200 | The Ultimate AI Showdown: Comparing Data Analysis tools with Liza Adams

0:00Hello and welcome to episode 200 of the Leveraging AI podcast. this podcast episode is going to be absolutely epic. Now, episode 200 is obviously a big milestone, and I'm very excited about this, and hopefully you are too. But at the end of the day, just a vanity metric, it's a number. What I really care about, what really matters, is how much value can I actually provide you, the listeners, business people, my community members, and anybody who stumbles into this content randomly in one place or the other. Overall, it's tens of thousands of people every single month. And that's a serious responsibility that I don't take lately.

0:44And so the first thing I need is to thank all of you, the people who are following me, who are consuming the content, who are connecting with me on the different platforms, who are sharing the value that they gained with me, from me. So basically telling me how they're doing different things and how much it's helping them in their businesses. And it really, really, really fills me up. Like different people approach me on the different places and tell me what they're doing. Just this week, one of my clients is launching an internal AI contest to drive innovation and adoption across their people after we've done training for their people.

1:18In our Friday Hangouts, the recent Friday Hangouts, those of you who don't know, it's a community thing. It's open to all of you. Anybody who wants can come and join those Friday Hangouts. just this Friday Hangouts, one of the guys who's been a member of the community for a long time and has taken my course, has shared that he has built on his own an app for Zendesk that allows his salespeople to provide faster, better answers that are aligned with the company policies and tone of voice. On his own, using AI, building an application, he's not a coder, and he was able to do that. Yesterday, I spoke with one of the recent course graduates, and she has told me that her The company just took a project that they wouldn't have been able to take without taking the course.

1:59These are the things that keep me going, doing this week in, week out, many hours, two episodes every single week. It's that I can help exactly these kind of things happen. So with that in mind, I was looking on how can I make this the most impactful, right? How can I make sure that episode 200 really provides the most amount of value that I can to all of you? And I started looking through what kind of questions do we get on the YouTube channel, on the Friday AI Hangouts, on public courses. When I speak on stages, like what are the things, specific company workshops, what are the things that people ask the most?

2:40And what I found is that the thing that people ask the most is which tools should I use for which thing, right? People are so overwhelmed with having all those different tools and they don't know which one to pick. And most of them just stay with Chachapiti because of that, which is great. Chachapiti is an awesome tool, but there are many other tools that can do a lot of other things. And so I decided to go with that. And the problem with that is obviously it's a moving target, right? It's going to be true today, like which tools are best for which use cases today might be a different answer by the end of this day today and definitely two months from now.

3:17But I assume that giving everybody a snapshot of what's the current situation across multiple use cases is going to be extremely valuable to everyone. And based on the excitement that I got from everybody who heard about this, I think I was correct with my assumption. Now, the other thing that this will do, by hearing from top experts how they evaluate and how they use the different tools, will also allow you to develop that lens, to develop that muscle, so you can, in the future, figure this out on your own based on the way they're doing it right now. So my next step was, okay, which topics do we want to cover?

3:52And again, I was looking for stuff that is up and coming, that's going to drive a lot of value, and that a lot of people are excited about. So we're going to cover four different topics. data analysis, which to me is the biggest value from an ROI perspective to a business, vibe coding, because I think that provides superpowers that every business person now has, whether you know it or not, to create applications that does things for your business, image generation and video generation, because it's fun, exciting, and everybody can use it across multiple aspects of the business. And so to do that, I gathered four of the top experts in each and every one of these fields.

4:25And each and every one of them is going to get 30 minutes to share with you which tools they're using, how they're using it, and why they're using specific tools for specific use cases so you can do the same. Now, as a first step, as I mentioned, I would like to thank all the people who are joining us live. There are a lot of people on the Zoom and a lot of people on LinkedIn who are with us today, and I appreciate you doing this. And I would like to ask you to introduce yourself, say where you're from and where you're joining, and also why are you here today? Like what are the key things that you want to learn in AI from this session?

5:03Which are the topics that interest you the most? Just share with other people and connect with other people. It's a great opportunity for you to network as well. Like I said, there are a lot of people both on LinkedIn and on the Zoom call. That being said, I have one final exciting announcement and then we're going to get started. So as you probably know, at least some of you know, I've been running the AI Business Transformation course for over two years now. We started in April of 2023. I've been teaching at least one course every single month, which is a lot. That means that hundreds or maybe thousands of people have taken the course by now.

5:37And we're teaching most of these courses privately, meaning most of it is organizations and groups and companies that invite me to teach just their people. But once a quarter, we open a public course. and that public course allows anybody who wants to learn AI to sign up and some of these people are solopreneurs, some of these people hold managerial positions in different companies and the company itself is not moving them forward so they decided to take the course as well. I shared with you some of the things in the beginning of what people are doing with the knowledge they got from the course and so the exciting news that I have is we open the registration for the next open cohort.

6:16The next open cohort is going to happen starting on August 11th, and it's going to go for four weeks, just like all of the other courses, four sessions, two hours each for four weeks, starting on August 11th. But because we are celebrating episode 200 today, I'm doing something that I've never done before, which is to provide$200 off the course for anybody who signs up today. So if you're here on the live and you're listening to this when you're with us and you want to take a course that can change your career and your business. And just a point to prove that it's just not me saying this, in the recent report by PwC on the state of AI in business, they shared that based on their statistics, people who has advanced AI skills make 56 % more money right now than their peers.

7:07That's a very significant amount of money. So in addition to the fact you'll know AI and you can future-proof your business, you can make more money right now. So if you want to take the course, use promo code LeveragingAI200, all uppercase, one word, all connected. In the sign-up, we will share the link to that in the chat. And so with that, I want to move forward to the actual show. So the ultimate AI showdown, as I mentioned, is all about understanding which tools to use and how to use them in different use cases. And the very first use case is data analysis. And for that, we are in the reason, by the way, I chose this to be first.

7:51Like I said before, I think data analysis is the thing that provides the most amount of value in businesses from a strategic perspective, because it helps you make the right business decisions. And AI gives you superpowers when it comes to doing data analysis. And our expert on data analysis is no other than Lisa Adams. Lisa has over 20 years of experience in senior marketing positions from VP to CMOs across different sizes of companies in different industries. And all of that at the age of 28. Yes, I know you're doing the math. She started when she was eight. She's incredible. No, I'm lying about the age, but she does have 20 years of experience in senior marketing positions.

8:29and she brings amazing experience of running strategic thinking in businesses. In the last two and a half years, she's been a fractional CMO, helping companies implement AI as part of their strategy. And she is the queen of data analysis with AI. And she's going to share with us how to maximize AI for strategy insights in seconds using AI. It is not her first time on the show. So I am very excited to welcome Lisa back. Lisa, welcome to Leveraging AI.

9:04Hi, Isar. Good morning, everyone. And sorry to tell you the truth on why I'm going first. I am live from Kauai, Hawaii. I am on vacation with my family. However, when Isar asked me to do this, I couldn't refuse. I had to make it work because I learned so much from him and from his community. and what a way to give back, to share some of my key learnings, my experiences, what works and what doesn't work. Thank you for having me and congratulations, Isar, on your 200th episode. Thank you so much. So stage is yours. Yes, thank you. Thank you. All right, guys, I've got slides and I also am going to attempt to go live on three AI assistants.

9:49I'm going to be using ChatGPT, Cloud, and Gemini to do some data analysis. Let's just see what happens. All right. To do data analysis, I actually had to do three steps. I needed to do some deep research to gather the data. Then we had to analyze the data and lastly, visualize the data as well. And what we're going to do in step one is we're going to do deep research on three project management software solutions. And we're going to pretend that we're evaluating project management solutions. And we're going to evaluate Asana, Monday, and Smartsheet. And we're going to ask our AI assistants, Chachi Petit, Claude, and Gemini to help us with this evaluation.

10:30And as you guys know, many of you know that when it does deep research, it outputs a very long report. So I wanted it to summarize its deep research into visual representation. So we'll see what it came up with. And then step two, what I've done here is I have taken the three outputs, one per AI assistant. I then took each, those three outputs and put it back into each AI for each AI to evaluate those three outputs. I did not give it information around which report came from which AI, but I would assume that they roughly know which one. And I had them analyze their outputs, their own outputs for a number of criteria to determine how well they did on research and analysis, and then also asked it for a visual representation.

11:20And then lastly, I asked each AI to craft an interactive tool or guide, it doesn't have to be interactive, with a quiz or some sort of interaction where it asks anywhere from five to ten questions of the user, and then based on the research and the analysis reports that it crafted, it will make a recommendation of which project management platform is best suited for that user based on how they responded to the questions. So we're going to go through this pretty quickly, but I'm hoping that you will see the progression in the next 18 minutes. I'll pause you just for one second to add two cents.

11:59I love the fact that you're feeding it back to the AI for like a feedback loop. What I do is something similar, but a little different. I have a custom GPT that is actually built into an automation on NA10 that actually takes the output. So similar to you, I run it on Gemini, I run it on ChatGPT, I run it on Claude, I run it on Perplexity, and then I feed them all into this automation that actually creates a table that compares all the facts that they all found. And on the top of the table, the columns are the different tools, and it shows me which one has found the facts. And basically what this automation does, if all three or four of them found the same stuff, it's most likely real and not made up.

12:37And what the second step of the automation does is it looks for the outliers, the stuff that only one of these found, and it goes and does additional research with a different tool to see if it can find the same fact. If it does, it tells me it's also real. It's just something that only the other tools didn't find. And that gives me a very solid way to then write the report knowing that all the information is accurate. And I was able to kick out all the stuff that one of the tools hallucinated on something specific. Yeah, that's awesome. And although I'm getting a little bit meta on this, given that I'm feeding the results of AI back into AI, at the end of this presentation, I will show you my picks, the human Lisa Adams picks on what I believe is the best, just in this use case, because it varies depending on the use case.

13:22All right, step one. Okay, so we did research and analysis, and this is my prompt. And I don't know, how you're planning to share the decks, but I'm open with everything that I do. So feel free to share these decks. And here's my prompt. I'm a research project management solution for US-based organizations. Please conduct comprehensive research on Asana, Monday, and Smartsheet using multiple sources, vendor websites, analyst reports, buyer's guides, so on. And then I said, rate each vendor one to five, five being exceptional on these criteria, ease of use, features, functionality, pricing value, customer support, integration, scalability, and overall, present in the table with a brief rationale for each rating.

14:04And then I said for each vendor, provide key strengths, key weaknesses, best use cases, worst use cases, red flags to watch out for, and critical questions to ask during an evaluation. So as you can tell, we're a prospect. We want to deeply understand these potential project management solutions. So just to give you at a high level. Here's what I did. Cloud 4 Sonnet is what we're using for deep research for Anthropic. They do have deep research, but today it's on beta. So I'm giving it a lot of grace, right? So it took about five minutes and 10 seconds, 396 resources, research, and then it gave me a seven-page report.

14:42It was the fastest of all of them. And this is what it came up with, Monday as the top, then Asana, then Smartsheet, and then Asana. Gemini did a deep research, not as fast, less sources, but very comprehensive report at 30 pages. And then what's interesting about this one is it wasn't like picking a top vendor. It was actually saying situations where one might be better than the other, which I thought was very interesting, right? Because not everybody has the same top one. So I really like the nuanced response of Gemini to that 2.5. And then I use ChatGPT 03. And as you guys know, there are eight different models of ChatGPT, but I wanted to use a reasoning model and one that is fairly good in strategy.

15:32I could have gone on ChatGPT 4.5. So you could, you know, try to replicate this and see what you get depending on the model, but this one took a bit of time. Nine minutes, 37 sources, 178 research, but it's roughly about a 30-page report as well, like Gemini. What's interesting about this one is Monday, again, came up on top, very similar to where Claude Sonnet identified, and then Asana and Smartsheet came in at second place. So I wanted to just show you what this thing looks like, just so you know you see how this worked so basically here is Claude oops

16:17you can see all of my projects and here is Claude and then you can see there's my prompt and then here's the research and if you click on that research it gives you the full report it gives you all of the sources and what I'm going to show you here and then here's Gemini same prompt and then the research is over here and then here is ChatGPT same prompt and then I gave you the research report. So I'm going to go back to my... Two cents on that. You see everybody that the process is very similar. The outputs are very similar. They open in the side panel that is called differently in each and every one of them, whether canvas or artifacts or whatever.

17:13But the interesting thing is in ChatGPT, when it's doing its research, which I really like, it asks you follow-up questions before it does the research. They're kind of like, oh, what about this? What about that? Do you want to know this? It clarifies what you want to get before you actually hit go. And that's why you saw there's another section. It happens every time you run deep research on Chachipity, which I think is really smart. Well, you know, I don't like to read really long reports. So it is really hard for me to distill in my own simple brain. So I always ask for a visual representation.

17:49So here's my prompt. For all three of them, I said, please create an infographic that highlights the key insights from the report. Make it easy to read, easy to understand, and visually appealing. My brand colors are deep blue, vivid orange, shades of gray, and white. Now, I could have put hex codes in here, but I didn't do that. Just so you know, you can just have a very simple prompt like this. I'm going to show you the visual representation. So here is the visual representation from Claude. So it basically ran, just show you super quick what that looks like. So here's the prompt right here.

18:25Create an infographic that highlights as I read it to you. And then the infographic is right here. Amazing. Right? I think just again, to generalize this for a second for everybody listening. First of all, I think the output is incredible because like you're saying, it makes it very easy to follow. And it's going to be a lot more user-friendly to distribute for sure. The cool thing about this is you can do this for anything, right? So we're now doing an example about comparing project management tools, but you could do this to compare your competitors. You can do this to compare your clients. You can do this to compare your potential offerings to potential.

19:04Like you can do this on any topic you want and run this across these tools and generate a very similar, really beautiful output that you can immediately share together with a full report. This could be your executive summary, if you want, is that kind of like the cover letter could be the infographics and then share the entire report with whoever you need. Yeah. And to your point, Isar, you know, while I prompted this as if I'm a prospect of project management software, you could actually be a product manager or a marketing manager at any of these companies and you can determine how the market might be perceiving you.

19:38So as a marketer, you can take a look at this and determine whether you're positioning in the right way or if the market is perceiving you in the way that you want to be perceived. All right. So that's Claude. And then I'll show you Gemini. Gemini looks like this. So Gemini was fairly comprehensive in its infographic. But Gemini just recently did this, I think about a week ago when it started pumping out, you know, infographics. The challenge with Gemini compared to Claude, Claude, you can actually share that infographic because you'll get a link. I can show you that link. Anyone can see that infographic with Gemini.

20:21For some reason, we still can't share anything. So basically, this is what I get, right? And I can't share it with Esar unless I just cut and paste and screenshot these things and I send them an email or a Slack. But there isn't a link today that could be shared with Gemini. And going back to this, one of my favorite tools for sharing is actually perplexity. And the reason for that, if you do the same process in perplexity, you can turn the output to a web page that is publicly available to anyone. And then you can share the link to the web page. It's just a web page. The other cool thing, which is a twist on this whole thing, if that webpage, because you requested, has links back to your website showing how your product is better than other products or comparable or what are the strengths of it, you now have a reputable website like Perplexity pointing back to your website with an article that says how great your product is.

21:17So just food for thought for people on other stuff you can do with this process. Yep. So it's super simple. It's right here. This is the report. You go to create and you have a number of options. I just clicked on infographic and it created the infographic for me. I could even say create a Jeopardy game based on this report. And then, you know, it'll take just a little bit, but it can create any app that you want. And then you can speak with it in natural language. So we'll come back to, oh, well, it's still programming. It's still coding. So we will just keep going because I only have six more minutes.

22:04You're doing awesome. And then this is ChatGPT. ChatGPT is by far the one that struggled the most when it came to creating the infographic. So ChatGPT is right here. I asked it to create an infographic of this report. Now I can't find the infographic.

22:28Okay, we will just go over here, but this is the output of ChatGPT. And I'm not exactly sure what it's saying. And it has some spelling errors and obviously it went down the path of not creating an interactive version. So would not recommend ChatGPT for at least for this use case. And then let's just take a look at Gemini and see what it's done. All right. So here's my project management Jeopardy game that Gemini created while we were toggling between my spreadsheet. And let's just see. I don't know. OK. And then we type in an answer and there's your game. Anyway. All right. So just something fun to think about.

23:10All right. So step two, what I did is I took all three AI outputs and put it into each AI and asked for its evaluation of its research and analytics capabilities. Here's the prompt. In your knowledge, because I uploaded all three reports in their knowledge, are three research reports analyzing the same project management software competitors. Please evaluate each report, one through five, on research thoroughness, source quality, analysis depth, actionable insights, and overall quality. Present ratings in a table with rationale, which I love to do because tables are easy to understand. And I always ask for its reasons because then I can understand how it thought about the process.

24:00Please do not identify which AI created which report. Refer to them as report A, B, and C. A is Claude, B is Gemini, and then C is ChatGPT. Be honest and unbiased. This is how you can be most helpful to me. And then after it did that, I did a second prompt that said, please output an infographic or a visual representation that best highlights the key insights and results of your analysis. Please reflect the actual AIs. Now that it's done the analysis, I wanted to see in the infographic the actual AIs that were evaluated. But please make it easy to understand and visually appealing but still professional.

24:37And then I gave it my brand colors. All right. So here is the output of Claude. Claude basically preferred itself. And then Gemini and then ChatGPT. And then here's its visual representation of it. Basically, you can see here the ratings by category and then the overall quality. So just keep that in mind that it preferred itself for the following reasons. And then here's the output of Gemini. Gemini also preferred itself, rating it as the highest at 4.3, and then ChatGPT, and then finally Claude. And then here's the output of ChatGPT. ChatGPT, interestingly enough, preferred Gemini. So it went Gemini, and then itself, and then Claude.

25:34And then there's reasons in here, obviously, right? But I will tell you my final picks towards the end. And then, like I said, it doesn't do very well in visual representation because it didn't even follow my instructions when it came to, you know, hey, now say that it's Claude and it's Gemini and these are not the right logos. But you can see that it picked Gemini. So what I did then is the third step is I asked, hey, create an interactive tool or guide with recommendations based on the user response to key questions. And here's the prompt. The guide will help a user determine the best solution based on a set of five to eight evaluation parameters that are most relevant given the research.

26:18Include your reasons for your recommendation. Please make it easy to understand. Here's the recommender tool. So I can go to Claude and show you this.

26:31I'm just going to go down here and pick up my artifact.