Automate Business Plan Creation: Integrating Market Research with Zapier & ChatGPT
Let's build with Zapier and AI (100+ videos)📅 2023-10-27
Transcript
[00:00] welcome back to Corbin AI where I'm showing you daily how to start leveraging artificial intelligence in your personal and your business life and you clicked on today's video as we're going to learn how to automatically analyze a market research report that's going to Output an entire business plan based off the data we find in that PDF if you like this kind of content and you want to see more content related to PDF analyzation you can check out our channel here without further Ado let's go to jump in here we're going to go and jump in over to Google Drive as this is going to be our main trigger event essentially the way we're going to work this is we'll provide the search report
[00:30] into a folder and then everything else will be automatic and the way we're going to do that is we're going to use zapier we're going to use Chad gbt and then just free services provided by Google so the research report we're looking at today here is a free one that I found about hardwood hardwood floor manufacturers as you see it is a sample report but in other context you can get a research report on any Market real estate markets artificial intelligence e-commerce whatever the market research is on I mean you go all the way to Vapes
[01:01] you can go all the way to you know a restaurant industry whatever the market research research is based on you're going to be able to use the same flow and get a potential business idea with an entire business plan so let's go to learn how to do that today we're going to come over to zapier here and we're going to go ahead and create a trigger and the trigger here is going to be a new file in the Google drive folder therefore we can identify and distinguish to use a specific folder for this context I also want to point out that you're about to learn how to do large PDF files as you see here this is
[01:32] a pretty large PDF file this is over 27 pages of a bunch of data we're going to have DBT handle all of that information for us so we don't have to lift a single finger so we're going to say new file and folder we're going to choose our courses and the folder we're going to choose is called PDF text let me go and move my audio up here a little bit all right so hope that's a little bit better and we're going to choose our folder here PDF text there we go to continue here we're go ahead test this trigger here and what we're looking looking for is going to be file exist but not shown
[02:03] as if you're familiar with PDFs and when it comes to zap year it won't give the entire output of text but don't worry we got some code to help you we're going to do another Google Drive block here and the purpose of this Google Drive block here is the first thing we need to do is we need to convert this PDF into a Google doc this is going to allow us to grab the underlined data here so we're going to go ahead and do the event of upload file we're going to continue here continue here in our courses account and I want to make sure you know that it's very important and we're going to also do file exist but not sham that the folder that you put this in is not the
[02:34] same folder this needs to be a different folder so we're going to put this in business proposition the reason why is that you'll create an infinite Loop if you chose the same folder because the original trigger on this is a PDF text so if I chose PDF text here then it would you know You' add a PDF it would create a Google doc in PDF text then it's you know another new file and then create it again and then you know infinite Loop you don't want to deal with so make sure the folder you're putting in is a separate folder from here though we can go ahead and convert the document we're going to say true when they continue here you can also give it a special name so in this
[03:04] context maybe we'll just give it a title of the PDF so we know exactly what we're talking about here when it continue here we're want to test this step jumping over to this folder we should see it pop up here it's going to be a Google doc but the reason we're going to make it a Google doc is because now we can actually going to be able to get the underlying data associated with that PDF so if I double click this as you see it might not look perfect sometimes in output but this can take any type of template and you'll get all the data that has associated with this PDF this is great information for us to use so
[03:34] coming back over to here we're going to do two major things the first thing we're going to do here is going to be adding a code block that's going to be giving uh all the relevant information all the relevant data found in that PDF so we're going to run JavaScript here I'm going to go ahead and grab our code block I'll make sure to add these in the description down below so you can use them yourself we going to continue here and if you're familiar with the Channel you already know the very famous code block here so the purpose of this code block is it outputs the entire data for
[04:05] us to use if we didn't have this code block as you'll see when we pull up this we don't have all that information right we only have file exist but not shown so what we want to do here is want to do file text as we want to get the plain text that is found in that Google doc and then the input data this is very important you need to name a text as that as that's how we identify it in our code and then as you see here we up to 20,000 words I'm going to go Ahad and increase this so we output up to 30,000 as I would do 40,000 as I'm not sure how many words are in that PDF but that just basically you know gauges how much data
[04:36] you want to do so typically obviously do a little bit more than you would expect to be found in a PDF so if you deal with PDFs in the sizes of 5,000 10,000 You' probably put 20,000 going continue here test a step and we should get all the relevant information in that document all right so there we go we got all that information as you see it is a lot of information and you might be seeing yourself oh my gosh like I can't read that or it's the the you know the way it structure does not look good don't worry about it AI doesn't care they just retext or they just Red Data so what we do here is we're add another JavaScript
[05:07] block here as having that many words and inputting into a gbt prompt is counterintuitive for two major reasons the first one is that too much data in a gbt prompt as you'll see here is going to lead to worse outputs the second one is that you'll lead to more likelihood to for the automation to break because of delay times or timeouts and stuff of this nature so this is how we circumnavigate stuff like that we're going to go ahead and do JavaScript here we're going to do another code block so the purpose of this code block is going
[05:37] to be to separate the data into separate blocks the reason we do that is so we can kind of compress the data uh using gbt 16k uh in order for us to use later so we're going to go ahead and make sure we choose the right variable here which is going to be text again and the underlying data we're going to use here is going to be the previous result which is the entire data that was associated with that PDF and as you'll see here we have separated so we have three major chunks and we're going
[06:08] to go ahead and separate by 5,000 words a chunk I think that should be sufficient I'm going to go ahead and just increase that or maybe add another chunk here just in case let's go and see if we need to add another chunk here so I'm do this I'm go ahead this right here we just do chunk four and three so this is like their way of creating an array I'm going to text test the step here and we should get three separate chunks here so we got chunk one we got scroll down here chunk
[06:39] two and okay so it's only two chunks so we have actually less than 10,000 words we don't have to worry about the rest two chunks perfect now let's go ahead and compress those chunks so we're going to go ahead and add a CH gbt block here and it is very important after for this CH gbt block you choose the model of 16k as this is going to allow us to compress large amounts of data for specific use cases so we're going to do conversation here and we're going to go ahead and hit continue continue and we're going to do 16k now in theory because of the purpose and the use case of this video is to build out a
[07:10] business plan I could approach this in two different ways the first way I could approach it is maybe give a lot of information on what's typically encompassed of a business plan and what we're looking for so in theory you could kind of find tune a little bit further here but I'm going to take this at a very broad uh aspect so rather than me listing out you know as executive summary business description market analysis organization and management we're going to get into that later in this automation I just think for the initial compressing of the data we want as general as possible so we're not leaving out maybe specific data points
[07:41] we would need so for here we're going to say based we're going to keep it General right so we're going to keep it General because then you can basically upload any type of market research report into this folder and get an output so we're going to say based on this market research report then we'll do uh report title and then we'll do uh report section
[08:11] one and then we'll go ahead and add these little parentheses here we're going to do title here so the title is just going to be you know our original title here and then the section one is going to be chunk one for us so we're going to come down here let go all the way to chunk one there we go we're going to say and we can honestly talk very uh blatant we're going to say compress this data to find what will be relevant for us to make a business plan
[08:41] later I could probably proct do this better I could have probably made this better but let's go to see what this looks like real quick Biz out memory key random 32 character string consistent outputs let's go to increase I don't typically increase tokens as it doesn't follow it but I'm going to increase it just for now we're going to say a th uh purpose of tokens is how large of an output you would be looking for in the context of how many words it's very loose it doesn't typically follow it but sometimes it's just you know good practice to add it just so you can give it more breath okay so based off those 5,000 words found in the first half of
[09:12] this market research report we can go ahead and scroll all the way down here and we should get a reply here okay there we go so based off those 5,000 words we went ahead and got it looks like around 15 major points with some additional information so just to prove why compressing data is so important here we're going to go and copy this and see how many words are associated with that compressed data all right so coming over to our word counter here all right so 440 words so it compressed 5,000 words all the way up to 440 words I
[09:43] think that's like what uh like a 80 80 to 90% compression rate so really good stuff here um now let's go ah and just copy this we're going to duplicate this we're going to give it the same logic here we're going to go ahead and identify that we are in section two of the report and we'll provide just the same deal here and we're going to go do chunk two here and we're going to compress this data even further So in theory we could if you are dealing with smaller reports you can compress one report one 16k block but I wanted to
[10:14] show you the functionality of chunks as uh if you're dealing really large PDFs this is going to be your best bet when trying to do Automation and Analysis and using AI alternatively there's probably softwares out there that cost you you know an arm and a leg every month to pay for it I mean this is using it as direct as possible in our current context so from here we're going to take another 5,000 or around 5,000 words we're going to compress that as well so just to give an example of maybe one of these outputs I'm going scroll down here real quick uh
[10:45] we got um the market is dominated by 3M but they have production and delivery problems CP adhesives is a low cost alternative but api's new formula has advantages over both competitors so just to remind you we are dealing with a market research report about hardwood floor manufacturers I have zero knowledge on this industry so if you do have knowledge on this industry you probably understand a little bit more of that terminology of what was said there from here though let's go ahead and continue the logic so we went ahead and compressed the entire PDF pretty well
[11:17] let's go ahead and take this one step further here we're going to go ahead and do chat gbt here and we have two major options here I can either jump directly into building out the GB T4 logic as a gbt 4 logic is more comprehensive in the context of building out A business plan is what you're going to want to use but the very important thing to understand about gb4 is you don't want to frontload it with too much data even 450 words seems a little much to me so what I like to do in the context where I want to make sure that it is as nice as possible
[11:49] is let's go ahead and uh tailor a little bit more with uh gbt 16k 16k is amazing 16k is very powerful if you know how to Leverage 16k you can do most AI automations than you would expect a lot of services out there you really don't need anymore as you can really do it all on the back end of zapier so from here I'm going to say um based on this market research report
[12:21] summary want to do report summary so if you're familiar with code uh you would know that a lot you know you're making cont you're making varia it's really nice just to speak to gbt in a similar way where it's like you identify that we're dealing with a report summary okay what is a report summary here's a report summary and here is the variable Point uh from here though we can go ahead and add our reply one and then you know we could just add a comma here it's not too important as end of the day they don't read it like a human they just read it like a robot so we provide our entire report summary here and we're going to say generate a
[12:54] general or generate based on this data the following so I'm going to give it a very broad idea of a executive or sorry A business plan so I'm going to copy it over I already kind of got it outlined from chbt here we're going to go ahead and do this so and then we're going to uh let's go ahead and see what this output actually looks like I'm actually curious but we essentially gave an executive summary identified specifically what we want an executive summary business description and you
[13:25] know a bunch of stuff of this nature so we're going to go ahead and up this to 16k this might not be perfect here there might be uh limited amount of stuff I want to do here so actually let me do this real quick so I'm going to go ahead and just put in some words here and for the stuff that takes a little bit more comprehension let's go ahead and well first actually let's go see this output if the output's good we can kind of proceed from here if it's not then I'm going to try a different route here where instead of giving so much analysis to 16k I'm going to put it on four okay so here's what we got we got the business plan outlines and strategies
[13:56] and actions required for an advanced Products Inc to tap into the wood floor manufacturing market segments API is a lead manufactured okay the market for wood floor manufacturing is strong which with an increasing demand of high durability adhesives and bonding techniques although the market is dominated by 3M APS new adhesive formula provides advantages and gives us a general overview so this is kind of more specific in the context maybe more analysis on API um funny enough API is a very common term in software uh from here though let's go ahead and fine-tune
[14:28] it cuz maybe that was the main point of the research right that's what they found in the market now let's go ahead and say based off this research though find me some find me a way to start a new business or a potential in this market so now that we have our output here which is good let's go ahead and add a formatter block this is going to allow us to split the data into identifi identifiable data points that we're going to then be able to leverage with chbt 4 so I'm going to go ahead do text here we're going to do an event of split so split is going to be a very important tool to use in the context of anytime
[15:01] using an AI prompt this is used a lot in code as well when we're dealing with outputs that are reliable from open AI sometimes you want to use JavaScript as a formatter in order to ensure consistent outputs at scale because sometimes AI can uh throw in some extra stuff in there that is not necessarily needed so we're going to do the input of the previous reply here we're going to do their documentation which is going to be new line and this is going to Output every single uh you know data point as a set uh as a separate field that we're going to be able to use later on so I'm expecting a executive summary and then
[15:32] some sentences here okay perfect that's actually really good okay so now that we've been able to split up this data into separate points here one chbt block for basically 10 different data points amazing we can go ahead and add the logic of chat gbt 4 so I'm going to go ahead and just showcase one block of this and one execution of this and then we're going to show you how to push it towards a Google doc in theory though you would repeat this kind of Logic for all the other different blocks here that you deemed necessary that you want
[16:03] analysis and further indepth you know research on so for now let's go ahead and give like the most important one right so we can go ahead and say uh okay we just do the executive summary and then you know we're going to say based off this give us uh you know a good business idea right but in theory you'd probably go through all this information loaded for this Google doc here we're going to say based on this business
[16:33] executive summary semicolon parenthesis uh Spell executive correctly and we're going to do this generate a new business idea that will leverage this information all right and I may need to restr structure this prompt we'll see what it is but notice we went from a over five you know 7,000 to 9,000 word
[17:06] dock to now just a nice concise data point uh from gbt 3.56k press compressing therefore now we can feed this into gbg4 which is a very much more comprehensive model and you know provide it with very relevant information that we care about so let's go and see this output here okay so we got a new business idea before we read it let's go Ahad and jump into just actually showing you how to push it towards a Google doc so we're going to do Google doc here um if we come over here and you know this could be push to a presentation as well
[17:37] it's a little bit more complex you have to build a preset template in Google Slides but in theory you can build out a whole you know presentation off a market research report which is really cool uh from here though we're going to say create document from text and we're going to go ahead and just input our relevant text here so I'm going to choose our courses account and from here we can go ahead and put our text so we just got there so in theory though as well to keep in mind we can actually um use HTML here so in theory I could have had it so that the output was
[18:08] in HTML so it's formatted a little bit better and honestly let's just go ahead and do that real quick so we're going to go ahead and take the same logic here we're going to say uh formatter output in HTML and as you know since we want consistent outputs I'm going to refresh the memory key so we start fresh here and let's see what our first out put looks like and if we are satisfied okay so we got a limited amount of HTML I probably could have said you know add bullet points formatting you know add a little bit more pizaz here um have
[18:39] bullet points and uh bold important words add a little Pizzazz here make it look bit better so when you have it all interconnected and maybe have every single block that's doing analysis you can have it output HTML so the output looks a little bit more legible A little bit more professional and stuff of this nature here we go so we got some cool HTML formatting here let's go a and check it out live here so we're going to go ahead and put in this reply that we just received here and this should be good
[19:10] let's go ahead and put it in a folder that is specific to us which would be the business folder that we created here business proposition and let's sayit continue here test this step and we should see a new Google doc show up here based off the information and the data of this market research report we went ahead and got a Google doc that is going to be an analysis from gbt 4 here in HTML formatting and here is the new business idea obviously we could have got this more complex added more blocks but here we go so we got bearing in mind invaluable business executive summary information we can launch a new business
[19:42] venture focused on Staples of engineered wood flooring adhesives this business will capitalize on untapped potential and fast falling engineered flooring adhesives by leveraging API Innovative and high performance adhesive formula offering Superior durability from prevailing Market players like 3M NCP adhesives product development target market penetration building strategic part Partnerships competitive positioning and market research now we could of got more in depth here as we have all that data that was Associated found in these chunks found in these
[20:13] outputs but for now you have successfully learned how to take a data research report and all you would need to do is simply drag it into that folder and then have automation deal with the rest and give you a structured out HTML output into a Google doc so if you feel like you learn something make sure to leave a like it's complete free and helps us here at corbon AI if you're more interested in learning more about PDFs and how to leverage them check out just recent videos here or just type in corbon a PDF we've done a ton when it comes to PDF whether it's creating uh from an invoice to excel analysis on General topics and PDFs uh we even did
[20:44] one where we showed you taking a client's information and automatically post a call creating a l a personalized uh lead PDF right crazy stuff here I'm going to leave a playlist at the end here called AI for tools for mod business you can check out a bunch of new stuff here without further Ado I'll see you in the next video thanks for tuning in and yes surprise I'm an AI Avatar make sure to explore more here at Corbin AI where we demystify AI for your personal and business life until next
[21:20] time