Leveraging AI

234 | From initial idea to Business Clarity: How AI Transforms research, Visualizes Insights, and enables data driven decision making

• Isar Meitis • Season 1 • Episode 234

Share episode

Are you missing out on smarter, faster decisions because you're underutilizing AI?

Most leaders know AI can help, but few truly know how to make it work for complex research, data analysis, and real-time strategy.

In this solo episode, host Isar Meitis breaks down a real-world example: helping his son analyze the rise of Marvel for a school project. The catch? He used AI to do it all—from research to data visualization—and the result is a masterclass in how business leaders can leverage AI for actual impact.

If you've ever wondered how to turn AI into a legit tool for decision-making (and not just content generation), this episode is your blueprint.

In this session, you'll discover:

  • Why deep research mode in ChatGPT and other AI tools changes the game for analysis
  • How to build complex, multi-source, multi-scale data visualizations in minutes
  • A real use case showing how AI can combine apples, oranges, and analytics to support strategic decisions
  • The step-by-step prompt strategy Isar used to turn chaotic data into business clarity
  • How to scale AI thinking inside your organization (or even your home!)
  • Key lessons on teaching others (kids, teams) how to use AI ethically and effectively
  • A fast, repeatable system to turn data into insight without hiring a data science team

About Leveraging AI

If you’ve enjoyed or benefited from some of the insights of this episode, leave us a five-star review on your favorite podcast platform, and let us know what you learned, found helpful, or liked most about this show!

GMT20251020-153718_Recording_avo_1280x720:

Hello and welcome to the Leveraging AI Podcast, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. This is Isar Metis, your host, and we got a really cool episode for you today. It is based on my own per. Personal experience from the past few days, and it's showing a lot of really critical aspects and how to very easily use AI to do proper research, to do data analysis, to present the data in ways that can support data driven decision making process and all of that while potentially helping other people, your kids, or other employees, learn how to use AI effectively. So because I went through that. On my own. I thought that would be extremely powerful to share all these things in a quick, very targeted, practical episode that you can all benefit from. So, a little bit of background on how this all started, and then we're gonna dive right into the content itself and how to do the process. My son was working on a school project. He's in high school right now, and the project had to do with a very detailed, thorough analysis of the success of Marvel. He's learning it in a class about media, but it was a very large and detailed project that he had to do over, uh, over several days. And he came to me with a question, and the question was, how can he. Check, what was the impact of, what was the impact of the issue of the first Marvel movies and then the acquisition by Disney on the success of Marvel. And obviously the first thing that came to mind is how can we use AI in order to do this? And I found this as a great opportunity for me to test some of my existing skills while teaching my son how to properly use AI in a way that is ethical. Meaning not having AI do the work for him, but just use AI as a way to accelerate and improve the research process. As well as a way to explore different ways to present data in support of this, in support of very specific requirements And the very first step was doing a little bit of research, trying to figure out what are different aspects in which the distribution of Marvel capabilities is actually generating revenue and showing growth. And that was the initial research. And that showed us that they're actually growing on multiple different vectors, including movie theater, ticket sales, obviously, but then also TV shows that started popping up, gaming on computers and gaming consoles, actual comics, sales, and several other aspects. And the idea was how can we see the growth on all of them combined to see a trend and to see how each and every one of these are. How each, every one of these vectors are playing a role in an overall transmedia strategy that has led to the growth of Marvel from being just a niche unknown brand to the global phenomena that we know them today. Now, while this is a school project, this is perfectly aligned with more or less every new strategy that you want to think of implementing for your business. Because if you want to change a direction or add new process. Or add new services or products to your offering, you need to investigate multiple different aspects of how that may or may not be successful and compare them one looks to the other when they may or may not even be similar things. You need to find a way. To combine apples and oranges in order to make the right decisions. And this is why I find this so attractive. So let's start looking at the actual process that I did with my son, step by step on chat two pt, and I will share with you what I'm seeing on the screen. If you want to actually see the screen and you can right now, meaning you're not driving or walking your dog, there is a link to our YouTube channel on the show notes, and you can jump and continue watching this on YouTube. The first thing that I did is I attached this image that has a different research that was done by a third party provider that is showing the ticket sales and how the growth, as well as the reviews on IMDB and Rotten Tomatoes to the different movies. The reason I attach that because it provides great context for the AI to get started because it has the names of all the movies as well as the years in which they were released and some basic initial information to start the process. And then I used the following prompt. I need your assistant in analyzing the rate in which Marvel released movies and also TV shows. I gave you the attached image as a reference so you can have the names of all the movies. I would like to get a detailed analysis of the release states of every movie in addition. I would like you to research the following, any TV series or any other media released related to Marvel and its release state. In addition, if possible, I would like to find the revenue from other channels such as merchandise, ride ticket sales, et cetera, in order to have multiple benchmarks to evaluate the popularity of Marvel through the years, such as comic book sales, et cetera. Now I did this as a deep research project in ChatGPT. Those of you who don't know how to do that, when you are in ChatGPT or in any of the other tools, there is a plus button in the actual chat interface. And when you click on that, one of the options that opens is deep research, and if you click on that, the AI will invest significantly more time in the research and will provide you significantly more detailed results after reviewing many more sources than it just would if you wouldn't click on the deep research option. The same thing exists claude and on gr, et cetera, on all the different tools you have, the deep research option. So I selected the deep research option, which immediately asks you some follow up questions. So it asked me three different questions, and the questions were to provide a more comprehensive analysis. Could you please verify a few points? Number one. What is the time range you're interested in? Should I include everything from the first Marvel Studios movie in 2008, which was Ironman 2025 or limit to a specific set of years? Question number two, do you want only MCU, which stands for Marvel Cinematic Universe? TV series and media releases, or should I include non MCU, Marvel shows like those on Netflix, like Daredevil, Jessica Jones, et cetera. And number three, for merchandise and other revenue channels, eg. Toys, rides, comics. Should I focus only on Marvel Studio's Disney Era post 2009, or should I include earlier dates as well? Okay, so a few things before I dive into my answers. These are great questions, right? Because these questions provide the AI more context and more specificity to provide me relevant answers. Now these questions will automatically pop up every time you start a deep research project, but you can do this and I highly recommend you do every time you use AI for something that is not very basic, ask it to ask you question for additional context and we'll see an example of that later on because that helps you provide more context to the AI without you knowing exactly what to add, and the AI is helping you providing more context, which in return will give you better answers. So my answers were number one, start at the release of the first movie from a video aspect. But if you can find information about merchandise revenue and comic sales, go as far back as possible. Number two, any Marvel related media release Number three, please include all the information you find, but please highlight the date of the first movie and the Disney acquisition because this will obviously provide us more context to understand what changed by the release of the first movies and what changed after the Disney acquisition. So then it started the research. It worked for 17 minutes. It looked at 31 different sources, indeed, 281 different searches to end up with a report that it created. Now the report gets delivered in a long text format, and it has several different interesting things. First of all, you can see it's arranged based on different headers and subheaders, so it's written as a report, but it also has. Links after every single parameter and every single piece of information that you can go and check if the data is actually accurate, and you should, because despite the fact that it's giving you links, which gives you the understanding that it's actually pulling the information from somewhere, you want to go and verify the information. You wanna verify the information because of two reasons. Reason number one is sometimes it would still hallucinate and make stuff up despite the fact it's giving you a link. Sometimes the links are made up and they don't even exist. Reason number two is you wanna verify that this is a legit website. So it's not Joe's blog, no offense to Joe, but it's actually a website that has information that you believe that you can trust. So as an example here, it says, uh, MCU related Consumer Products has generated around 41 billion in global retail sales for Disney by 2020. And then there's a. And if I click on the link, you can see that this link comes from a website called Starburst, and you can go and research that website and in there if you scroll down. The other cool thing when you're using deep research is that it'll highlight the segment of the article from which it pulled the information so I don't have to go. And read these four pages of article. I can go straight to the highlighted section and it says, film franchise has always been big merchandise, revenue generators, blah, blah, blah, blah, blah. And then at the end it says, MCU Merchandise has earned Disney around 41 billion by 2020, more than every single Marvin movie combined, right? So this is a very interesting piece of information for our research, but again, I wanna broaden what we are talking about. Think about this for any research you do for your company. This will go in research, multiple different websites. I've seen it go up to hundreds. I think the most I've seen was north of 500 websites, that it's done on a single deep research project. And then for each. Then for each fact that it founds, you can go and very quickly verify the source and verify that the data that it's pulling into the report is actually accurate. So I got this really detailed, long report. It's a few pages long, it has multiple things, but it's just long. Amount of text and it's very, very hard to understand what's actually going on. I'm not gonna read this to you, but if, just for those of you who are not watching the screen, I will read the first paragraph and it says, Marvel Media released Timeline 2008 through present, and then there's 2008 to 2012. The MCU begins on May 2nd, 2008. Marvel Studios released Iron Man, the first film, blah, blah, blah, blah, blah. And it's just lots and lots and lots of texts and you can scroll and scroll and scroll and keep on reading. And yes, the information is there, but it's very hard to actually use it in any useful way. And the same thing might happen to you if you do research for your business. So then I said, okay, what if we turn this into a table? A table's gonna be a lot more helpful for me to see the information. So I wrote the following prompt. Please create a table that shows the year in column A, and then each of the parameters in the following columns, as an example, column B can be the number of movie releases. Column C can be the number of TV episodes. Column B can be the number of comics, books sold, et cetera. So then it actually me follow up question. It said, okay, what are exactly the columns that you want me to use? And it used the columns that I gave as examples, but it also added columns e for estimated merchandise revenue, column F for theme park ride revenue and column G for video game sales, which is something I didn't even ask for, but it thought that it would be relevant, which is awesome. And so. I said, yes, this is great. Please create the table. And then it created the table. So now I have the data with column A showing the years 2000, 2001 and so on. And I have other columns, one showing Marvel movie related theaters. So which movie was released then? Uh, TV episodes that were released and on which series? Then comic books sold merchandise, retail sales theme park rides video game. Revenue and units. Great. Now, I have something that is way more useful as far as understanding the information that was provided to me based on the online research, but it's still very hard to follow. And if I wanna present this to other people, it's not a great way to present this because this starts at the year 2000 ends at 2025. So there's 25 rows, there's multiple columns. The chances somebody's actually gonna read all of this. Is not very high. The other good news is, again, I have the resources to go and click and verify the information once again. So that is another great benefit of this process. But then I said, okay, let's please create a graph from the table. Since the parameters are in different scales, please create a separate Y axis for each and every one of the parameters. Make sure all the parameters fit in roughly the same scale. Use the years as the e axes. So. As you probably know, when you create charts of more than one parameter, in some cases they're in very different scales. As an example, you have something that will be in hundreds of thousands and something that will be in single digits. So how many movies were released versus the revenue, as an example, the revenue will be in hundreds of millions. The number of movies are gonna be single digits, and so you will not see the single digit graph at all, but it's going to because it is going to be flat. On the bottom of the chart. Now in Excel and in Google Sheets, you can create a secondary Y AEs, meaning you can have two separate Y AEs for two separate series, but that's it. It is. Practically impossible in day-to-day tools to create multiple Y-axis scales for multiple series of data. And here, all I had to do is ask for it. And I verified which data I want to include in the graph, which is everything that was in the table. And I said, yes, go ahead, please create it. And it created a line chart because I didn't exactly explain what kind of. Chart I want. And again, those of you're seeing it, you can see the line chart with each and every one of the parameters in a different color, and each and every one of them with a separate scale. So on the Y axis, I have TV episodes in 20, 40, 60, 80 as far as the digits. Then I have comics sold in million with 65 70, 75, 80, and so on. So these are in tens of millions. Then I have merch revenue in billions with 1, 2, 3, 4, 5. Then I've got theme park events and so on and so forth. So you can see that what it's doing, it's allowing me to put all the graphs in a single view. In each and every one of them. I can see the trend now. I could see the trend in theory, but because I have six different graphs, all one laid on top of the other, it's very, very hard to understand what's going on, especially in a line chart, and they start crossing each other and intersecting each other. So what I did then is I asked it to turn this into a bar chart, and with each bar representing the data series per year, and so it created the bar chart. Now it's a lot easier to actually see the data. So what I have is I have separate bars for every single year, and for each year I have six different bars showing the six different parameters that we are looking at, and they all fit in the same page, but it is still very hard to understand what's going on overall because it's very hard to understand. The relationship of each and every one of them to the other bars because there are six bars for every single year. So I went on and I said, okay, please turn it into a stacked bar. Into a stacked bar chart for each year, and it did, and now is where the magic starts happening. Now, we can actually see the trends across all the different years, across six different vectors. Of comparison of data that are unrelated. Again, it's apples and oranges. It's number of TV series, episode released versus revenue from theme parks, right? These are not things that you should be able to combine, but because AI very easily can do this with simple requests in English, it provide. An incredible graph that shows us the trend. Now, the numbers on the left side, on the Y axis are completely relevant, right? Because we're combining things in single digits, in millions, in tens of millions, in billions, all into one graph. But from looking at a trend, trying to analyze, let's say, demand across different sector for our services is something that was. Very, very hard to do before, and you needed serious data scientists to do this for you. And now my son and I were able to do this in about 20 minutes. What I did then is if you, those of you who can see the graph, you see some of the parameters are still very, very thin, meaning they're very, very small and we can barely see the trend, meaning their ratio, the scale they are compared to the other ones. Are not good enough. So I started going back and forth and playing with the scales of the different parameters until they're all significant enough in the graph. And then in the end I ended up with a graph, and I keep on scrolling down because we did a few other additional research, so I'll explain the additional research. We understood that some of the parameters we were checking were not the parameter we actually wanted, and that there could have been better parameters, such as looking at tV series, watch time in minutes instead of how many episodes were released because what we're trying to see is the impact on global audience. So the fact that six different series with 35 episodes were released doesn't tell me what was the impact on the global audience. But being able to find the number of minutes that were watched on those different TV series is obviously providing a much better parameter. To track the success of the Marvel Universe on the world. So we made a bunch of changes. Again, just asking AI for it and what other piece of information could be more useful to do this analysis. It gave us that we sent it to do additional deep research to find that information, and then we created new graphs. So you can see here the updated graph, again, the bar chart, not the stacked bar chart, but there's the bar chart is almost impossible to follow because every single year has six different lines, and it's very, very hard. It's really, really complicated to see what's going on. But then when we stack this one, we have the final answer to everything that we needed, a small additional change of scale. So as an example here, please show the TV viewership in hundreds of millions. Because in the previous chart, it was almost not showing up because it was in billions, and now when I change it to a hundred of millions, it becomes 10 x bigger, which now starts to impact the overall trend. Now again, I I wanna explain something again. The numbers don't make any sense, right? Because some of them are in the billion, some of them in the millions. What I'm trying to see is the trend, and so I need each and every one of the parameters to DA scale that changes the overall trend and being able to just ask for it in English. Is magical. So what are we looking at right now? We're looking at a graph that is showing as a stack bar chart for every year. The movie tickets sold in millions. The TV viewership in hundreds of millions of minutes. How many comics were sold in millions? What was the merch revenue in hundreds of millions Theme park events. As far as the events on the different theme parks that were launches of different rides and so on, and game revenue in millions of dollars, all of them stacked on top of one another for every single year. And what it shows is an incredible trend showing the growth from the year of 2000. Being more or less flat, like a little bit of growth with the initial launch of the initial movies, but staying roughly in the same ballpark all the way through 2009, and from 2009 all the way through 2018, a huge growth up that comes to about four to five x. The amount of total volume of, let's call it audience engagement. Compared to the year 2000. So yes, the numbers don't make any sense. It's apples and oranges, but the trend and the total volume is very accurate. You can see it is five to six times bigger than what it was before. You can see that the trend grows every year. You can see what were the biggest contributors to this trend every year. You can also see the decline. You can see the complete collapse in the year 2000 because of COVID, obviously. So there's no. Movie tickets and there's very few t TV series, which is actually very surprising to me. So I went back and did additional research on that. So why weren't there a lot of people watching this on tv? And the reason is the last TV series they released was actually in 2018. So in 2020 got there and we were all stuck at home. Everybody already watched the Marvel related TV series, and so not a lot of people, uh, view that. And the only thing that is still staying very, very high is game revenue, which makes a lot of sense. People are stuck home and one of the games we were playing are Marvel related games, but then there's a huge spike in. 2021 because they recorded a lot of new TV series because they understood that there's an opportunity for that and they did not know if the pandemic will end. So five different series related to Marvel characters were released in 2021, which ro another big spike and then there's a slow decline from there. As you know, there weren't any big movie series hits after end game Avengers. So it's slightly declining from there. So what is this showing us? It is showing us an incredible ability that literally was not possible before. It is the ability to collect data from different sources in different scales and different types of data to evaluate basically anything you want The. Current demand for your product based on sales of your competitors, combined with reviews of your product, combined with appearances of the thing that you're selling in different conventions around the world. You can combine different types of data that previously was not possible either from a research perspective. It was just very hard to do all this research as well as from combining it in a logical way, and now you can do this within less than an hour see data in ways that were literally almost impossible before that can support better decision making for your business. So this is one aspect. The other aspect that I find really interesting is the education aspect of this. I did this with my son, showing him how he can effectively use AI in order to support a project that he has. I could have done the same thing with my employees, right? And show them how to properly use AI in order to get better answers to things they're trying to get answers for, and showing how you can go really, really deep and in a very highly professional, detailed way without spending weeks on a specific topic. So I really enjoyed that process with my son. I highly recommend to each and every one of you that has kids to find ways to show them how to use AI exactly this way. So don't send AI to do the project for you. He still spend. Two days writing the actual report on his own. He did not use AI to write the report, but to do the research, to do data analysis, to review information, to verify sources. The process was significantly faster with AI and he still had to do the work on his own, which is exactly how would expect kids and adults and professionals to use AI in their work. The other thing that it was great to show him how to do is how to check the data, how to go and verify that the sources that are being used are legitimate sources and that the data that it's using is real data. So that was very helpful as well. So summary of this episode, first of all, learn how to use deep research. It's an incredible, powerful tool. It exists on all the different platforms, and again, to operate it to select it, all you have to do is click on the plus sign or there's different symbols inside the actual chat bar and select the research, and then the AI platform will invest significantly more time, look at significantly more sources and provide you way more accurate data. The second thing, go and verify that data, so all these tools provide you the links to the data that it's has used in the different sections, and you can go and verify the data on your own. It'll take you a little bit more time, but it will make sure that you're not using made up data, that the AI will hallucinate and it will. Then you can create different kinds of visuals. Some of them are straightforward, some of them could be really complex. And the cool thing here is, let's say you don't know, you can ask AI to help you. You can say, this is the data that I have. I'm looking for some kind of visualization. Gimme three great ideas on how I can visualize this data to support. This thesis or this argument or this data analysis that I'm trying to do, and it can suggest ideas or you can do what I did and just experiment and try different things and see what helps you to get to an outcome that will actually help you make better decisions. Then the final step, once you have those visualizations, you can manipulate the visuals and make minor changes or bigger changes in order to get the exact outcome that you want. Meaning you saw that I change the type of parameters I was looking for by doing additional deep research, as well as I also change the scales of the different components in order to really stack them in a way that will show me clearer trends. And all you need to know is how to ask it in simple English. You then create a final version, and if you really want to go the next step and use ai, you can have AI write the initial report for you based on your template, so you can upload your template and say, this is what I want the output to look like. You get an initial output, you can do it. In Canvas. So Chachi PT has a canvas tool. I recorded a whole different episodes about this that you can go and check. Canvas is an amazing way enough to do. All you have to do to activate Canvas is to ask Chachi PT to open it in Canvas. It opens it on the right side instead of in the chat, and then you can edit the output. It becomes like a Word document that you can edit and you can highlight specific sections in the document and ask AI to highlight. Ask AI to elaborate or add more information or make it shorter, or whatever it is that you want it to do for each and every one of the sections. And then all you have to do in the end is copy and paste it into Microsoft Word or a Google Doc. Add your header, footer, table of context, make final adjustments, and you have a report. That's it for today. I. Hope you found this highly valuable. I think it is. It touches across multiple aspects of how to properly use AI both in business as well as for personal life and helping your kids or employees learn how to use AI better. Please gimme some feedback. If you like this kind of episode, I can record a lot more of these because I do these kind of things every single day. if you enjoyed this episode and if you're enjoying this podcast in general, please hit the subscribe button so you don't miss any single episode that we release. As well as click the share button. Yes, there's a share button on your podcast player, and share it with a few people that can benefit from it. I'm sure you can think right now of a few people that can benefit from this episode and definitely from the podcast overall, you can help them learn. And I will really appreciate it if you share that information. And until the next episode, have an amazing rest of your week.

Head Shot

Isar Meitis

Host