Everyday AI Podcast – An AI and ChatGPT Podcast - Beginner’s Guide: How to visualize data with AI in ChatGPT, Gemini and Claude

Episode Date: December 3, 2025

FYI -- Today's LinkedIn livestream broke, so you can access the custom instructions here. This is Vibe Coding 001. Have you ever wanted to build your own software or apps that can just kinda do your work for you inside of the LLM you use but don't know where to start? Start here. We're giving it all away and making it as simple as possible, while also hopefully challenging how you think about work. Join us. Beginner’s Guide: How to visualize data with AI in ChatGPT, Gemini and Claude -- An Everyday AI Chat with Jordan WilsonNewsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion:Thoughts on this? Join the convo and connect with other AI leaders on LinkedIn.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:Combining Multiple Features in Large Language ModelsVisualizing Data in ChatGPT, Gemini, and ClaudeCreating Custom GPTs, Gems, and ProjectsUploading Files for Automated Data DashboardsComparing ChatGPT Canvas, Gemini Canvas, and Claude ArtifactsUsing Agentic Capabilities for Problem SolvingVisualizing Meeting Transcripts and Unstructured DataOne-Shot Mini App Creation with AITimestamps:00:00 "Unlocking Superhuman LLM Capabilities"04:12 Custom AI Model and Testing07:18 "Multi-Mode Control for LLMs"12:33 "Intro to Vibe Coding"13:19 "Streamlined AI for Simplification"19:59 Podcast Analytics Simplified21:27 "ChatChibuty vs. Google Gemini"26:55 "Handling Diverse Data Efficiently"28:50 "AI for Actionable Task Automation"33:12 "Personalized Dashboard for Meetings"36:21 Personalized Automated Workflow Solution40:00 "AI Data Visualization Guide"40:38 "Everyday AI Wrap-Up"Keywords:ChatGPT, Gemini, Claude, data visualization with AI, visualize data using AI, Large Language Models, LLM features, combining LLM modes, custom instructions, GPTs, Gems, Anthropic projects, canvas mode, interactive dashboards, agentic models, code rendering, meeting transcripts visualization, SOP visualization, documenSend Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info) Start Here ▶️Not sure where to start when it comes to AI? Start with our Start Here Series. You can listen to the first drop -- Episode 691 -- or get free access to our Inner Cricle community and all episodes: StartHereSeries.com Also, here's a link to the entire series on a Spotify playlist. 

This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live and Adobe Firefly, the All In One Creative AI Studio. Just describe what you want to create and the assistant handles the rest, orchestrating multi-step workflows across Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome. The assistant accelerates execution. Combining multiple features in a large language model is where you can really start to unlock superhuman-esque capabilities.

So today I'm giving you the keys to the castle. We're going to be combining two of my favorite features in most front-end large language models that hopefully will change how you think about working with LLMs. Ultimately, over the next 20-ish, minutes. I hope to show you live how you can not just visualize something inside chat, Chbbyt, Gemini, and Claude, but how hopefully you can rethink about the way you work. Because this is really an easy entry into vibe coding 101, but I'm going to do it all for you. All right. I'm excited for today's show. I hope you are too. Let's get into it. What's going on y'all? Welcome to Everyday AI. My name's Jordan Wilson. And if you're new here, this

is an unedited, unscripted daily, live stream podcast and free daily newsletter, helping everyday business leaders like you and me, keep up with all of the large language model and AI updates, make sense of it, cut through the BS, pull out the important insights that we need to grow our companies and our careers. If that's what you're trying to do, sweet, me too. Starts here with the unedited live stream podcast, but if you really want the most important insights, make sure to go sign up for our newsletter at your everyday AI.com. we're going to be recapping the highlights from today's show.

So if you miss anything, don't worry about it. It's going to be in the newsletter as well as all of the other AI news and whatnot that you need to know. All right. This is AI at work on Wednesdays. This is a series we've been doing now for, I don't know, nine months or so. And it seems like you all really like it. But one thing that I hear from a lot of people is, hey, it's almost like there's too much information or sometimes things are a little too advanced. So I wanted to take it back to the fundamentals.

all right. But I'm going to explain that here in a minute, but we're actually just going to kick it off live. We're going to do it a little different today. We're going to be doing a lot more on the live side. So if you are listening on the podcast, appreciate your support as always. This might be one of those that you're going to want to check out the video version. So again, just go to your everyday AI.com and then find today's episode. And you can watch the video version. So this is episode 665. So go watch the video version on our website. All right. We're going to start a couple of things live. All right. I do this sometimes. We got to put the ingredients in the kitchen. So what was I talking about by combining multiple features in a large language model? Well, what I'm talking about is combining projects or GPTs or gems. All right.

So I'm going to talk you through that later. So combining those and some of the capabilities that using those, things unlock, but also using each large language model's ability to essentially write and render code. All right. So like I said, we're combining two separate kind of modes or features in the three big models in chat, GBT, Gemini, and Claude. All right. And then we're going to be showing you how to essentially vibe 101 your day, your daily work.

All right. So I'm going to explain what I'm doing here for our podcast audience. So I have a GPT open inside Chad GPT. All right. And stick around to the end. I'm going to give you everything, right? Because there's some custom instructions in here that were not easy to get working correctly. I don't know for whatever reason.

I kind of put out a tweet out there. So I might have to end up reaching out to someone at Open AI. But the GPTs, I don't know why in the last week or so, they've been finicky. So, you know, GPTs and gems are a kind of a customized version of, chat GPT or Gemini respectively. And then projects is a little different inside Anthropic, but essentially it's a way to create a customized version of the main model that just acts specific to your needs.

So I have some pretty in-depth in advanced custom instructions that I'm going to be given away to everyone. Don't worry. So that's what I'm going to be doing here. But for whatever reason, like I said, GPTs are having problems with consistently reading uploaded files. So I'm going to be doing a couple different tests. So in the first one, I'm just pasting it a bunch of information.

So in the first test, like I do on our weekly putting AI at work on Wednesdays, I'm thrown in some podcast stats. But as we go through this, think, what is the data that you constantly work with? Or, you know, long meeting transcripts, you know, long SOPs, right? Don't just think about visualizing data. Think about visualizing your work, all right? Because I do think that's going to, as we look into 2026, that's a new era of work that we're going to be moving toward is just working in much more visual live and AI powered

interfaces. So this is a good way to get used to it. So I hope you can follow along with me. All right. So here we go. I'm clicking enter in chat. GBT. All right, we're getting these things started off so we can do it live. So that was a chat GPT custom GPD. Now I am in Google Gemini and I have paid accounts for all of these. In this case, I'm going to upload a file. So all this file is is a CSV with all my podcast stats. All right. And I'm going to get this one going. There we go.

Our Google Gem, again, a custom version of Google Gemini. It's off and running. Now, I am inside Claude. A little different here. I have what's set up as a project. So Claude doesn't have the equivalent of a GPT or a gem. They have a project, right? Which Chad GBT also has projects.

Google Gemini is going to be rolling out projects any day now. There's pros, cons, overlap, right? But for all intended purposes, we're doing the exact same thing. I have those custom instructions inside this project. And all I'm doing, again, all I'm doing, I'm uploading a file and I'm hitting enter. That's it. Nothing fancy. No prompt engineering, right?

You don't need to know anything. I'm uploading a file, hitting enter. And the custom instructions, I'm going to be giving away to all of you. who repost this episode, by the way, is going to take care of the rest. Well, hopefully. All right. So now we have our pies in the oven, so to speak. All right.

And I'm going to be kind of keeping my eye on those. And we're probably going to be testing out a couple other ones live. But I'm hoping that these ones that I just kicked off are going to actually work. All right. So let's. get into it. So like I said, the custom instructions to get all these to work was not easy. All right. So if you want access to this, I built it for you, right? Talking about combining multiple modes, right, being able to kind of like a puppeteer, be able to control certain elements of

Chad Chbg, Gemini, and Claude. So if you want those custom instructions and hopefully you'll be able to see and understand some of the results here at the end when we check in on these. Just go repost Today Show and I will send them all to you. Give me a day or three. It just depends on how many people repost this. All right. So here's what we're going to be going over on the rest of today's show. Well, I'm going to simplify the three best ways to visualize data, visualized data in the major

LLM providers. And I'm going to talk about some of the pros and the cons. We're going to go live even more than the three that we just put in. I'm going to try a couple of other. things because here's the reality. Not everyone works with data all the time. So think whatever document types that you work with, this should work. So yeah, gonna go live, doing it, doing more, and then I'm gonna give you all the goods, right, at the end, so you can do this immediately. All right. So why? Adobe just introduced an entirely new way to create,

bringing the power and precision of its creative suite into one conversational experience. audience. Meet Firefly AI Assistant, now live in the Adobe Firefly app, the all-in-one creative AI studio. Powered by Adobe's creative agent, Firefly AI assistant lets you start with your vision, just describe what you want, and shape the outcome as it takes form with the assistant. The assistant orchestrates multi-step workflows drawing on 60 plus pro-grade tools across Adobe Creative Cloud apps, including Photoshop, Illustrator, Premiere, Lightroom Express, and more to help bring your ideas to life. You can also get started with creative skills, a growing library of pre-built workflows for common creative tasks, like batch editing photos, creating mood boards, portrait retouching,

and creating social variations. Every step the assistant takes is visible so you can refine, redirect, or take over at any time. You stay in the driver's seat as the creative director. Adobe Firefly AI assistant now in public beta. See it today at firefly.adopi.com. You might be thinking, why do I need to do this? Right? There's no, you know, you might be looking at your workflow and say, hey, there's nothing broken. If it ain't broke, don't fix it. Right.

You could make that argument, sure. But I hope you'll see by the end of this, how this changes how we work. Because no one wants the equivalent of the blinking cursor when it comes to AI. And I think so many or so much. of the sheer power of large language models is their ability to code. But I think when you hear the word code, I think 99%, especially of our audience, which I know a lot of you are non-technical, when you hear that, you're like, no, that's not for me. Right.

So this is really just to challenge you because here's the thing. Let's just say you work in HR. Guess what you're drowning in? Documents, right? Onboarding forms. You're trying to personalize certain policies. You're trying to update, right, when certain laws come into effect, certain labor laws, etc. Right.

You're drowning in long documents, right? So this same thing can be used to help you better visualize and understand key elements of your work. Right. When we talk about visualizing data, sometimes data is just words, right? Sometimes it's just piles of documents. So I want you to think of what is that thing for you, right? For me, a lot of times it's stats, right? It's podcast stats.

It's stats from our newsletter. It's stats from social media, right? But not just that, but it's really triangulating those different stats and finding the connections, right? What type of podcast episodes do you like, right? What are things that, you know, bring in a lot of new audiences and retain them across multiple platforms? That's something, right, to do at scale.

Generally, you need a big team to do that, right? A lot of the, you know, popular podcasts out there that, you know, our show is luckily rubbing elbows with, you know, on kind of the top podcast charts. They have large teams, very large teams. We don't. Our large teams are exactly what I'm showing you, right? I'm literally showing you how we run our day to day. That's putting AI to work at Wednesdays.

I'm showing you how I use it. So I want you to do it as well. And I need you to think of the future. Talk is cheap. So is text, right? Unfortunately, I think AI has really cheapened the written word, which sucks for me as a former journalist, as someone that, you know, I would say is above average at writing. I've won a couple writing awards in my day, former Pulitzer Fellow, all that good stuff.

But text is cheap now and overwhelming, right, because of large language models, AI slop, right? Obviously, when we think of AI slop, we now all think of, you know, AI photos and AI videos, but the OG AI slop was text, right, walls and walls of text. And I think that not just what we output is going to change and what we're expected to output is going to change, right? I'd say for the majority of our listeners, you know, what are your outputs? You're responding to emails, you know, maybe PowerPoints, Excel sheets, right? Normally text-based things, even, you know, PowerPoint's pretty much text-based and you try to make the text-based. And you try to make the looks pretty. It's going to change in the future. Right? I love one of our recent guests said

something, demos over memos. And I think that's the future of work. It was Richard, chief evangelist over there at Google Cloud. He said demos over memos, I think that's how we're going to be working in the future with AI. Right. And just putting out text-based work, whether it's emails, presentations, documents, Excel sheets, whatever, it's not going to cut it for much longer, right? Because I think how we learn, how we interact, and how we build value for our companies is going to be, well, in a very visual, multimedia and interactive experience, like what we're going to be showing you here today. Because what we're going over, this is like vibe coding.

I'm not going to call this vibe coding 101 because it's not. This is vibe coding, oh, oh, one. This is like if you were in a college class and you wanted to take vibe coding 101, This is like one of the first prerex that you need before you go to vibe coding 101. Because ultimately, this is what we're doing. We are having chat GPT, Gemini, and Claude us useful information that we as non-technical business leaders can do. And like I said, this is so easy, but I think so few people take advantage of this. And it's best, I think, when you stack features and make it simple.

Because even what I'm doing right here, it's technically not easy. right? Because not only am I having to, you know, get the data together to give to a large language model, but normally what you're faced with is writing out these long instructions, right? So this is where the GPTs inside chat GPD, the gems inside of Google Gemini and Claude's kind of projects and the custom instructions and projects really comes in handy because if you get it working right and you can build in that reproducibility and scalability, right, in these custom instructions that I'll share with you, It doesn't matter or it shouldn't matter, right? I'm going to fine tune it a little bit more before I send it out to everyone,

but it doesn't matter if you're uploading a bunch of spreadsheets, a bunch of documents, a bunch of text, meeting transcripts. It should be able to provide you instant feedback and instant visualizations that you can really take advantage of. All right. So let me just real quick give you the pros and the cons of each. All right. So chat, GBT's Canvas mode.

Okay, technically chat Chb-T's canvas mode, there's a duality to it. The same thing with Gemini's canvas mode. Number one, it's kind of like an interactive text editor, right? So here, I'm just going to talk about chat Chb-T Canvas and Gemini Canvas in the same breath. A lot of people think of it as an interactive text editor, right? So instead of going back and forth in a chat style with a large language model, you have like a dock, right? And you can go in line, right? So if you have a 10 paragraph

document and oh man, paragraph number six stinks, you know, normally in a chat interface, you just tell chat TPT or tell Gemini, hey, keep everything the same, but change paragraph six and oh,

shoot, it changed paragraph two and number eight as well. So, you know, in a canvas mode, it's interactive and in line, right? You can put your cursor in the middle

and highlight something and say, just change this. And it just changes that. People think that's what Canvas mode is in Chad, Chbett and in Google Gemini. Well,

that's the half of it. The other half is it can run and render many different types code. So you can have these interactive environments. It's almost like you're spinning up apps or you're spinning up demos, right, demos over memos, in chat without knowing how to code, just being like, yo, chat, yo, Gemini, here's a bunch of information. Go build me something useful, right? But then having that scalability of the custom instructions on top of it. So I think that chat chitbcannvas, it can be finicky, right? Especially in my case right now, when I'm using it with

a GPT because if you look in your GPT settings, you can enable, people don't know this, you can enable or disable Canvas mode, right? So obviously in my example, I haven't enabled. But yeah, it can be a little fiddicky when attaching files in a GPT while using Canvas mode. So I know that's a lot there. One of the plus sides of using chat ChaptipT canvas is if you tell it in the prompt to use Canvas, you don't even usually have to click the Canvas button. In Google Gemini, you do.

Even if you explicitly tell it in the gems custom instructions to use Canvas mode, it won't always do so. All right. I think the highest, the best visual outputs in the fastest amount of time, if we look at speed, quality, and reliability, I think it's Google Gemini. Hands down. The highest ceiling might be clawed, right, with their artifacts. And I'm going to talk about that a little bit more when I show you that. So that's their version. So we're using projects, but their version, the thing that renders code is just called artifacts.

It's not an interactive inline editor like canvas mode, but it does render code. And it's really good. The ceiling is high. It's finicky, though, right? So in terms of reliability, I think reliability and speed, I think Google Gemini takes the cake. In terms of flexibility, I think it might actually be. chat GPDs just because you can use GBT's in a variety of different ways, right? And when you can bake in canvas mode in a GVT and take advantage of all the different scaffolding essentially in

the agentic nature of the GVT51 thinking models, it's really good. It's really robust, but it's slow. But what it works, it's great. And then I think of the middle ground there, you have clawed, super high ceiling, but still fast enough, but finicky as well. So hopefully that helps and it's really just down to your own experience. All right. So enough on that. Let's check in live and maybe we'll cook up one or two other quick experiences here. All right.

So let's go and check out. There we go. All right. Cool. We got it. So on my screen here, I am in chat. GBT. So this is when I just pasted in the information because like I said, I am using a GBT.

And for whatever reason, when you upload, a file, it can be a little finicky. So I just paste it in all the information. And now you'll see it created a dashboard for me here. It says, Everyday AI podcast performance dashboard. All I have to do to launch this experience is click the preview button. Okay. So now I'm going to try to resize this a little bit here to make the dashboard,

the interactive dashboard a little bit bigger. And here you see, I'm going to zoom out a little bit on my screen. And I have a pretty nice dashboard. So it did a nice job. So it gave me kind of a running total of the number of downloads of my podcast over time. Really cool. It's interactive. I can hover.

It's actually really impressive. It goes day by day over the course of three years. And it shows me running total downloads. That's pretty cool. I don't even have that in my podcast platform. It breaks down the number of total downloads by category type, which is cool. Another thing I don't have, right? Great thing about large language models is it takes unstructured data, right?

So just looked at the title of my podcast and then it started to categorize them automatically. I don't have those categories baked in. So, you know, it looks like strategy is my top performing podcast category, then news, then how to tutorials, et cetera. So pretty cool. If I scroll down here, I have another, looks like some of the last 10 podcast, the total number of downloads. I have a, oh, cool here. Okay, nice.

Yeah, this is pretty good. Chad, Tbti, you know, I kind of knocked it and said it can be a little finicky. Really knocked this one out of the park. So I have a different window here. I can do the last seven days and it's giving me the episode and then the last seven days of downloads, last 30 days of downloads, last 90 days in all time. That's really, really good because, again, these are stats that I do not have in my podcast. host, right? So you can see immediately where the value already lies. For me, this is amazing,

because a lot of times I have to go export all my data, you know, go in, put a couple filters in a spreadsheet, right, pre-Gen AI, run some formulas, spent a couple hours, Googling how to write these formulas, they're not working, right? Essentially, you're, you know, building a very archaic simple man's version of an algorithm inside of a spreadsheet. I have it all here. It did it for me. This is very, very impressive. All right. So let's see if there's anything else. So nothing super robust in terms of the amount of data,

but the data that it did give me really, really good. So chat, jubt, good here. All right. So let's look at Gemini. All right. So unfortunately, I can't kind of zoom too far in and out. But same thing. It gave me a very, very good.

looking dashboard here. So similarly, it gave me linear growth over time for my podcast. It gave me average episode downloads. It's not correct, though. So it looks like there's maybe some issues here. A growth trajectory. Oh, here's why. For whatever reason, it only sampled 13 episodes in my data set. All right. I've done this similar thing before on my Ultra account, and it did a little bit better. So this is probably a limitation of the paid account that I'm using, but to have all of these in one window, I just did this for simplicity. So maybe not the best example for Google Gemini, but hey, it looks like maybe I'm proving myself wrong. So if I'm going head to head here, I like the chat chabit version. But visually, I think the Gemini one is better. Utility,

which is ultimately more important, the chatypt one is better. So now let's jump into Claude. Let's see how Claude did. I can resize this. So a little more simple and straightforward, not quite as flashy visually. So this gave me kind of everyday AI podcast analytics, different tabs that I can click that are, well, I thought they were sortable, but they're not. All right. And I did use Opus 4.5. All right. That probably is worth noting. So I used Gemini 3 Pro. I used Opus 4.5. And then I used 51 thinking. So I used kind of the most powerful models for each of these. So the Claude artifact, not that great.

Like I said, there's these tabs here. They don't do anything. So it didn't work correctly. Again, I'm one-shodding things here, right? So it has some monthly, monthly trends. I'm scrolling through these, and these are, maybe they're correct. Performance by topic. So again, this did a good job of breaking certain things down by topic.

and then giving average downloads. So it looks like AI agents is top. Then how to tutorials, business ROI. Okay, cool. That's pretty helpful. It looks like, hey, Claude even found my Claude in Anthropic episodes are not very popular compared to Chad GPT and Gemini

that are more popular. All right. And then it has a content performance. That's cool. So it says when I have a numbered list podcast, it usually gets a little bit more downloads, tutorials. There we go.

Cool stuff. And then it has a couple key insights from my data. So maybe a little bit to be desired from a user experience in visual standpoint, but it did the job. So all three of them, just right there. Even if you're not able to see my screen, again, go to your everyday AI.com, watch the video version if you care. They all produced one shot, right? I wasn't iterating back and forth. I did this live. Probably with one or two follow-ups on each of these. I'm telling you, this is, right? When I say this is, coding zero one.

And when I told you all last year that it would become very common for non-technical people to just be building their own mini apps, this is it. I told you. My podcast provider doesn't even give me some of these things that I had, right? Because it doesn't have this layer, number one, it doesn't have the layer of large language models that I can, you know, manipulate and better understand data and turning unstructured data into structured insights. but then obviously having that, you know, a combination factor,

the custom instructions that I gave each of these models to really deliver me what I want. All right. Let's go ahead. I don't know how this is going to go, y'all. We're going to try. All right. I'm going to try. Let's see.

I have another file. I'm going to do, what are we going to do here? I have a document. All right. this is an old, I'm just going to say, please visualize. So I'm doing nothing. I have a transcript from a meeting. This is an older meeting.

I defined one where there wasn't an external client. This is just two of my coworkers. This is, I don't know, a couple of years old. We were having a meeting about advertising on meta. I think it was, you know, we did some ads for everyday AI, maybe like a year and a half ago, two years ago on meta. So I think that's what this is. So I'm actually not sure.

We'll see once it maybe does this. All right. Let's see. So again, we're having this issue here in chat, with the GBT and uploading files, this combination for whatever reason, not working very well. So I'm just going to copy and paste this and then do nothing else. So we'll see if that works a little bit better. All right, we're going to do the same thing in Gemini.

We're going to go back into our gem. Let's see if this works. I'm just going to upload this meeting transcript and absolutely nothing else. All right. I'm going to make sure, whoops, I don't want nanobanana, although that could be fun. All right. Let me try that again. So here I am in my gem.

I'm just uploading this transcript doc, right? And then I'm enabling canvas mode and going to town. And then the same thing. I'm going to go back into a in, let's see here, going to my projects. That's not the right one. There we go. All right, go to my projects here inside Claude. I'm going to upload that same meeting transcript.

And let's see. Let's see if we can get something usable here live. And we're not giving this one as much time, right? I started the one before with more time to spare, right? Gave myself a little time to put it in the oven and cook. But, well, now we can kind of see how they do this live. And, you know, I'll tell you a little bit more about some of the custom instructions that I put in here. I did this to handle any type of data.

So like I said, whether you throw in multiple spreadsheets, PDFs, aside from the bug that's going on inside chat TPT, which hopefully they squash. Aside from that, you can really put in any type of data, including meeting transcripts, right? And the cool thing that I baked in here, you know, I said, don't just visualize, right? Don't just visualize things.

Help solve problems. So even in the example I said, and we'll see if it works because I didn't even, test this part out. But all of these models are agentic by nature, right? So Opus 4.5 from Claude, Claude Opus 4.5, Google Gemini3 Pro, and GVT51 thinking, they are agentic. So what does that mean and what does that mean when we are kind of stacking or combining these capabilities? Let me break it down for you really simply, right? They're not just going through in coding, in parsing data, in building things. They are doing that. But,

they're thinking about things, hopefully step by step, right? Chain of thought. But also, I've built them to go solve problems. So hopefully what it should do is not just tell me what went on in the meeting and hopefully give me a visual dashboard to show some of those things, but what it should be doing, hopefully, we'll see. Now I'm going to be kind of looking through the chain of thought a little bit here. It should be solving the problems, right? So there's probably some next steps that we talked about in the meeting, right? And maybe you pay for some, you know, AI software that does this for you. Well, here, this does it by default, or it should do it by default. It should pull out kind of action items, next step, to do's, right, which is pretty cool. But then it should hopefully

go out and start researching those and providing potential solutions. Yeah, no AI platforms do that. But I mean, I don't want to over promise a set of custom instructions. But hopefully it does, right? But that's what? These models are agendic by nature. They should be able to go through, see that, and then go start researching things, be like, wait, not only do I need to extract to-dos. I need to visualize this meeting transcript, but I need to go out and help these people that had problems or next steps or things that they were going to research and do it for them.

That is what an agentic model should be doing, right? When it can start down Path A and be like, nope, this isn't right. It can go back, research more, and go down a different path. All right. So hopefully we'll see that here without me having to, go on for too much longer. Yeah, because this is a little bit more of a complex situation. So let's just see as an example. I'm opening the chain of thought here inside chat chpd, so I can kind of see what's going on. And I'm going to see if it did this right. So first, it says it's building a user

interface for a meta ads overview dashboard. Cool, because there's probably some statistics that we talked about out loud. So it's probably going to be pulling those and creating a dashboard. I really wanted to see if it did what I snuck in there. Yes, it did it. All right, ChatTbtee did it. And I assumed that I was going to be able to see that because one thing that I love and one of the reasons why I still use Chad ChbT as much as I do and not Gemini more is because in the chain of thought, at least right now, Gemini doesn't show when it goes out to certain websites, which I always need to know. I have talked to the Google team. I do think that they're going to be building that in here. But now I can see by.

reading the summarized chain of thought. All right. This doesn't, it's not crazy. Just, just go in. I'll give you these custom instructions. You put your own stuff in there. There should be a little button on there that says thought for. All right.

So it's kind of hidden. You have to click it. And then I can go see exactly what this GPT with my custom instructions using this very powerful model did. And I see it did it correctly. And it went and it looked up. So cool. It went up and looked,

uh, looked up potential solutions for things. we didn't even know at the time and it went out and started solving our problems. So I'm excited to see how this one is ultimately going to show this information. So it's still working, but cool to see that that little custom instruction working in action and it should work across multiple different, different things. All right, let's look at Gemini. So Gemini, like I said, fast. It's done. My gosh, this looks really good. All right. Yeah. Okay, sweet. So, interesting.

Yeah, this is nice. So like I said, this meeting was maybe like a year and a half ago. Okay, yeah, here's the date. Okay, yeah, I was right. It was from July 2024. So it was an older meeting. So this was with some of my colleagues and we were talking about some different tactics that we were doing on meta ads to, you know, we're just more or less, you know, testing different

customer acquisition costs, right, to get people to sign up for our email newsletter if we were to do meta ads. I completely forgot about most of this. But yeah, so it gave us like, hey, a list of our inactive targets, ad relevancy. Very nice, right. Again, interactive. So number of users, views. So it looks like it pulled out different things that we talked about out loud, right? So we We didn't spin out every single metric out loud, but those that we did, it put them in a nice chart there, which looks good. And then we have, this is like a meeting dashboard, which is really cool. So then it says tech stack status. So it looks like there was a failure on one of the pieces of technology called Split Hero that we were using.

Something about Beehive integration said that it was going correctly. That's good. There's an action plan here with little status indicators. Really cool, right? So we needed to fix the split hero because the MAV2 was not triggering. The owner on that was Brandon. Right. So it literally created like a dashboard, almost like Canban style, you know, follow up action plan.

And then it had a creative and offer pivot. So it went in offered some examples on what we could do. So very cool. I don't think it did a fantastic job of going out and finding new solutions. So I don't know if it, you know, just used its own. training data. I don't know if it went out and, you know, looked at certain up-to-date websites, but regardless, y'all think of all these meetings that you're in. Imagine if you could have a custom personalized dashboard like this that just visualizes everything. So like I said,

yeah, there's a lot of AI note takers out there that give you, you know, next step, action plans, all that, but combine it with like Gemini 3 Pro. It's good. All right, podcast audience, you can't see my face, but this is one of those where I'm like, what the heck, this is really, really good. All right, I want to see what Chad GPT is cooking. All right, here we go. Because I'm excited because the Chad GPT one, I can see the websites that it went out to. So it might have done a really good job at kind of solving some of these problems for us, right? When we talk about the future of work, this is huge, y'all.

All right, so I'm going to click preview. All right. So now it's previewing. Let's see what we got. All right. So here is our meta ads overview. kind of the people in the meeting. Oh, this is cool.

Top talker. So it analyzed who talked the most. The number, the words, the key ad topics, very cool. So it did some kind of like word clouding. Very cool. So it counts how often key concepts came up, right? Not just by the words, but it categorized what we talked about most. Really cool.

So it looks like we talked about the creatives the most, the PPP course, then the newsletter, then the audiences, then community. versions, very cool. A number of people speaking and the number of words. That's awesome. All right. Let's see what we got here. Okay. It gave kind of, I'm not sure. It didn't create a label for this whole section, but it, again, it looks like it kind of had some follow-up pieces here. And then, okay, and then it looks like it pulled in some new information from the web. It got some benchmarks and it gave us some average meta CPMs from October 2020. 25. Like I said, this was from July of 2024. So I'll tell you this. And then it says like money, you know, there's a money mentioned section. So pretty cool. I wish I wish I could combine these two different ones. Right. And here's the thing, y'all. I can go back and just say make this better or, you know, I can just say, you know, make, you know, I'm not going to, I might not wait for this because it might take too long. I'm going to say, you know, create better headings.

next steps and research potential solutions for these issues discussed, right? This is all iterative. Everything I did so far was one shot. I didn't even put a prompt in, right? It was built into the custom instructions. I just uploaded a document and this was all done for me. Right. So imagine it looks like it's still, still rendering here because it's spitting back all of this

information in text. So imagine just going through a prompt or two, right? At that point, that's when you have bespoke software for whatever you do. So then imagine if you have the same type of meeting once a month and it's huge and it's a big part of what you do. You have now a literal personalized piece of software that you can use reliably and consistently every time just by updating it. And that can be your kind of home checkpoint every single day. Maybe that's where you start your work, right?

So very cool. Let's check in on Claude. All right. So Claude, here we go. Surprisingly, nice visuals from Claude. Not that that's what I care about the most. All right. So it says the meta ads performance review. I'm going to see if the tabs work this time.

For whatever reason, when Claude builds tabs, they don't always work. Okay, this time it did. So this one's pretty robust if all this works. Okay, this is nice. Claude probably won this one, Opus 4.5. impressive here. I do want to look, though, in the chain of thought here. I want to see. All right. I'll have to look at this a little bit more later because there's a lot of steps. But it did give me a nice overview. It talked about total ad spend, landing page users, all these

things, performance benchmark, gave me some nice charts and graph. It kind of, okay, this is cool. it kind of labeled people, even though we didn't label ourselves, right? It said, Brandon was the marketing lead. George was the ad specialist and myself, the decision maker. It's kind of true how that meeting was set up, right? So it did a good job of even understanding who is who based on the discussion, right? Same thing. It broke down key discussion topics.

Issues, great. It put them color coordinated. I think Claude definitely won this one. action plans really, really good. And then let's see what else it did here. Recommendations. Really, really good. So, hey, on this one, Claude did crush it.

So Opus 4.5, impressive. Whereas on our demo last week, it kind of missed the mark. I said, don't write off my one little show here. Kind of redeem itself a little bit. All right. So that is a wrap. I wanted to do a couple. I didn't want this to turn into an hour-long show.

So here's here's the reality, y'all. This is the future of work. It is demos over memos. All right. And I think you, hopefully if you saw this, you can understand the potential. And why, you know, when I said a year ago that by the end of 2025, everyday non-technical people are going to be building themselves micro apps. They're going to be building themselves little pieces of software. You saw me do it just by combining two simple things, right?

custom instructions with any type of data or documents. And that's something that we can all do. Regardless of your role, if you sit in front of a computer for the majority of your day, if you are a knowledge worker on the internet, you can do what I just did. It's very simple. You don't have to know anything.

All you got to do, well, you got to go share this show. All right. So I will send you over the custom instructions. You might want to tweak them a little bit yourself. You can read through them, but go repost this show on LinkedIn.

All right. So if you are listening on the podcast, I always put the LinkedIn link for this very show. All right. So like I said, go find episode 665 on LinkedIn, repost it. Give me a couple hours, couple of days. And I'll send these custom instructions over to you. And you can get going right away. So that is it.

That is putting AI to work on Wednesdays and the beginner's guide to how to visualize data with AI in chat, Chbite, Jibb and Claude. I hope this was helpful. Well, if so, tell someone about it, then go to our website, your everyday AI.com. Thanks for tuning in. Hope to see you back tomorrow and every day for more Everyday AI. Thanks y'all. Meet Firefly AI Assistant.

Now live in Adobe Firefly, the Allman One Creative AI Studio. Just describe what you want to create in your own words and the assistant handles the rest, orchestrating multi-step workflows across Adobe Creative Cloud apps, including Photoshop, Premiere Express, and more in one conversational interface. You direct the outcome while the assistant, accelerates execution. Stand control with the ability to step in and refine at any time. See it today at firefly.adobie.com.

And that's a wrap for today's edition of Everyday AI. Thanks for joining us. If you enjoyed this episode, please subscribe and leave us a rating. It helps keep us going. For a little more AI magic, visit Your EverydayAI.com and sign up to our daily newsletter so you don't get left behind. Go break some barriers and we'll see you next time.