Doing a certain process so um for us U measuring that like how many hours they are saving in a given month um it's one very important metric so for this specific um kind of system that we have built so we helped our clients save more than 20 hours uh on a monthly basis because these of course was a very time consuming process and we're going to explain you more about this um and also.
As we are talking about this system of building newsletters of building podcast episodes um not only we have the client save um those hours but also we have the client we have the clients impact a big audience so um like everything that we created here all the newsletters all the all the podcast episodes are um kind of going out and reaching a big audience of more than 50,000 people that's that's.
Significant and thank you thank you very much and for everyone that's exactly what we're going to be going through today this is going to be a very practical webinar how did Ed built those automations in order to uh achieve those results in order he was automating uh the generation of newsl and also podcasts uh so just to quickly go through the the agenda for today so we're GNA go very very quickly through an introduction about me and about a.
Agency and then we're going to look at that challenge that his client was facing and then we're going to jump right into a live demo showing how did they uh achieve those uh results very quickly just some uh house housekeeping if you do have any questions believe do not leave them in the chat leave them at the Q&A we have a Q&A tab so if you have any questions please uh leave them there at the end of the present presentation we're going to answer your questions the webinar will be recorded no worries the webinar will be recorded going to be.
Related to uh Automation in general that's all let's Jump Right In so uh hopefully you all know about Mak I'm not going to spend a lot of time speaking about that make as an automation platform that enables you to automate your workflows no matter how simple or um or complex they are uh and it enabled you to connect so many different uh apps we have more than 100 uh 1,800 apps at uh at the moment and so many customers.
Have been using make for years and years and years uh that's all now I'm gonna move it to it to uh give a quick in about access automation yes great so yeah um I've been building automations for the past um eight nine years and um for the past three years I've started to do that for other um companies as well so I started X automation which is an AI automation consulting firm so we build automations using automation tools like like make.
Using apis um web hooks customer scripting with python and JavaScript and a lot of different AI model as well um we share a lot of content online especially on Twitter linkage in YouTube um we have our own newsletter as well so we are impacting right now more than 30,000 people with um everything related to Ai and in automation we have worked with more than 100 clients and we are a team of um seven people here awesome.
Thank you thank you very much uh thank you very much it's and actually where are you joining us from where are you speaking from today speaking from s paalo Brazil yeah because like I did not see anyone from Brazil commenting with the chat so uh if you are from Brazil please yeah do do write it down so right now so yeah let's go to that challenge that you were trying to solve so what was it yeah so um so yeah we started working.
With um a client which is a big um real estate company and um one of the challenges that the client was facing uh was the process of creating and sending out newsletters for their client base for their lead based client base um so they wanted to kind of make sure that that client base were keeping like being nurtured um start engaging with the client base and they were using newsletter for that it was a monthly newsletter that they were sending out.
Posts Related:
- Noob vs Pro Excel LOOKUP Values with two spreadsheets
- 3 Easy Ways to Create Your Power BI Calendar Table / Date Table
- Excel 2019 VBA Intermediate Tutorial (Part 2 of 4)
Kind of by curating and getting the most important news articles related to real
Estate especially related to commercial real estate for local um like business in their own City but also National um um local new national news and they were kind of getting the content searching for those news articles they were um writing all the summary for all the um new newsletter articles that they were.Um inputting to the email and also sending out uh the template sending out the actual campaign on on this case on mail chain so all that process of getting the those new articles writing out the content for the newsletter and actually sending out the the newsletter were like taking more than eight hours a month um for my client which was of course was a very timec consuming and there are some months that the client was just just didn't have time to to do.
That so um he ended up skipping one month or two and of course that was very bad in terms of client engagement um um for his for his business and another case study was about the podcast so the same thing the client wanted to create a way to kind of have a more skillable way to create podcast episodes Rel related to um different subjects mainly about the stock market and he wanted to have a.
A daily and weekly podcast show um getting the most recent um news articles about the stock market and creating those episodes um in a more automated way because of course um if you ever try to uh put together a podcast episode or even newslet you know that it takes time right so um yeah so that I I'm I'm I'm surprised that you were spending only8 hours per month on creating this new I would I would expect.
Actually they are they were spending much more uh than that and I think this is a use case that a lot of companies do have like newsletter whether it's collecting content from outside that not they are not the one who are creating and then just sharing these news with their audience or it can be your own content right so we at make we are sending newslet on a monthly basis as well and it does take time to collect all this content and then writing and summarizing that content and sending it out so I think this is a very relevant use case for uh for for the majority of.
The companies and that's why like I was very very excited about about this webinar so all right so that's how did you how did you approach that how did you automate this process yeah definitely um so the first part so for the newsletter automation we kind of divided this process into two different scenarios two different make scenarios the first one we connected to fidly to fatch the news articles um that were being published um on the certain on the categories that we needed to to.
Track right again so related to real estate related to commercial real estate uh to the let's say like business economy in general um so we connected to fidly and then we were getting those articles um um kind of through the the automation right so the first step was to connect to F to make sure that we were getting the right articles the the other step was to classify and rate those articles using Ai and of course using open AI so we we wanted to use.
Open AI to evaluate each article to make sure that it was a relevant article so what we did is basically created created a prompt to evaluate the impact of that article in the general um category so let's say that um we asked the we asked chpt we ask open AI from a scale from one to five how impact was that news article related to the bigger category so how impactful was that arle related.
To the commercial real estate market in general um and then once we F through the AI then we added all those news articles into the database with everything that we needed from those news articles so meaning the URL the the title and the rating system that we um created using the AI so that was the first scenario and then of course we're going to do a quick live demo I'm showing you each module here and then the second part the second scenario was.
To once a month create the actual newsletter so once a month
We got all the news articles that were relevant that had a rating of three or higher and then we used AI to create the summary of those news articles then the summary that we were adding to Mayo Chan awesome great uh so that was that was the newsletter are we going to jump into the demo right now yeah let me let's do.It great so this is the first scenario okay so as I mentioned this scenario we connected to fidly to fatch um the news articles that were being uh published um um of course automatically so uh we are getting we fetching all those news articles from fidley um as the trigger point of this make scenario okay maybe maybe sorry to interrupt you here.
Because like we know we have few a few people who not necessarily have a lot of experience with me right so if we if we zoom out a little bit here and just uh because this might look very complicated for for some people and just explain how how do you build those bubbles how do you build this this this bolts for someone who never never seen that before yeah definitely of course um this is a um a bit more simplified version of this of course you can you can still.
Uh make it more um more complex if if you want because make allows you to to do that but also it's very simple because all you have to do is to First create the first bubble so the first module um as the trigger point of this um of this scenario so you're gonna go to um you're gonna go to you gonna search for app that you want to connect here to to the scenario in this case it's fiddly and then you're going look for the the type of trigger that you.
Need that's going to activate all this scenario right so um in this case what we are doing is we are using the watch articles trigger so here you have all the available triggers for this app which is fiddling okay and then of course we selected the watch articles um trigger to activate this scenario here right great so with the trigger um in place.
Then we um created our first action so the same way that you have the the first module as the trigger then you're going to have the other modules which going to be all the actions that you want to perform on this um scenario okay so the first um the first action is with open Ai and here here is of course is basically you are connecting to open AI you're connecting to one of the openi um openi modeles so you can pass a prompt so you can get a response so instead of.
Going to chat GPT um website and interacting with chpt you are using the openai API so you can do that using automation okay um and sorry when not no that's perfect go okay cool um so yeah so for for this case we wanted to First analyze if the news articles were was related to a local news or a national news um for this client we were looking at news articles related to Seattle so.
That's why I wanted to identify if this was related to a local um real estate news related to Seattle or if it's a national um um piece of news okay because then this is going to be important on uh when we go and start building the summary they we're going to add to the newsletter um and then we we can then we divide it into local and National and how we are we are dividing to local and National so we are first using a a router it's basically a way that you.
Have to divide the the make scenario into different paths um so you can perform different actions according to one output of a previous um of a previous module here so in this case um I wanted again I used um open AI to analyze if this was a local or a national news so I created a router then to divide the mix scar into local or national news using the output from CH.
GPT from open AI to see okay like if the if the output of the openai contains the word local then it's going to go to this to this path here if the output from OPI contains National you're gonna go to this path here again so I can we can divide and better organize our make scenario here yeah okay um and then after doing that what we are just doing is rating the.
News articles in a scale from one to five in terms of the impact of that news articles into the brother um category so the impact in this case uh for the national path the impact of the the news articles into the national um real estate market for example or the national um business Market okay and then of course I just want the AI to reply with the rate so from one two.
Three four or five because then I'm going to add um then I'm adding again as I mentioned at the beginning I'm adding that news articles with the rating to our database okay so then we're going to use the database in the second scenario so that's that's why it's important to just get the number from the chat from the from the open AI um module and I'm adding another field here to avoid.
Adding to to the database any article that is rated as one so there is no impact so that is not irrelevant article for us to consider for our newsletter makes sense makes sense great um so moving on to the second scenario here let me do a quick zom great so for this second scenario this is the the scar that we are doing to actually build the newsletter to actually build um the email that we're.
Going to send out to our audience okay so what we are doing is we are fetching um we are fetching the articles that were um published in a given date so the past 30 days and also they had um the rating the that rating from AI from three and or higher three four or five five so again we are only considering relevant articles here for our um for our newsletter okay so you are able.
Basically to set the trigger to go look at Google sheet and only look at ENT that has been added in the goal sheet within the specific time frame and your example is 3050 days but it can be like s days or 14 days depending how frequently you are sending this news dat correct yes correct of course like um this is just the the example of the the database that we are using so we adding The Source the URL the headline the category the date if it's local or.
National and the rating right of course that you can use makes data stores to do that but we created the Google Sheets um database because the client also want easy access to to to to the to the database so that's why um it was another way to do that but again you can you can do do that even inside make using data Stores um um to kind of manage all those those articles ratings and then to fatch them into this second scenario here okay.
Um yeah and then what these second scenario is doing not only fetching um the the actual articles from the from the database like I mentioned then we are just combining them into a tax aggregator so we are basically putting together the headline with the URL on the tax aggregator so just putting the same kind of structure both the headline and the URL of that news article and then we are using again open AI to write.
A summary um of that news newsletter okay of that of that article sorry first before writing the summary we need to select which article are we gonna um select to um to go to the newsletter because let's say that we have 10 articles that have a rating with three or higher we need to select the three most important ones and we are also using open AI to select those articles.
Okay because we need to make sure that we're only adding to the newsletter the most relevant piece of articles that we have on our database so we are using open AI to to help us select um those more those relevant articles and then of course we are um on the prompt we are asking um for the open AI to actually write the summary of each article that we are selecting a brief summary on a specific writing style that the client.
Needed okay that's and this is important so we can keep the newsletter with the same writing style as uh um my client so it can sound like as the same person that is that is writing that newsletter yeah okay um and then of course we are just on the prompt we are also instruct the um AI to have the output in the HTML format because that's how we're GNA uh pass through the mream.
API um and then after writing each article's um summary then we're going to write an all overall summary so that's the summary that's gonna that's going to sit on top of the newsletter that's going to open the newsletter so it's a summary of all the summaries um of everything that's going to be covered on the newsletter so it's a brief summary that we are adding here the introduction to the newsletter exactly it's the introduction thank you yeah um and after.
Having the introduction then we are sending to macham so we are basically crean new campaign on machain using the template that the client was using so keeping the same design um the same template design and then just adding the content using um the the API um their MCH API here yeah yeah and like definely I think like something that is very important here that in this case you using M CH but like obviously you can switch that to whatever email uh uh.
Campaign tool that you are using right so if you're using a different a different tool you can uh you can use that the same thing for Google sheet if you don't want to use Google sheet you can either use data stores at Make Or Another another another tool another tool as well yeah correct good point so again this this system can work with a different database like data store this system can work with a different um email marketing provider it can be mailchip it can be convert K it can be active campaign can be any other one that you can use to create those um.
Those email campaigns automatically it can even work with Gmail so you're if you're just want to send out um a very kind of basic and simple newsletter using Gmail that's probably the easiest way that you can start building newsletter right so if you want to start uh building an audience if you want to start talking to an audience engaging the audience that's an a very easy way that you can do to start getting the content and sending out the emails to your to your audience so um yeah you can even work with Gmail one thing that.
We are doing internally here at Mech uh that we are sending it to slack it's not even email just internally uh for for the team who are following some specific topics online uh so you like and every week like this is the summary of the news that was published about this specific topic let's say AI for example and we just sending a slack message every week in one slack channel so there are different applications that you can uh you can use this for and also like for me what was really uh uh interesting.
Is the fact you see how many open AI modules that Ed was using here right so normally you don't need to go to chat gbt and then write a prompt and go write another one actually you are automating that your once like uh every step you're going back to open ey automatically to ask to ask another uh question or add another prompt and it gives you uh another output and then you use this output to uh in in The Next Step which is uh definitely looks like this is how you can uh uh make use of AI rather than.
Just going there and chatting with it yeah 100% And and this is something that I always tell my clients like um if you are just using AI you are still doing manual work uh you're still going to the chat GPT website you're still including like the the data you still um writing down the prompt and everything so you it's still a manual process right but when you start combining AI with automation that's when you start building some powerful systems as you can see I'm using AI uh to create the.
Final output of this process which is the the newsletter but I'm also using AI um as a way to organize the scenario like as a way to filter out the the things that we don't need and as a way to divide into different paths so AI is not only you being used here as the final output but also to better organize the entire system the entire automation here cool no thank you very much again just reminder for everyone attending if you have a question please add it to Q&A.
Tab we're going to answer them at the end and also Ed will be sharing uh the blueprints if you're interested you're GNA be sharing the blueprints as well so you can uh C it in all right awesome so what's next great so next is um the podcast automation which is U very similar to the to the newsletter one so um for the first step we basically use the same structure uh because as you as you guys.
