Everyday AI Podcast – An AI and ChatGPT Podcast - EP 544: AI Magic - Convert Outdated Content into Engagement Gold

Episode Date: June 11, 2025

Got old docs that need updating? Yeah, yeah, yeah. You can do that with AI. But that's as basic as a Pumpkin Spice Latte in October. What if, in a few minutes, you could not just bring life to your old docs with AI by making them interactive, but also ADD AI functionality into those docs? We show you how it's done.In our new segment -- Working Wednesdays with AI, we tackle practical use-cases that even non-technical people can pick up and run with. Don't miss this one. Newsletter: Sign up for our free daily newsletterMore on this Episode: Episode PageJoin the discussion: Have a question? Join the convo here.Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineupWebsite: YourEverydayAI.comEmail The Show: [e-mail] with Jordan on LinkedInTopics Covered in This Episode:Transform Outdated Content with AIAI Tools for Efficient Document UpdatingGoogle Gemini Deep Research CapabilitiesAI Studio's PDF Transcription FeatureEnhance Presentations with AI StudioInteractive Presentations with Gemini CanvasEmbedding AI in PresentationsGoogle AI Studio's Developer FeaturesTimestamps:00:00 Transforming Document Creation with AI03:53 Rethinking AI in Content Management09:29 Updating Small Language Models Presentation11:55 AI Workflow Insights14:30 Google Gemini Evaluates Web Data17:27 Exploring Google AI Studio20:54 Unscripted Presentation Update Strategy25:17 "Deep Research and Presentation Updates"29:23 "Redefining 'Small' in Language Models"33:00 "2025: On-Device AI Revolution"35:43 "AI-Enhanced Slide Summarization"38:42 "Interactive Live AI Chat Widget"41:34 "AI's Efficiency Explained"43:41 "Everyday AI Practice Wednesdays"Keywords:Google Gemini, Gemini 2.5, AI Studio, Google AI Pro plan, Ultra plan, generative AI, convert outdated content, engagement gold, AI magic, interactive presentation, Commonplace AI tasks, Deep research, AI workflows, Google AI Studio capabilities, AI document transcription, Google Gemini Deep Research, interactive and slick interface, embedded AI capabilities, transcribe PDF presentation, factual keyword updates, targeted deep research, Google search grounding, transforming outdated documents, Canvas mode, integrate AI with presentations, automate mundane tasks, creative AI applications, AI-driven efficiency, enhancing content with AI, live AI capabilities, small language models, presentation refinement, knowledge worker efficiency, research improvements.Send Everyday AI and Jordan a text message. (We can't reply back unless you leave contact info)

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 in 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. I've spent thousands of hours of my life using AI to create documents, which is kind of wild when you think about it.

But yeah, I actually did the math since 2020, myself and my team, we've been using generative AI, large language models, to create documents. but as of the last couple of months, that process has completely changed. And I think it's something that all of us, if we're a knowledge worker, if you get paid to sit in front of your desk and create value for a company, which is what many of us do,

you probably also spend hundreds of hours a year. Now, using AI to bring new life to old documents. And it's something I think that a lot of us think, oh, well, if I'm using AI to do this, I'm doing it probably the correct way. And I'd argue you're probably not. So that's what we're going to be tackling today on everyday AI. A little what I like to call little AI magic and how you can convert outdated content into engagement gold. We're going to be doing it live today for our new weekly series Put AI to Work Wednesday. All right, I'm excited for this one.

I hope you are too. What's going on, y'all? Welcome to Everyday AI. My name's Jordan Wilson and I'm the host. And if you're trying to grow your company and career with generative AI, you are in the right place. We do this every single weekday, Monday through Friday with our unscripted, unedited, daily live stream podcast and then putting it all out in our free daily newsletter. So if you're trying to grow your company and career with generative AI, you are definitely in the right place. So if you haven't already, please make sure to go.

to Your EverydayAI.com. Sign up for our free daily newsletter. We're going to be recapping everything that you need to know from today's episode in that newsletter. All right. If you want the daily AI news, sometimes we go over that on the live stream. Just go check out the newsletter today. But let's get straight into it.

And you all wanted this FYI. You said, hey, yes, let's do this new weekly segment here on Everyday AI, where I show you how I'm using AI for my company here at Every Day. or I might do just some other demos. Maybe they're not as applicable, but you all said overwhelmingly, yes, once a week, let's show up on Wednesdays and learn together. So if you're on the podcast, I'm going to do my very, very best describing this. But just so you know, we always put a video.

Yes, this is an unedited, unscripted. It goes out on video as well. So make sure to check the show notes in your podcast. This is one of those. It's going to be a little bit visual at the end. If everything works, right? Love doing live demos with generative AI. Nothing ever goes wrong there.

But this might be one of those after you listen to it on the podcast. You're like, wait, I have to see how that's done. So make sure it's going to be in the newsletter as well. But make sure you can go watch that video. So here's the problem that we're going to tackle. And if you think about your work, right, don't pay as much attention to the actual example that we're going to be going over today. But I want you to think about your work.

How much time do you spend updating old documents or repurposing old content? And you're using AI. And I think sometimes we fall into this false sense of security. Like, oh, well, as long as I'm using AI at some part of the process, I'm probably, you know, being efficient. And are you maybe? But I feel that there's so many new features that the big AI companies have rolled out in the last, just the like last two months alone that have honestly kind of even

challenged me for someone that's, like I said, I've used generative AI. AI since 2020. And I'm even having to rethink and rework my current workflows, but the average person, if you're a knowledge worker, you're spending probably hundreds of hours a year, whether you know it or not, updating old content and probably using AI to do it. Or maybe you're not. And if you're not, then this is really going to be helpful. And just doing it sometimes manually.

And when I say manually, yes, even if you're using AI, as part of a workflow. I think so many of us are still doing many manual steps. So today, if you stick around for the next, you know, 25 minutes, I'm going to try to make this one a tight one. Here's exactly what we're going to learn. Here's what we're going to go over. We're going to use AI Studio to transcribe a visual PDF presentation.

I'm going to tell you why that's important. We're going to perform a targeted deep research based on that content with Google Gemini. We're going to use AI Studio, Google's AI Studio, to bring factual keyword factual new life to an older presentation. And we're going to embed, get this, we are literally going to embed live AI capabilities into this presentation. Yeah, we're essentially going to build a piece of AI software from an old document. That's why I'm calling this episode AI Magic because it is literally so easy to do now.

And I think this new announcement from Google literally got swept under the rub. like no one's talking about this and it's actually pretty amazing. So stick around. That's exactly what we're going to do. But before I go in and we're going to start working through this live, I want you to think, especially if you're on the podcast, what's that one thing you do? Right. The one thing you do maybe every day, once a week, a couple times a month, handful of times

a quarter, right? Maybe you're up, you know, maybe you have an onboarding form, right? you know, or you create your company's onboarding, but there's new state laws. You have a new employee handbook. So you're kind of having to pull information from multiple documents and updating an old document. And then you're probably having to look up some information on the web as well. And maybe you're using, you know, chat, GPT to, you know, pull something from these two documents. And then you're using Google Gemini to help you expand one.

And then you're, you know, using perplexity to help you research. And yes, you can do that. But I want you to look at the process. we're going to go over today. You know, it's not fully automated, right? I'm not going to, you know, sit here and build an automated workflow. And, you know, with, you know, agentic AI sprinklings. I want to do something very simple, very basic.

You don't need to even have any experience. All right. I'm going to walk you through how to do this. And the crazy thing is most of this can be done, even if you don't even have a paid account, right? But even if you just have the basic Google Gemini, Pro plan, $20 a month. Or if you're a student, it's literally free for the next year. All right.

That's all you need. So again, I want you to think what is your use case before we go any further? I want you to think. Do you have that use case yet? All right. I'm going to take a sip of coffee because this one might get a little wild. And I want you to have your use case. Even, hey, if you're following on the live stream, go ahead, do this live with me.

Pull up Google AI Studio on your computer. if you want. So that's just AI studio. Google.com, a pull up Google Gemini as well. And then we're going to get into it here. So let's start live. Okay. So first, I'm going to show you exactly what we're going to be trying to do here. And live stream audience, let me know. Can you see my screen?

I think you can. But if you could let me know, I always appreciate it. Because one time I was going on for like eight minutes. And then I finally looked in the comment. section and someone's like, oh, Jordan, we can't see your screen. You'd think after 540 times, I would have this down now. But yeah, sometimes when you're just doing these live demos, you know, they play tricks on you.

Okay, so here's what I'm trying to do, right? I said, hey, what is your use case? What old document do you want to bring new life to? Do you want to make interactive, right? That's another thing with generated AI. I think we have to relook at how we present information, whether that's internally or externally, right? I think so many things that even used to be boring, power.

They can be interactive websites. That's essentially what we're going to do now. We're going to turn an old, boring website. We're going to use AI to quickly update, validate, expand that information. And then we're going to not just make it interactive, but we are literally going to add AI elements to it, right? Crazy. All right.

So here's where we're starting. I did a presentation last year. Okay. So live stream audience, you should see this on my screen here. I did an episode called small language models, what they are, and do we need them? I believe this was, I should have looked this up probably about two years ago, maybe a year and a half ago. So what I'm trying to say is I want to do a new presentation.

I want to do a new episode on small language models. And this is, you know, the outline is pretty good. But I know a lot of it's going to be old. And I also want to make this a little better. Okay. So that's ultimately what we're going to be doing here is we are. are going to be turning this old presentation into something better, something interactive, but we need to completely change and update the content.

Because the content in here, right? So, you know, it's what is this small language model? 14 facts you need to know, right? So I go through here. And although the definitions necessarily haven't changed, a lot of the information in here is now extremely outdated. Okay, so we're going to bounce around because certain things take a little. little longer. All right. And I kind of have some copy and paste prompts ready. So what we're going to do,

because this is what's going to take the longest, I'm going to go into Google Gemini. So that's just Gemini.com. I'm going to click the deep research option here. All right. And I'll talk you through deep research as this starts. I just put a prompt in. I'm going to read it out loud here in a second. Google Gemini is creating a plan on how it's going to do deep research. I'm going to have to approve the plan. And then we'll, I'll be able to talk through that once we get. Or actually, just let me tell you the, the prompt right now. So something simple. I'm actually going to click start research.

So I said, put together or sorry, please put together a comprehensive report on month by month happenings of small language models in 2025. Be detailed, accurate, finding examples, understanding trends, et cetera. So again, nothing, nothing crazy there in the prompt. Essentially, it's just being like, hey, go fine month by month, what has happened with small language models in 2025? Because like I said, I think this presentation is either from late 2023 or early 2024.

And regardless, I know large small language models have changed so much in 2025. I don't even want information from 2024 or something like that in this new document, right? When I'm doing a new episode on everyday AI, I do like to understand historical trends in content, context, right? But I don't like to present old information. This is one of the things, right, as we do this AI work, you know, put AI to work on Wednesdays. I want you to kind of go into my head because people will always be like, oh, Jordan, like, how do you use AI, right?

This is one example. I have much more complex and automated workflows, but I wanted to start with something simple, even though there are some manual steps here to put these different pieces together, I wanted you to kind of see what we're doing. Okay. And if you haven't used Google Gemini's deep research recently, it is really, really good. Okay. And here's why.

Because back in what was that? That was February? I believe I should know that. When was I in Las Vegas for Google I. When was that? February, I can't even see it now. May. No.

No, I wasn't at, gosh, this is why I need to sleep more, y'all. I wasn't at that one. I was at Cloud Next. All right, that was in April. All right, there we go. So Google updated their deep research in April with Google Gemini 2.5 Pro. This is by all really measurements, the most powerful model in the world. Google even updated it last week, even though it was the most already the most powerful model

in the world by almost every single benchmark imaginable, including blind, the kind of blind taste test that is LM Arena. So one thing, when they updated their deep research, they made it much, much better. So if you don't know what deep research is, as we do are like, let's put AI to work. Remember what perplexity was, right? Like a year ago. Think of that times a thousand. But the difference is this is using a thinking model.

So Google Gemini 2.5 is. is obviously a hybrid model. So when it needs to think and go very slowly and plan like a smart researcher would, it will do that. When it needs to just be fast, it'll do that as well. So I can go here and it is thinking about the research, right? So I can go through. I'm not going to read this,

but it's actually first thinking about my very simple research prompt I gave it. And then it started to do a round of research. So it looks like it went to about 10 websites. And then after going to those 10 websites, it actually started to first reflect and think of the information that it found on those 10 websites first. Right. Pretty interesting here that out of the first 10 websites, it went to a lot of Microsoft websites, which I find interesting. Right. So, you know, in this use case anyways, even though I think Google's Gemma 3, Gemma 3N is the best small language model, it's actually looking at information from all across the web.

So, right? So I do like that about Google Gemini. It's not just, you know, saying like, oh, Google's, you know, small language model is the best, even though it pretty much is. All right. So then it's finding this information and then it's saying, hey, based on all this information, I've actually now figured out, I have to go research more research. So first, it went out to 10 different websites. It thought about it. It reflected on what it found in its research based on my original query. And then it decided, yo, I actually need to go find way more information.

All right. So then it went out to, uh, looks like another, uh, 12 or 15 sites. Same thing, right? So it is agentic in its nature, right? So I didn't have to do a lot of tricky prompting in order for it to accomplish this level of research, right? So as we're scrolling through live stream audiences, seeing it, it's doing this now that's seven times, eight times, and it's still going. So we're going to let this finish because we're eventually going to use this information in part to partially update our presentation. But now you have the first part of our little AI magic tutorial here.

All right, before we get going, have to pause first for a word from our sponsors at Google. This podcast is supported by Google. Hey, everyone. David here, one of the product leads for Google Gemini. Check out VO3, our state-of-the-art AI video generation model in the Gemini app, which lets you create high-quality, eight-second videos

with native audio generation. Try it with a Google AI Pro plan or get the highest access with the ultra plan. Sign up at jemini.gov to get started and show us what you create. All right. So this is probably going to take another couple of minutes.

So I'm going to move on to our next step. And then we're going to check back on this deep research. I do find, and this is not just with Google, Gemini, same thing with chat, GPT, other platforms. sometimes the deep research looks like they stall out, but it actually doesn't. So if you're ever like looking, kind of like on my screen, it's like, wait, is it still going? A lot of times what you can do is just there should be a toggle here, kind of like a, you know, it looks like a clock.

And you can make sure it's like, okay, is it still moving? Is it not moving? I've had this happen with just about any platform that does deep research. You can always just refresh the browser, FYI, and you're not going to kind of lose your work or lose your place. All right. So what is ultimately is going to happen once this Google Geminii research is done, we're going to have a document.

And then we're going to use that inside Google's AI Studio. All right. So let's go ahead and jump into AI Studio right away. So if you don't know Google AI Studio, right now it's free. Right. I say right now because there's rumbling is on the internet that, you know, Google is eventually just going to move away from this kind of free model. And you're going to have to use your own Avalier.

PI key, but y'all, I've been, and I've been saying this for literally months on the show, I feel criminal how much I've been using Google AI Studio. I mean, I should actually do the math and see, oh, if I was paying for this, how much, but I've been using it. Yeah. So, sorry if I'm one of the, you know, I'm sure there's a lot of people like me, who use this for a lot. But Google AI Studio, it's technically made for developers, right?

But I'll say this, over the last six months, a lot of the features, the user interface, the user experience, it's actually made it a lot more beginner friendly. So even though this is technically a platform built for developers, and there's going to be some settings in here in AI Studio that you're like, wait, what is this? Don't worry. It's actually very simple. And I'm going to just show you the basics of how we're going to be using it. All right. So I have a simple kind of prompt that I have ready here. And I'm going to start reading this and uploading a couple of files as we go along.

All right. One other thing to know on the right hand side of Google AI Studio, there's going to be a lot, especially if you are a beginner or at least a beginner in AI Studio. And you might not understand all these things, right? But there's a different, you know, the model drop down. You can use the different models up there. We're going to use obviously the newest version of Gemini 2.5 Pro preview 605. So this is only like three days old now.

There's something temperature. That's essentially the creativity. So you can turn that down. And, you know, it's going to kind of take away the creativity or you can turn it up. It's going to be more creative. There's some other toggles here, such as structured output. You know, if you want always Google Gemini or sorry, Google AI Studio to, you know, output something in a structured way.

You can go and build that. We're not really going to do any of that. The only thing that we're going to toggle on here is we're going to toggle on this grounding with Google search. So all that means is, yes, we're telling Google AI Studio, hey, you can go use Google, right? You can go use the web as well as the model that you're using in any information that you're going to be inputting. All right.

I do have a little bit of a long prompt here. Don't worry. I'm going to explain what we're doing. All right. So I'm saying to AI Studio, I'm sharing two PDFs with you, a presentation PDF and a research PDF. So the research PDF is still being. made, right, inside this, let's see if we're done.

Not done, but it looks like it's almost done. So we're creating a research PDF from the Google deep research. All right. And then the presentation PDF is this older version. Like I said, this is the old small language model presentation. So a lot of times when I do these live streams, I have a set of slides on the screen. So, you know, when I say it's unscripted, you know, I have like bullet points, right? but I'm just talking not like off the cuff, but I'm talking unscripted.

But, you know, I generally have some kind of slides to show the, to show the live stream on it. So I'm telling Google AI Studio, these are the two documents that I'm uploading. And then I'm telling it, ultimately, I want you to update this content for a new presentation, because I'm saying the presentation, small language models, presentation PDF is a PDF presentation I gave nearly two years ago. The second PDF is a research PDF call. than the name.

And I'm saying this is a starting point for updating the presentation, a starting point. All right. So I'm saying I would like you to do a couple of things. First, please transcribe the PDF presentation called small language model presentation PDF because guess what? How many times? How many times? Have you been handed a presentation, a document, right?

And it's your job to update this now. And you're like, who put this together two years ago? No one knows. Who has the original, you know, PowerPoint file? Who has the original like PDF? Like who has this doc, right? Who has the Google slides? And no one knows, right?

And then you're like, oh, great, what am I supposed to do? All right. So this is again, one thing that the newest models, even six months ago, the base large language models were not very good at this. Now they are. So first we're going to have Google, Google Gemini, inside Google AI Studio, transcribe the PDF. Here's why this is actually important and pretty impressive.

This PDF is a bunch of images, right? So it's literally going to use computer vision go through and grab all of this text. A lot of this information, and you'll see here, like as an example, slide three, it's just a screenshot from our website. But it's just this is a bunch of JPEGs. It's not like a proper PDF. I created it in Canva. It's a bunch of like images, screenshots, right?

In Google AI Studios, actually going to go through using computer vision. Yes, there's some tax that it can, in theory, use a process called like OCR to grab that, but it's going to use computer vision to grab all this as well. All right. So first, that's a huge win right there to be able to transcribe that entire PDF. Then I'm telling it, I need the transcription verbatim. Then I'm saying, number two, next, please analyze the document called research on small language

models, 2025 PDF, right? That is the document that I am creating in this Google Gemini deep research. As I jump over, we are done. Great. All right. So Google Gemini put this dock together. I'm going to go ahead and export it. All right.

So it's opening now in Google Docs. That's a feature. I'm going to go ahead and rename this. All right. And I'm going to save this as a PDF. All right. On my computer. So again, I'm doing this all live.

the old presentation saved. I have the new information saved. All right. And actually for use of time, I'm going to start this prompt as I finish it, right? I want to make sure that I have both of my files uploaded. All right. So it's uploaded. I clicked on the grounded grounding with Google search. And I'm going to go ahead and get this going. All right. So let me tell you kind of read out loud the rest of what I told it to do. So again, first I said, here's the old PDF that's the presentation. Then I said, here's another PDF that's the base of the research. Then I'm saying you need to fill in the gaps on what additional research or

clarification needs to be done. You should research deeply on the subject to understand trends. Focus first on finding additional information from May 2025 to June 2025. So if you remember, originally, I told the Google, Google Gemini deep research to focus on 2025. Okay. But now I'm telling Google AI Studio and the Gemini 2.5 pro model inside Google AI studio to just focus on the last like five weeks, right? Because it's been absolutely nutty. But I'm also saying you need to look.

Now you need to do a separate thing here because the deep research, I didn't say, here's my old presentation. what I'm trying to do. Now inside Google AI Studio, I am. So this is someone, again, this, I've spent thousands of hours in my life doing this, essentially combining, you know, two to three documents and then doing additional research to fill in those gaps, right? So if you've ever done anything in, you know, analytics and marketing, content creation, business, you know, business intelligence, you probably do a lot of, you know, document juggling. It's something that so many knowledge workers spend so much time. But I'm telling it, you need to go find the gaps. And then you need to go to additional research that's just information from May 2025 to June 2025 to understand

the current state and clarify any questions you have. And then I'm saying essentially number four here is after this extension of research, you need to update the presentation outline in its entirety. So in part of the first step, it's going to spit out the old presentation verbatim. It's going to go analyze the two documents that I upload. It's going to go do additional research based on the holes and the gaps that it identifies based on what I told it to do. And then last but not least, it is going to spit out a new version of that presentation. All right. So again, I just click go on this about two minutes ago.

It's already done. All right. So you can always click and I always encourage people to do this. You should be reading the chain of thought or the summarized chain of thought here. right, so you can understand what the model is doing. And a lot of times what you'll see is it's going to start to do something. You're like, wait, that's not exactly what I wanted it to do, right? And it's still going to kind of finish the task and adjust on the fly.

But this is the only way that you're going to get better working at large language models. And one of the ways, I'm being honest, right, because even this version of Gemini 2.5 Pro preview, the 605 version, it behaves much differently than the version that was released a month ago, the 506. I know that's different. You're difficult, right? May 6th versus June 5th. It behaves differently. So the only way that you can really know and understand is by kind of reading this chain of thought.

Anyways, it goes down here. It says, I'm building a framework. I'm developing a presentation. I'm synthesizing foundational details, focusing on presentation refinement. Again, this is what Google AI Studio is doing under the hood. And then it says focusing on presentation refinement. Again, a lot of that. Summarizing recent updates.

Because again, I grounded this in Google search. So it's going out and it's finding new information. So here on the outputs or the deliverables, here's what it's giving me. First, we have the verbatim transcript of the original or the older small language model presentation. PDF. Remember that, what was it? Page three, right?

That was just a screenshot, right? JPEG? Yeah, it did that. It went through. It said a grid of images with the file. following labels. Yeah. So it's very,

right, today's large language models are extremely impressive at their ability to accurately be able to see and analyze and create transcripts from very long PDFs. All right.

So part one, perfect. It went through. It properly transcribed the old PDF. Parts two and three. It did some, it says it did some research. So you can go through and read that if you want.

And then part four, it is the updated presentation outline. So this is great. All right. So I'm not going to go through and read all this right now. Maybe I'll go through and, you know, put the human in the loop later and look at this a little bit and refine it and maybe do a new updated, uh, uh, episode on small language models. So yeah, I don't know. Live stream audience. If you want to see that, just type in small, you know, if enough of you say it. I don't, I don't know. I don't know if you guys are super interested and I want to learn more about small language models. Uh, anyways, so it went through and it didn't

very, very well here. So, so I'm looking. Okay, page two, redefining the small in small language models. This is good, right, because it's saying, you know, how the definition has evolved of what a small language model even is, which is, you know, super interesting with some of the more recent small language models. So I'm going through here and I'm looking at least on first glance. It looks like this is pretty accurate. So, You know, we have the Gemma three models that were just updated and released, the Microsoft five models, including Phi four. So I'm seeing it did a lot of good research and it did additional research as well.

So not only pulled in the information that I gave it from the deep research, but I can see according to the citations in here by looking at the kind of the output is I can always click on this information. So it's citing things, right? So I clicked on this and it went to a website. you know, custom gpt. AI, right? So I see not only did it go through, number one, it transcribed my old PDF.

Number two, it looked at the additional research that I did inside of Google Gemini deep research, but I see it also by grounding Google AI studio in Google search, it went out and filled the gaps because that's important, right? Because I almost like, it was like handing it off to another assistant. It was handing it off to another person, be like, yo, here's a bunch of work. You need to double check everything. You need to find the gaps. and then you need to go out there and explore and find the gaps.

All right. This is a lot. We're not done. I said I wasn't going to keep you for too much longer. So the last step that we're going to do, two steps. We're going to turn this into an interactive presentation. And then we're going to add AI. And I'm going to challenge myself to see if I can do this for in less than five minutes.

So this is one of those times. It's like, all right, shut up, Jordan and let's put AI to work on Wednesday. All right. So all I'm doing podcast, uh, podcast, uh, podcast crew here is I'm copying and pasting these new updated slides. All right. So it looks like it gave me kind of nine slides. Perfect. So I'm copying that. I'm going back into Google Gemini. I'm opening a new chat. I'm making sure to have Gemini 2.5 pro. I'm pasting all of this information in. Okay. I'm clicking the canvas mode. That part's important. All right. And I think I have a little bit of a prompt here. Something simple, right?

You know, I'm going to be great at prompt engineering or know anything tricky. All I'm saying is turn this into a more interactive and slick interface, include everything verbatim. Okay. So I did do one test of this because I was just kind of curious and it turned out really, really cool. All right. So the thing with generative AI, y'all, is it's generative. It's a roll of the dice.

It's going to be a little bit different each and every time. So this is now we have Canvas mode inside Google Gemini. So if you don't know Canvas mode, you know, a lot of the big, large language model players have something like this. So Open AI also, their mode is called Canvas. I think GROC has their version. Perplexity has labs, which is kind of different, kind of the same in some ways, right? It can go do research and build apps.

You know, Anthropic has a very popular version of this called artifacts. But essentially, what's happening on my screen right now is I paste it in all this information. And it's building something with code, all right? And you'll see now, It's done, right? So it took like all of 30 seconds. And this is really, really good. I hate to say, this looks way better than my website.

It essentially created an interactive presentation. So it says small language models in 2025, the year of on-device, agentic and specialized AI. And then the subhead is how efficient, powerful models are moving from the cloud to your pocket. And then it says Jordan Wilson, founder and host every day. day AI and even put the, the website there, your every day.com at the bottom, really, really good. The thing is, this made it interactive. I see this little slide thing right here.

So I could, if I wanted to, instead of doing my, my old and ugly kind of slides that I build in Canva, I could literally just do this. Live stream audience. Number one, have you used Canvas before? Number two, are you impressed with this? If you're not, just wait. So let me slide through. So when I click next, it's actually really nice. Okay.

There's this nice little slide animation, right? I didn't have to build any of this. It actually looks like the branding looks pretty cool, right? It's kind of dark. It's got these pops of colors. The main page had this like gradient. Like it looks really, really good. Like this, especially the title page, this looks like a designer, like an actual designer made it.

There's literally a gradient. layer across the text, it looks really good. When I go slide to slide, there's a nice transition. There's an outline box. This looks like I used like a high-end template or I paid a designer to put this together. It looks really, really good. And then it's interactive when I flip. All right.

So there we go. So we're not done. We're not done. Here is something. And this is one of those things. Remember I said this little thing got swept under the rug. And remember I said, wait. We're going to not only update an old presentation with AI more efficiently and more effectively,

and we're going to create something that's interactive, but we are going to embed AI capabilities into it. And guess what? With Google Gemini, one click. So it's very small. This is one of those features. I wish that Google literally just did a full like keynote presentation about this at I.O. It was part of their presentation, but it should have been like, a main feature. This is wild. All right. So it just says add Geminii feature. So it's this little, you know, button in the lower right hand corner. I'm going to click it. That's all I have to do. So now I have in, well, not now in like probably 30 seconds, I'm going to have AI capabilities embedded in this presentation. And then I will be able to share this with people. All right. So it's

telling me what it's going to do. So it says summarize slide. So it says, summarize slide, click this to get a concise AI generated summary of the key points on the current slide. And then it says, deeper dive. This button uses the slides headline to ask the Gemini API for more detailed information. So that's pretty cool. It doesn't look like it's done yet. So I don't know if you understand what's going on.

But think of all the different AI tools that you've used before, right? A lot of them, you know, they're using either the Google Gemini, the open A. the Anthropic API to bring some sort of AI functionality to a website or a SaaS that you use. We literally just did that in one click. All right. So let's see an example. All right. So I'm going to go to page two.

All right. Here we go. So before, okay, it's still an interactive presentation. But now I have AI capabilities built into this thing and I didn't have to write a line of code. All right. I can literally just click this summarize slide button here. All right. And it did this nice little, you know, animation.

It says contacting Gemini. All right. And it gave me a kind of shorter and more like in kind of plain English definition. Right. So it just simplified my content that was maybe a little more in depth, maybe a little more technical. And then there's also this other button called deeper dive. All right.

So let's see what happens here. So this one, it says content. contacting Gemini. All right. We'll see what the deeper dive version does. Maybe it's just going to make it much longer. All right. We'll see what happens here.

So yeah, live stream audience, thanks for sticking with me. This one to take it a little longer. It's doing a very deeper dive. Oh, wow. Oh,

geez. Okay. This went like super deep just on this one slide. Okay. Um, yeah. This is pretty important. So this slide was on kind of the moving goalposts or how the definition of a small language model is changing.

All right. So the one button made it much shorter. The second button made it much, much longer. So I could go through here, I believe, in text prompt. Let me actually just see if I can do this. I'm going to say instead of summarize slide and deeper dive, I just want to. I just want a chat box where the user can ask questions of Gemini slash the slide itself. All right.

So I think I've done this before. I don't know if this is going to work essentially updating the kind of the AI embedded capabilities. We'll see if it does or not. I actually don't remember. Hey, but this is why we do these things live. But let's see. It says, of course, I can make that change. I've replaced the summarize and deeper dive buttons with a single collapsible chat widget

in the bottom right corner. Now you can use the Ask Gemini button to open a chat window and ask any question you have. The AI's responses will be contextually aware of the slide. That's nuts. Contextually aware of the slide you are currently viewing, allowing for a more natural and interactive Q&A experience. All right. So it is, again, y'all,

I've written zero code. And if you'll see here inside the Canvas mode, it is writing the actual code here live, right? So there's hundreds of lines of codes of code. And then in the preview window, so this is what Gemini, one of the features of Gemini Canvas, which is a very powerful feature, by the way, just one of the features. So it essentially wrote the code for me and now I'm previewing it. So let's just do an example here. Let me go to that same slide two.

So it says the small language model, a constantly moving target. All right. So now I'm going to say, ask Gemini. Let's see if this works. Okay. Cool. All right. So I'm not even going to say, I'm going to say explain this slide in basketball terms.

Keep it simple. All right. So let's see if this works. So literally last year mod, it's it embedded like, a legit like chatbot into this presentation where it looks like I can just chat with Gemini. So let's see if it is contextually aware. So this slide is about how the definition of small language models are changing. So it said, okay, let's break this slide down using simple basketball analogies.

Think of the small language model as a small player on a basketball team. Small keeps changing. Then being a small player used to mean you were really short. like maybe under six feet. Now the definition has changed. We now consider some players, small players, even if they are quite taller, like six foot eight. It's very true, right? It's very true.

Like 30 years ago, you know, if you were a small forward in the NBA, you could be like six foot five. Now if you're a small forward, you're like six, ten. So that's a great explanation just on how even, you know, I'm straight up flabberg. Do you see now when I, like, when I encourage you all, this took no time, you know, 90% of it is me blabbing on, right, trying to do a good job, you know, describing how to do this step by step. But I could have done this in less than 10 minutes, right? Yes, you still need a human in the loop. I would have spent more time on the front end, more time on the back end.

But the majority, 80% of the work got done in less than 10 minutes. And it is extremely impressive. If you would have even told someone like me, someone that covers AI every day for two and a half years, that something like this would be possible, that I could go find my old documents, just hand it over, dump a bunch of information,

that there would be an agentic thinking model that would go, pull information from multiple documents, do additional research, spit out something great that then I could use inside a similar platform and it could build an internet, interactive presentation and embed AI in there.

All right. How is that for putting AI to work on this Wednesday? All right. So I hope I hope this was helpful, y'all. So I think you all said you want to keep doing this thing on Wednesdays. So let me know, option one, option two, option three. What should we do next week? Do you want to do option one, notebook LM, advanced workflows?

If you want that, just type option one. I'm also going to have this in our newsletter. So at Your EverydayAI.com in today's newsletter, you know, make sure you go vote. Option two, co-creating with Gemini Canvas mode, or option three, building useful AI dashboards for non-technical people. All right. So what do you want to do next week?

It's up to you. I work for you. I hope this was helpful. If so, please, if you haven't already, go to Your EverydayAI.com. Sign up for the free daily newsletter. That's where, right, we can learn all we want. you know, watching this live, you know, learning together with our live stream audience, podcast audience, appreciate you guys.

But where you leverage this is the newsletter. That's where I go, I go like write this with my fat fingers, right? And I'm going to say, hey, according to everything we talked about here, here's the most important insights. And here's actually some new additional information that maybe we didn't have time to get to in the show. So we do that every single day on the website, your everyday AI.com. I hope you go put AI to work for you today, this Wednesday.

I can't emphasize enough how much just simple practice, right? Simple practice doing this every single day and it's hard to keep up with everything. But that's what we're going to be doing on putting AI to work Wednesdays. Thank you for tuning in. If this was helpful, please repost this, share this with someone. And I 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.