79:43 test so what we're going to do is go in 79:46 and use our sales lead qualifier twice 79:49 and hopefully that starts to push us 79:50 towards the checkout flow once we run 79:53 out of credits and just to reiterate 79:55 this is all on a test data account so 79:58 it's going to let us purchase without 80:00 putting any funds forward when we switch 80:03 this to a production environment you 80:05 need to set this up again with those 80:07 different credentials or just update 80:08 them in the relevant places we have here 80:11 okay great it seems like it's ready to 80:13 test so let's start testing again let's 80:16 insert the LinkedIn URL here and we'll 80:19 run it twice to effectively run out of 80:22 credits oh we run it once Okay so great 80:25 so it's shown us this insufficient 80:26 credits you need eight credits to use 80:28 this tool but you only have two credits 80:30 that's really nice that it's um set all 80:32 of this up and also linked us to the 80:35 purchase credit so let's try and go 80:36 through this purchase credit route and 80:38 it's sending us to the payment link page 80:41 let's see if it works so we'll say it's 80:43 getting stuck on the payment link page 80:45 from the purchase credits warning button 80:48 it's probably not linked it to the 80:49 correct payment page so let's just get 80:52 that fixed great so it says it's updated 80:54 the purchase credits button let's give 80:56 it another go redirecting to payment 81:00 error creating checkout session you have 81:03 to be very patient with a tool like 81:05 lovable but don't forget that what 81:07 you're doing is stuff that would take 81:10 you if you were a software developer and 81:13 considerably longer amount of time to 81:15 actually do this but each time you run 81:17 one of these changes it's going to take 81:18 about one two minutes uh to update you 81:21 test again Etc but what you're replacing 81:24 here 81:25 is a very high development cost and you 81:28 can get a SAS prototype up and running 81:31 for very cheap so it's definitely worth 81:34 the time investment and it's really cool 81:36 and I'm very impressed with the outputs 81:39 here there were always tools like 81:42 webflow WordPress Wix Etc that could 81:45 create good front ends but tools that 81:47 integrate that front end with the back 81:49 end were very few and far between and 81:52 often had a big technical barrier but 81:54 using something like lovable completely 81:57 takes away that technical barrier you 81:58 need a little bit of understanding of 82:00 architecture but other than that you can 82:02 do it just by prompting like you would 82:04 with chat GPT so very impressed with it 82:07 so lesson learned here that the issue 82:10 actually was that the embedded window 82:13 here inside lovable was not showing the 82:17 stripe page but when we opened up the 82:20 page here we go to tools we try the same 82:23 process we purchase credits it actually 82:27 takes us straight to the stripe page so 82:30 just a little bit of a learning there 82:33 and what we'll do is we will just enter 82:35 some test details and we will hit pay 82:38 there and you can see that this is in 82:40 test mode anyway and that's going to 82:42 process and hopefully that will update 82:44 in all areas it will update our database 82:47 which will receive a notification that a 82:49 purchase been made so it's thrown us 82:52 back to the login page here which is not 82:54 deal but let's go to the database and 82:57 have a look to see if anything is 82:59 updated so we can see that the database 83:01 is not updated it's also thrown us back 83:04 to the login page and we've got two 83:07 credits still up here so let's go to our 83:10 stripe developer portal and inside the 83:13 logs we can see everything that's 83:16 happened so I can see that the checkout 83:19 session loaded correctly and then the 83:22 payment methods end point was contacted 83:25 suggesting that the payment had gone 83:27 through if we go to customers we can see 83:30 that there is a customer in here with my 83:32 name and it's saying that the payment 83:34 had succeeded and it gives us details 83:37 around that if we go back to our data 83:40 base we should have a striped customer 83:43 ID in here and it's still showing as 83:45 null so what we should do is check in 83:47 our Edge functions in super base and if 83:51 we click into the stripe web hook we 83:53 should be able to to see that something 83:56 has happened here and it has shown that 83:58 something has been received so there is 84:00 a connection there let's go to the logs 84:03 and it's showing that there were a few 84:05 errors so great so it has actually 84:07 received the information and what I'm 84:09 going to do is just copy that we're 84:13 going to go back to lovable and we're 84:15 going to attach the image of the error 84:19 so I'm going to say went through on the 84:21 stripe side super base seems to have 84:23 received it like we saw but then shows 84:25 this error let's see if it can help us 84:28 resolve this and it seems to think it's 84:29 fixed the issue now so let's go and try 84:34 again in the tools run the tool purchase 84:37 the credits enter the test details and I 84:40 probably didn't say before but when 84:42 you're using a test account this is a 84:45 well-known test card information so 84:48 424242 Etc you can put in any dates but 84:52 it's a VISA test card that you can use 84:56 it might decline other cards and we'll 84:59 process that there so it's thrown us 85:01 back into the login page let's refresh 85:03 the database same problems so let's go 85:07 back and check what's happened here and 85:10 this time it doesn't seem to have 85:13 received it 439 okay it has received it 85:16 again and in the logs and if we go to 85:19 the overview here I can see that 85:23 something has happened at it's got a 4xx 85:25 error which usually means it can't find 85:28 something so if we go to invocations 85:30 over here go to this which is my most 85:33 recent run and I'm just going to copy 85:35 all of this metadata down here take it 85:40 back to lovable refresh the page tell it 85:43 it's not working again just refresh so 85:46 got a 400 era in super base copy that 85:49 error code in and we'll see if it can 85:51 tackle that so we seem to be having the 85:53 same issues here so I've come into the 85:55 stripe dashboard and all of the payments 85:58 that we've tried have succeeded in 85:59 stripe so it must be as a result a 86:03 connection issue between the stripe 86:05 dashboard and our superbase database in 86:09 the logs that we were looking at before 86:13 it seems to show that things were 86:14 executed but they're always not found so 86:17 with some persistance I've gone back and 86:19 I've just refreshed all of those things 86:21 we did before not sure why it wasn't 86:23 working I think we might have got it 86:25 working now so there's six credits in 86:29 the account we're about to top up we're 86:31 going to add again some credits and 86:35 sometimes you just need to refresh and 86:38 let's see if that updates and tops 86:41 hopefully tops up our credits here so 86:44 it's thrown us back but we can see now 86:48 hooray 106 credits so we'll go to the 86:51 tools and we're now able to use it and 86:54 and if we go back to the database we can 86:56 see that it's 106 and we can see it's 86:59 put our stripe customer ID in there as 87:01 well so that is a success we know that 87:04 it stops US purchasing when we've not 87:06 got sufficient credits and now it's also 87:09 connected to our stripe customer ID so 87:11 hopefully now you can see that it's 87:15 actually extremely easy orbe it takes a 87:18 bit of time with prompting and 87:19 reiterating and testing to set up a full 87:24 AI SAS Marketplace and hopefully you've 87:26 got some idea of the products that you 87:29 could host on this Marketplace so today 87:31 we just set up a sales lead 87:33 qualification flow as well as an extract 87:36 from PDF flow but looking at a 87:38 Marketplace like agent. a you can see 87:41 they have thousands 98 pages of 87:45 different flows we've got things like 87:48 pulling recent LinkedIn posts a meme 87:51 maker fit to purpose image generator 87:54 industry analysis so all of these things 87:56 could be replicated with an N backend in 88:00 a similar interface to this it works 88:02 with n in the back end and your users 88:06 could pay and come to one platform to 88:08 access all of these tools so the 88:10 possibilities really are endless if 88:13 you're considering implementing this for 88:15 yourself obviously you followed the 88:17 video through but we've got all of the 88:19 different prompts implementation plans 88:22 the setup guide for stripe payments the 88:26 credit management system so we just did 88:28 some basic stuff with credits that 88:31 wouldn't stand up in practice so we've 88:35 got a full guide here on the different 88:37 things that you might need to implement 88:38 with loveable and test to make sure that 88:41 you could actually run this in 88:43 production thank you so much for 88:45 watching give this a like And subscribe 88:47 below if you enjoyed the content and 88:49 want to see more like this