Episode Categories:
Resources:
Join the discussion on LinkedIn: Got something to say? Let us know on LinkedIn and network with other AI leaders
Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Connect with Jordan Wilson: LinkedIn Profile
Start Here Series in our Inner Circle Community: Join for free access
Unlocking Business Growth with ChatGPT Images 2: How Non-Creatives Can Now Drive Visual Success
OpenAI’s recent release of ChatGPT Images 2 has shifted the landscape of business creativity and operations. From previous eras of design bottlenecks and creative barriers, businesses now have instant access to highly realistic, on-brand, and text-accurate visuals—without needing specialized creative teams or technical, design-specific prompts. This article distills the tangible, business-driven use cases, capabilities, and process advantages that ChatGPT Images 2 introduces to the enterprise world.
ChatGPT Images 2 Visual Thinking Mode: Intelligent Prompt Interpretation
OpenAI’s Images 2 distinguishes itself with a new “thinking mode” that fundamentally alters how image generation unfolds. This mode does not simply translate prompts into visuals; it plans composition, integrates specific typography requirements, and applies business constraints before generating any pixels. As a result, business users can submit creative briefs—rather than intricate descriptions—and receive outputs that are not just visually compelling, but contextually aligned with broader company goals.
Where previous AI models required users to specify granular details such as focal length or stylistic nuances, Images 2 interprets intent and synthesizes context from chat histories and briefs. This supports the development of cohesive company visuals from the simplest prompts, helping non-designers participate directly in brand development, pitch decks, product demos, and more.
Business Use Cases: Immediate Applications for Non-Creatives
1. Product Packaging Mockups:
Generate ready-to-present product packaging renders complete with legible nutrition facts and accurate brand logos, streamlining the go-to-market process.
2. UGC-Style Ad Creatives:
Produce user-generated content visuals for platforms like TikTok and Meta, including native text overlays that adhere to specific platform guidelines.
3. UI/UX Mockups for Product Development:
Create detailed wireframes and product flows that product managers can use to brief coding teams or AI coding agents with zero design intervention.
4. Multilingual Marketing Assets:
Effortlessly localize high-performing creative for multiple markets—simply upload an existing design and prompt for translation and adjustment, even if original design files are unavailable.
5. YouTube Thumbnails and Social Graphics:
Develop click-worthy thumbnails with accurate, high-contrast text that grabs attention and reduces dependency on graphic designers.
6. Training and SOP Visuals:
Turn dense internal documents into easy-to-understand, scannable job aids to accelerate onboarding, training, and compliance.
7. Personalized Sales Collateral:
Produce industry- and segment-specific sales one-pagers at scale, automatically reflecting unique data and messaging for multiple regions or personas.
8. Boardroom-Ready Strategic Visuals:
Convert meeting transcripts or business plans into executive-appropriate visuals, maps, or strategy overviews with minimal manual intervention.
9. Live Data Infographics:
Integrate real-time competitor or market data directly into infographics and leadership reports using the model’s web-connected grounding feature.
10. Transformation Before/After Imagery:
Showcase outcomes for client proposals with side-by-side visual comparisons, generated from simple prompts and current data references.
Technical Edge: Cohesive, Data-Driven Visual Outputs
Unlike earlier models, ChatGPT Images 2 leverages live web search and real-world data to ground visual outputs in factual accuracy. Aspect ratios are highly flexible, supporting formats from presentation decks (16:9) to social stories (9:16) and more. Up to eight consistent, style-matched images can be rendered from a single prompt, and the model maintains narrative or design continuity across this range.
The text rendering capability now rivals industry leaders—flawlessly producing readable, on-brand, and multi-language text for packaging, slides, UI, and marketing assets. Interactive outputs, such as QR codes and scannable barcodes grounded in real destinations, further expand business applications, ranging from campaign launches to event collateral.
Workflow Transformation: From Brief to Execution in a Single Step
ChatGPT Images 2 collapses traditional creative silos by combining research, copywriting, design, and adaptation into a unified, prompt-driven workflow. A product manager can describe a new feature, prompt for visualizations, and hand these directly to engineering or AI coding tools. Marketing teams can rapidly adapt or update visuals for global campaigns—no longer dependent on cumbersome revision cycles.
Iterative processes are natively supported, allowing business users to refine outputs within a single chat thread. The model interprets feedback, updates data sources, and adjusts design nuances interactively, providing a realistic creative partnership experience without the traditional iteration delays.
Competitive Advantage: Outperforming Peer AI Visual Tools
ChatGPT Images 2 achieved a decisive lead in head-to-head visual quality assessments, winning 93% of blind comparisons in the LM Arena, compared to the typical 60-70% win rates achieved by previous leading models. It performs not only on the text and photorealism fronts, but also produces industry-grade UI/UX mockups and presentation slides with clarity and editorial polish—outshining specialized solutions in both versatility and outcome quality.
Democratizing Creative Execution for Business Growth
By lowering the barrier for visual content creation, ChatGPT Images 2 enables rapid, high-quality production of visual assets across every business function: marketing, product management, sales, training, and executive communication. Non-creative professionals now have a direct path to manifesting strategic and operational visions, accelerating timelines while maintaining quality and consistency.
In sum, this new AI-powered capability allows businesses of any size to translate ideas into visual assets with unprecedented speed, accuracy, and strategic alignment—from simple sales deck concepts to complex cross-platform campaigns—using only natural language briefs and contextual data. The impact will be immediate for organizations seeking to streamline communication, enter new markets, or simply keep pace in a visually-driven, global business environment.
Topics Covered in This Episode:
- ChatGPT Images 2: New Thinking Mode
- Real-World Grounded AI Image Generation
- Cohesive Multi-Image Creation from Prompts
- Near-Perfect Text Rendering in Images 2
- AI-Generated Functional QR and Barcodes
- Business Use Cases for ChatGPT Images 2
- Slide Deck and UI/UX Mockup Generation
- Multilingual and Localization Image Outputs
- Comparing ChatGPT Images 2 vs. Competitors
- AI-Driven Creative Workflow for Non-Creatives
Episode Transcript
Jordan Wilson [00:00:16]:
Maybe you've said this yourself or you've heard it a 100 times in your business. I'm not a creative person. Well, that sentence might have just stopped working. For decades, enterprise leaders have outsourced visual thinking and their business creativity to designers. Designers got the brief, the budget, and then at times the bottleneck, but that whole arrangement might have just collapsed. That's because OpenAI released their chat GBT images too this past week, and it generates legible text on photos, creates photos that look the right amount of real and pixel perfect UI mock ups. It maintains character consistency across eight images from a single prompt and handles multilingual outputs without hesitation. It thinks, use the Internet, and can leverage the history of your chat GBT accounts, and it does it all from a simple prompt.
Jordan Wilson [00:01:13]:
So if you've been telling yourself you don't think visually, I have great news for you. ChatGPT's new images two model does think visually, and with only a few simple prompts, you can expand your company's entire visual world. That's what we're gonna show you today on everyday AI as we put AI to work on Wednesday. So here's the big picture. AI image generation, I think, changed forever. Right? I think we've talked about, you know, AI images kind of having their chat GPT moment with the nano banana, nano banana release. But as we'll see today, chat chat GPT images two is so much better, not just than their last model, GPT image 1.5, but then literally anything else out there, and it's not even close. I think now the biggest difference is chat gbt images too.
Jordan Wilson [00:02:09]:
The generation reasons through business briefs and your chat history and your iterations rendering, before rendering a single pixel. So in the same way that if you use a non thinking large language model in a thinking model, you know the outputs are much better, but that same pattern across text based outputs just compounds across image outputs. And the great thing, this is already live on every single chat g p t plan today. Whether it's free or not, you obviously do get a little more with the paid tier. So you can start using this right now. So So that's what we're gonna be tackling on today's show, our weekly putting AI to work on Wednesdays. Here's what you're gonna learn. You're gonna know what's actually new and why the thinking mode changes everything for creators and noncreators.
Jordan Wilson [00:02:56]:
You're gonna learn why every noncreative business leader now needs a creative mindset and how you can develop that, and 10 specific use cases and pro tips that your team can deploy this week. Alright. Let's get to it. Welcome to Everyday AI. My name is Jordan Wilson, and while we do this every day, I hopefully can serve as your daily guide to understand the nonstop developments in AI across our, daily livestream podcast and free daily newsletter to help you learn and leverage AI to grow your company and career. So it starts here with the unedited, unscripted livestream podcast, but make sure you go to your everydayai.com. Sign up for the free daily newsletter. We're gonna be recapping today's show as well as all of the other AI news you need to know to stay ahead.
Jordan Wilson [00:03:44]:
Right? So have you used images too yet? If not, well, I think this one's gonna blow your mind. Let me just say that. Alright. And this is one of those podcasts, FYI. You might wanna go watch the video version. This is gonna be a very visual show, especially when we go hands on. So in the first half, I'm gonna go over what's new, how how it works, some of the use cases, etcetera. And then on the second half, we're gonna go live.
Jordan Wilson [00:04:12]:
So, yeah, we do, kind of live demos on our AI at work on Wednesdays. Sometimes they turn out well. Sometimes they don't, honestly. Right? So we'll see how it goes, but make sure you can go to youreverydayai.com. Click on the episodes, tab at the top, and then you can go watch today's video if you do wanna see what's happening on screen. Alright. And I did also FYI, I put together a PDF, brief with a bunch of different examples that you can look at as well as the prompts that were used to create that. So if you do want access to that PDF, make sure to repost this show on LinkedIn, and I will send that to you.
Jordan Wilson [00:04:46]:
So, yeah, in the show notes, if you're listening on the podcast, just go ahead and repost that LinkedIn show. Alright. Here's what's new. I think that there's a couple of big shifts. So number one, the biggest new thing in images too, according to OpenAI, is the new thinking mode. So this plans the composition of your image, the typography, and any constraints before rendering anything. Right? And that's huge, especially if you remember the very early days of AI image generation. Right? I remember using the original DALL E with OpenAI.
Jordan Wilson [00:05:17]:
You know, you could ask for a picture of a basketball, and it looked nothing like a basketball. Right now, as we'll see, it's extremely realistic. That's one of the biggest differences. The other thing, well, it's grounded in the real world with real data. That's because images too has live web search, and it grounds those visuals in real time facts and current data. So you can say, you know, create me, which I think is one of the, examples that I have in the, in the PDF if you want that. You know, I said, go look at today's, you know, newsletter, you know, from everyday AI and put it together in a newspaper form. Right? Or, you know, some of the, you know, kind of examples that have gone viral or, you know, find the, you know, the mark marketing industry's latest news for today and, you know, lay it out in a magazine layout.
Jordan Wilson [00:06:02]:
Right? Those things are extremely impressive because it shows off, both the thinking capabilities. It shows off the ability to ground it in live search and then, obviously, the design prowess. You can also generate up to eight cohesive images from a single prompt. Alright. Some of the other, things that really, I think, make this a a step change above anything else. The the the the text rendering is nearly flawless. Right? I think we got that nearly flawless step with Nano Banana, so this is not necessarily new across the industry, but this is new for OpenAI. In their first images model, the text running was not good.
Jordan Wilson [00:06:44]:
In images 1.5, it was pretty good. Right? You might have to clean up a few things. So far using images too, I don't maybe one thing I've had to clean up out of a lot of generations. I mean, we'll see how today's live Tesco. Right? Yeah. Watch. Today's gonna be the one where it just spits out all gibberish even though it's been, you know, pretty flabbed so far. But that that piece is huge.
Jordan Wilson [00:07:05]:
Right? Because even if you remember during the, kind of the earlier mid journey, you know, phase, kinda one of the other popular AI image generators, you know, even some of the later versions of of mid journey from, you know, 2024, 2025 did not have coherent text rendering. Right? I like I said, I think Google really changed that with Nano Banana, some of the other, you you know, Chinese, you know, AI image models did a great job with that as well. But, I do think now it's a completely new, leap that we've seen with images too. Also, there's fun things you can do. You can literally create functional QR codes, bar codes that actually scan of great world destinations. Right? So this isn't just something that creates images. It can create, well, interactive experiences. Know, some of the things that have kinda gone viral online is, you know, you can create a three sixty image.
Jordan Wilson [00:07:54]:
Right? So if you open it in a three sixty image, browser, you know, you can literally scroll around, right, like you are there. You know, and then the aspect ratios are all over the place, which is a good thing. Right? So whether you wanna do 16 by nine, nine by sixteen, three by one, one to three. Right? Whatever you wanna do, there's so many different, aspect, ratios. So here's how it actually works. Well, I think the simplest way to put it, if I had to summarize what this is in three words, I would say visual thought part. Right? It is like having a super smart designer, that you can just give a brief to, and then you can walk away, and they're gonna do all this research, and they're gonna, you know, grab all your data, you know, match everything up. And it is, I think I have felt this a little bit with, you you know, the newer versions of Nano Banana, kind of its ability to understand the context, and to ground in real world search.
Jordan Wilson [00:08:53]:
Like I said, we did get that with Nano Banana, but something about images too. It just feels a little bit better. The agentic harness, it queries the web, synthesizes data, and then it renders the final pixels. And then you can also, well, treat your prompts as just creative briefs and not necessarily literal visual descriptions. And I think that's huge. And if you have used AI image generators in the past, this one might make sense to you on why this is a big deal. If not, let me explain it to you. Right? Again, let's go back to Midjourney because I think Midjourney was one of the most popular, AI image generators, you know, of 2023, 2024, etcetera.
Jordan Wilson [00:09:28]:
But I think early on, especially in the, you know, the v three, v four days, I forgot the exact years on that. Right? But you almost had to speak MidJourney to it. Right? And I think some of the earlier AI image generators were like that as well. You know, you had to use a certain type of language. You know, you had to describe the the focal length of the camera if you really wanted these great, you know, these great photos and say, hey. I want that, you know, soft depth of field with with bokeh and, you know, crisp highlights and, like, you really had to almost have a a visual mindset to get anything visually great out of it. And this is not the same with images too. You can have some of these simplest prompts that maybe have hardly any creativity in them yet get outputs that reflect something that would you think would require a lot of creativity.
Jordan Wilson [00:10:21]:
So right now, here's how it absolutely crushes all other models. Right? We did share this in our newsletter last week, but I think it's worth showing. So, Ella Marina does blind blind tests. Right? So you put a a prompt for an image in, you get two outputs, you vote on which one's better. Nano Banana, two was great. Right? It was a huge jump ahead. Images two from OpenAI absolutely crushed, not just nano banana two, but every other AI image generator, for now. Right? I do have to point out, you know, Google's IO conference is in, like, three weeks.
Jordan Wilson [00:10:57]:
Who knows? Maybe we'll see something new from them. Maybe we won't. I actually don't have any inside information on that yet, so I can I can say that out loud? So right now though, the images too inside LM Arena with those blind taste tests, it set absolute records, and it wasn't even close. Right? So it won 93% of blind comparisons on the LM, on the L M Image Arena. Right? Where normally a leading model might win, like, head to head, it might win, like, you know, 60%, maybe 70%, some of the models. Winning 93%, that means it's it's almost like an undisputed hands down, right, at least according to, you know, all LM arena stats so far in the its ELO score. Right? So after that voting process, you get an ELO score. Arena said that it broke the record for the biggest lead ever.
Jordan Wilson [00:11:52]:
And I think this is important because, yes, there's other great visual tools out there. And I think one, you you you know, you might be thinking, oh, the new claw design. Right? Claw design's great, but images two does the things that claw design does well. Right? It does things like wireframes really well. Claw design does things like, you know, UI, UX mock ups really well. Images two probably does better. Right? And I think a lot of people, you know, and even how can this play the part of, you know, in my creative process. So if you already are a creator across multiple disciplines, right, I think the rest of the show I'm probably gonna more focus on are non creative people.
Jordan Wilson [00:12:30]:
Because if you're a creative person, I think you already see the potential in images too, and you're already crushing it. Right? But for, actual creatives, I think this new outputs from images too serves as a baseline. So using these as, you know, inputs for AI video. Right? Whether you're doing that in v o three, seed dance, whatever. You know? Sora is unfortunately, you know, no more. But starting with, these image outputs for videos, I think, is a a huge level up. And then the same thing. Starting with this for design.
Jordan Wilson [00:13:01]:
Right? I think that's one of the big, things that was actually talked about in the release on, you know, yes, GPT 5.5 and codex, and in general, not great at front end design. So if you're trying to build an app, inside codex or with the g p t 5.5 model, it's well documented, and it's definitely true. It's not that great at design by default, but you can use images too. Right? So if you're using codecs, I've been doing this all the time. I'll just create a render of something first with images too, and then just say, alright. Well, go build this now. Now that you see something that looks good because it has, in its training data, I'm sure has gobbled up every single, you you know, user interface that's ever won an award, you know, all the most popular ones, the trends. So the on the images side, because it has that, you know, kind of that world knowledge and it can think about things, it can create great starting points whether you're bringing something into cloud design, whether you want to code something, you know, in another platform.
Jordan Wilson [00:14:00]:
It's a great starting point even if it's not going to be your end point. Also, the type of create creative work that this unlocks is a ton. So it collapses research copywriting and layout into a single prompt for one operator. Right? You can have multiple steps of an output just with one text input that doesn't have to be overly creative or technically specific. You know, product managers, I think this is great. You can describe new features and, you know, into images too, and then give it to codecs, and it will write that front end instantly. Right? That's something I've been doing. You know, I'll take a a screenshot of an older app I've been working on.
Jordan Wilson [00:14:35]:
I'll say, hey. I want this type of feature. Can you render out what it should look like given this is what the app looks like? Right? I I I think that's a great use case. Also, marketing teams, right, to localize your master creatives globally without, you know, having to rebuild it or, you know, work with translation vendors. You know, you can literally just upload an image that you've already done. Maybe it's a graphic and have it, you know, translate the text and reformat it. Right? Maybe it'll be more characters, fewer characters. So maybe you lost the, you know, the Photoshop file or the design file that has all the different layers.
Jordan Wilson [00:15:07]:
Well, it can take care of that for you. You don't have to go in and, you know, get the eraser and and go to work manually like it's, you know, twenty twenty anymore. Alright. So let's talk about 10 use cases that you can deploy now before we start looking live. So, first product packaging mock ups with perfectly spelled nutrition facts and brand logos. Next, UGC style ad creatives for, you know, TikTok, meta, whatever, with native text overlays. It's actually really good at that. I was kinda surprised.
Jordan Wilson [00:15:37]:
Next, UI UX specs that front end, or sorry, that product managers can hand to coding agents for instant code. So if you are working with, you know, product managers and you're trying to describe a design to them, you don't have to necessarily get a designer involved first. Right? So even if you are not the end creative, you know, I had, Richard from Google on a couple of months ago. One thing he said that still stuck to me to this day is, like, demos over memos. Right? So think of sometimes you might have to describe something to a creative. Right? If you are not the creative on your team, but you're working with someone, it might take a long time, or you might have to bring in a designer just to help you, you know, translate or communicate a message. You know, maybe it's something for, you you know, your your customers, your clients, whatever it may be, but now you could just use images too for that. Next, instant multilingual localization.
Jordan Wilson [00:16:27]:
That's huge. You can do YouTube thumbnails. It actually does a surprisingly good job at that. You know, creating those, you know, eye catching, you know, text contracts, text contrast where, you know, jumps off the page, and it gets every word. Right? Next, you know, you can turn internal training visuals that, you know, turn those dense SOPs into scannable job aids aids. Sales one pagers, another great example with industry specific mock ups. Or, you know, imagine creating, you know, a sales one pager at scale personalized for, you know, 10 different, demographics against, you know, 10 different countries. As an example, if you have one winning creative that has done great for you for customer acquisition or, you know, client success, whatever it may be, imagine being able to localize that and personalize that at scale.
Jordan Wilson [00:17:18]:
You know, boardroom strategy maps that translate written plans, or meeting transcripts into executive ready visual briefings. That's a big one. Alright. Next, live data infographics, you know, pulling current competitor stats into ready to publish leadership reports. And then, you know, before and after marketing shots showing transformation outcomes for client pitches and the proposals. I mean, there's no shortage of use cases, but there's 10, you know, I think regardless of what job you work in, one of those 10, you can literally go try today. Even if you have a free chat GBT plan, right, you're not gonna get all the features of images too. But if you have a basic $20 a month chat GBT chat GBT plan, you can probably use two or three of those examples I just gave and instantly grow your business or at least grow your business opportunities, strengthen your internal or external communication, you know, give your homepage a little pop, whatever it may be.
Jordan Wilson [00:18:13]:
Alright. But this is our AI at work on Wednesdays, So let's get live. Let's go demo style. All right. Hopefully this works. I am. Oh, there we go. All right.
Jordan Wilson [00:18:24]:
So how did I know I was going to run into some issues? All right. So I am. If you, if you haven't seen, okay. So now, now, now we're going to run into some issues here. Let's see. Let's see if we can get this going. All right. I'm actually, if, if, if you are watching the video version of this, I'm I'm in San Francisco, not in my normal, Chicago home office setup.
Jordan Wilson [00:18:48]:
So, did get a new computer as well. So I'm having some issues sharing my screen, but I think I got it there. Let's see. All right. Hopefully we have it live stream audience. Let me know. You see my screen. Hopefully you do.
Jordan Wilson [00:19:00]:
All right. Let's get to work. Right? What could possibly go wrong, doing live, live demos of generative AI? Alright. So I'm going to go ahead and paste in, a little prompt here, and then I'm going to read it, to our audience as it works. So, nothing special here. I'm going into chat g b t. If you haven't seen this, all you're gonna do is hit the plus button in the prompt bar. You're gonna click the create image, then there's other options.
Jordan Wilson [00:19:29]:
So, you know, once you do click that, create image, you're gonna see there's different, options for aspect ratio, auto, square, portrait, story, landscape, widescreen, etcetera. I'm not gonna touch anything in there because I have it designated in the prop. So I'm just pasting this in, and I am hitting enter, and I'll read it. But for our livestream audience, you'll see this kind of this new in animation that I really like, that kind of shows that the model is thinking. Right? And you can it still has the thinking trace that I can go and look and literally read what the model is thinking based on my prompt. So, always do that. It's gonna give you better outputs. I know you're probably if you're a long time listener of the show, you're probably tired of me saying, like, read the chain of thought.
Jordan Wilson [00:20:13]:
But you absolutely have to read the chain of thought always because models always change. Generative AI is generative, not deterministic. So if something goes wrong, looking at the chain of thought will give you clear insight why. Alright. But here's the prompt, and it is working. So it's been about thirty seconds so far. So I said, I need a six slide presentation contact sheet, create a 16 by nine image, sewing six polished slide thumbnails for a fortune five hundred leadership briefing called the AI capability gap. Right? One of our, start here series episodes that we did recently.
Jordan Wilson [00:20:46]:
So this is a real use case. Right? And I can maybe go share this in our community based on, you know, what we put together. So I'm decide I'm just describing the scene backdrop. So I said clean ivory canvas with subtle grid and elegant slide shadows. I said I want six distinct 16 by night slub slide thumbnails arranged three across, two down, each with unique composition. I said, I want this in the style of a premium strategy deck with McKinsey clarity plus magazine editorial punch without copying either. I said, composition, all six slides are readable as a contact sheet, not one giant infographic. And then I gave it a color palette.
Jordan Wilson [00:21:25]:
I said, ivory, charcoal, deep teal signal, red, muted, gold. And then I gave the slide tech for each slide text for each, you know, the gap AI is ready. Most companies are not, etcetera, etcetera. Right? Alright. It's already done. So first version alright. Am I testing the first version of in my testing turned out a little bit better. This isn't bad.
Jordan Wilson [00:21:47]:
Alright. Let me tell you why I like producing it this way. So I just essentially created a slide deck with very minimal information. Right. So it did a pretty good job, stylistically. Right? But the reason I did all six slides on one prompt before I you know, versus doing them one by one, and I I made the prompt this way specifically. So this is a little tip or trick for you is if you are gonna create a slide deck in images too, you would do you would probably wanna do so one slide at a time. But I like telling it to first, you know, render three across, two down, so I can see how all of the slides look together.
Jordan Wilson [00:22:25]:
So this is a very minimal, output here. So what I'm gonna do now is I'm actually gonna open up, I'm opening up in another tab here our episode on this. Right? So I'm gonna copy and paste all of this back in, and I'm gonna say update it with this content. So I'm gonna give it no other, I'm gonna say update it with this content. No other, feedback. Right? So now I wanna see it did a pretty good visually, it looks good. Right? It followed all the directions. It gave me the six slides.
Jordan Wilson [00:23:03]:
Some of the content that I did give it, you know, kind of as placeholder content. It did a great job. Right? It it labeled the six slides as I wanted to. Number one, the gap. Number two, capability. Three, usage for profit. Five, bottleneck, and six, metric. Talking about the AI capability gap that we actually tackled on the start here series episode seven fifty five.
Jordan Wilson [00:23:22]:
So it did a really good job. Visually, it's nice. It's clean. I think it lacks a little bit of information. You know, I mean, it's you know, there's a slide that's just, you know, one bar chart with a lay right? It looks good from a zoomed out version, but I would probably want a little bit more information. Alright. So I did just you can't just iterate, inside the thread of any conversation. So one thing I've noticed personally, about, images two that I think is much better than images one and images 1.5 is its ability to iterate usually does a little bit better.
Jordan Wilson [00:24:02]:
Because previously, iterating within the the the same kind of context window or the same chat thread for whatever reason, it was very hard. And if you had something that you liked originally in the first prompt, and you were trying to refine just a couple of things. Right? Let's just say, oh, you know, I want this shirt to be green instead of red. Right? Then it might add glasses, or it might change, you know, the background completely. So I did I have noticed that iterating, usually does a little bit better. So now in my, follow-up here, all it did I just said update it with this context. It didn't actually do much of anything. And I'm wondering if because when I pasted it in, you know, there's this new feature in Chat GPT that says paste as an attachment.
Jordan Wilson [00:24:43]:
So it didn't really add any of this. I'm gonna try it again. So I'm gonna say update it with this content. I'm gonna paste it in, and I'm gonna make sure it shows as the text field. Alright? And then I'm gonna say, there should be probably at least, I'm gonna say what? I'm gonna say 20 to 50 words per slide. Alright? So we'll see if that works. And as we give that a second to cook, I'm gonna go ahead open a new tab, and we're gonna start a new example. So that was one slide deck example, that I think did a really good job and a tip as well.
Jordan Wilson [00:25:21]:
I like working instead of one slide at a time. I like working in that six slide view or the nine slide view. I haven't really tested it, you know, how far I can do it with one image. But even the fact that it was able to within one single image output, it was able to show a six slide preview, and everything on there is correct. Right? It didn't have the amount of information that I wanted, but everything on there was correct. It followed the design brief fairly well. You know, it created images. Right? So for the number one, it's about the gap.
Jordan Wilson [00:25:54]:
Right? So it not only, you know, created the the the text and the graphics, but it has this nice kind of image of a canyon gap, right, with some dark, background and light text on top. So from, an aesthetics standpoint, this first I mean, the first and the second version are really good. Right? It didn't do as well as I would have liked on the content side, but like I said, maybe that's, I actually didn't check as much of as I should have on how it handles those text as attachments. So, we'll see on the on the second iteration here. Let's see if it's done. Alright. It's still cooking. So, let's go in.
Jordan Wilson [00:26:32]:
Let's do one more, let's do one more live. So for this one, I'm gonna do a UI mock up. Right? I'm not gonna do all these images. Right? I wanted to do something that I thought would be great for the, you know, average business leader. Right? The average business leader, you you may not be going in here to, you know, play around with photos, right, and to create, you know, realistic photos of, you know, a man in his forties walking around downtown. Right? It does very well on that, obviously, but I I really wanted to get into the, bread and butter of some of the business utility, which I think is just, you know, being able to create decks, being able to create wireframes, being able to create mock ups, all of those things. Alright. So our second one is this.
Jordan Wilson [00:27:16]:
It is a UI mock up. So I'm describing this by category. I said the ass the asset type, it's a mobile app plus desktop and dashboard product user interface. And I'd said to create a polished product design mock up showing a retail field operations AI assistant across a phone app and a desktop dashboard. I said I want a clean studio product UI presentation with a subtle retail operations background. For the subject, I said it should be a phone screen for store managers, a desktop dashboard for regional leaders, and small task cards and exception alerts. The style, I said it should be a premium b two b product with UI, practical and believable, like a serious SaaS launch image. Then the composition framing, I said 16 by nine with phone on the left, desktop dashboard on the right, connecting workflow cards in between.
Jordan Wilson [00:28:02]:
So I did you know, I was somewhat specific. I didn't give it the actual data that it needed. Alright? And it's already done. So let's take a look here. In in in terms of prompt adherence, I mean, this is pretty good. Right? I don't necessarily personally like the design, because the the design itself of the, you know, of the user interface, to me, would not be something I would prefer. But I think that's a preference because I think stylistically, it's re really good. This does have a premium, you know, b to b SaaS feel.
Jordan Wilson [00:28:36]:
Right? I this looks like something, I don't know, that I currently pay for in one of my subscriptions. Like, it looks really good, but it has the the store ops assistant mobile app, on the left. It has the pullout of the tasks, kind of displayed, in the middle. You know? You know? And it's showing how all of those things make their way over into the desktop app. So in terms of prompt adherence overall, you know, maybe this played it a little safe, but I'm just gonna give it now just natural language feedback. Right? So I'm gonna say, you know, pretty good, but this feels a little, you know, 2 thousands cheesy. I'm gonna say, give it a sleeker and more modern, aesthetic. And then I'm gonna say, I'm also gonna say, let's test some of the grounding.
Jordan Wilson [00:29:27]:
I'm gonna say, you know, pull, you know, real world data from a, real company or market instead of dummy data. Right? So nothing more I love than doing live demos than typing live when my mic's in the way, and I gotta do my sideways T Rex arms. Alright. Let's go back to our first version. So remember, this was the capabilities, the AI capability gap deck. In the first iteration did pretty good, but there wasn't a lot of text. So now I'm looking at it and I see, okay. This is good now.
Jordan Wilson [00:30:06]:
It added you know, I said there should be about, you know, I think 20 to 50 words per slide, and it looks like we have that. So now what I'm showing on my screen, we have six stacks. It followed the color scheme. It's actually really good. I didn't even tell it to to rotate, you you know, having, like, every other. Right? So, you know, the first slide has more of a darker tone, still following all the guidelines I gave it. The second one is that lighter kind of ivory cream. The third one is dark.
Jordan Wilson [00:30:33]:
So good slide decks do that. Right? If you're saying, you know, a consultant company, etcetera, you know, good ones, they kind of rotate, so it doesn't look the same every single. So even though I didn't give it, that that requirement, it did it on its own. And that's one of the you know, that was one of my points when I started this. I said, even if you are a non creative person, you are going to be able to get creative outputs. I don't think an output like what I have on my screen now would have been possible with GPT one, or, GPT images one, GPT images 1.5, but it definitely is in, two. I do think you could probably get something like this in Nano Banana Pro with a very similar prompt. Right? But, in terms of the output, this is a really good output.
Jordan Wilson [00:31:17]:
Right? And then if I wanted to, I could say something. I could say, great. Let's design, each slot in full one by one. Right? So the the advantage, right, little tips and trick of doing it this way. This is also because I've, you know, been in and out of different types of design for, you know, twenty years just working, you know, small business marketing, etcetera. Right? I know sometimes you wanna be able to see that zoomed out view to see how this looks, as a whole. Right? And it looks really good. So now I can say, great.
Jordan Wilson [00:31:48]:
Let's design each slide in full one by one. Alright. And then it's very likely gonna go through and do that, but we don't need to wait around and check. So let's go to our UI mock up. And one thing I'm gonna pull out here is I told it, I said pretty good, but it feels a little two thousands cheesy. I said give it a sleeker and more modern aesthetic. Pull real world data from a real company or market instead of dummy data. And then if I look on the right hand side, I can see exactly what it's doing.
Jordan Wilson [00:32:19]:
So kind of it's the steps that it's going through. It says clarifying image with real world data, deciding on real market data for UI. It says I could use, a neutral data like The US grocery region view. Right? So it's going through in its chain of thought, like, hey. What kind of data would work well here? It says avoid fabric, avoid fabrication while using real stats. Alright? Clarifying image creation and sourcing data, editing image with updated modern aesthetic. Here we go. Using real data for dashboard.
Jordan Wilson [00:32:50]:
So it said, I found some solid data, but I'll avoid using logos as requested. It says, instead, I can mention specific Walmart stats like the number of US stores, 4,605, and associates, 1,600,000, along with their net sales of 462, 400,000,000. Right? So it pulled real data from Walmart. It didn't use the Walmart logo, but it does have kind of the Walmart vibes. Okay. So now, you know, when I'm looking at this, the design aesthetic did not improve greatly. It did technically take a step, in the better direction. So another thing, if you didn't know when you are looking at these images kind of in full screen mode, then there's thumbnails on the left.
Jordan Wilson [00:33:29]:
So, you know, I'm looking at the first version now, and then the second version. So the second version is definitely a little cleaner. Overall, it's still the a similar aesthetic, but it did make it a little more sleek and a little more modern, and it did pull in actual, you know, data. Right? I'm sure that number of stores is in Walmart has changed since, you know, it seems like a new one opens, you know, every day. But great a great use case and a great example there. Alright. So that is a wrap, but I did FYI y'all. I did put together a nice little guide for you.
Jordan Wilson [00:34:04]:
So let me show everyone quickly what that looks like. So, nothing nothing nothing I love than being on a being on a laptop, trying to do these, you know, things where I'm I'm sharing my screen. Always, always fun. Alright. So I did put together a little guide for everyone, just a PDF. Alright. Well, maybe may alright. You you know, I said I did just get a new laptop, and apparently, I'm not able to yet share that.
Jordan Wilson [00:34:38]:
Alright. Here's here's what I gotta do. Thanks thanks for sticking with me, everyone. Alright. I'm not gonna be able to show you the PDF, but, regardless, yeah, I gotta update my Chrome settings, reset it. I don't know why. Anyways, I put together a simple PDF with, I think, about eight different examples, so you can see exactly what they look like. You can see the prompt that was used.
Jordan Wilson [00:34:58]:
So, if you wanna try them yourself, including, the two that I just did are on there as well. So if you want, that document, I will send it to you. Just make sure to go repost this on LinkedIn. Alright. So what's my takeaway here, as we wrap up today's show? You do not have to be technical or creative to change what your company can do visually. Right? So, obviously, this helps on the marketing side. It helps on the brand messaging side. You can upload, you know, all of your brand guidelines.
Jordan Wilson [00:35:33]:
You can upload, you know, certain photos, certain designs, and iterate off those or have those serve as the base. That's important. Right? But I think this is a big step in the direction of democratizing true creativity for even non creative people. So if you've always had maybe a vision in your head but didn't have the skills, if if if you work in a large organization and, you know, maybe you're running into constant bottlenecks of getting certain creative approved. Right? And maybe or just to really be able to, you know, flash your vision out, you know, on paper, on screen. Maybe it takes too long with your current setup. I think this is a great way that is instantly not only going to save people time. Right? But it's also going to, I think, change what a lot of companies can do.
Jordan Wilson [00:36:20]:
And, yes, you know, this FYI, I'm I'm sure, you know, there's gonna be hundreds of, you you know, new creative, you you know, AI companies that are just using this image, you know, to bring products. So this isn't just with, ChatGPT and their images too. I do think this is gonna you're gonna start to see this kind of everywhere. So it changes what's possible. And if nothing else, I hope you see from today's demo, you just have to start thinking a little differently. Right? It doesn't take a lot. You don't have to be an AI image generation expert. You don't have to be a creative person.
Jordan Wilson [00:36:53]:
You don't have to be a marketer, you know, upload things you like, upload examples you like, talk in natural language, see the output, iterate with it in natural language. And like I said, I think it's gonna unlock some things that you didn't know were possible. Alright. I hope this show was helpful. If you're listening on the podcast, hey. Could you take ten seconds? Make sure you are followed and subscribe to the show. If you could leave a rating, I'd appreciate that as well. And then make sure you go to your everydayai.com.
Jordan Wilson [00:37:17]:
We're gonna be recapping the highlights from today's show as well as all of the other AI net news you need to know to stay ahead. Thanks for tuning in. Hope to see you back tomorrow and every day for more everyday AI. Thanks, y'all.