Exploring Advanced Features of ChatGPT: From Image Generation to Deep Research

In this episode of the Vera Shafiq Podcast, Vera takes a look at the latest advancements and features of ChatGPT as of April 9, 2025, with a focus on the newly released image generation feature. Vera discusses her experiences using ChatGPT for various tasks such as creating brand-consistent marketing assets, conducting deep research, managing long-term projects, and more. She also highlights the benefits of using these AI tools to improve productivity and competitiveness in marketing. Additionally, Vera touches upon the potential of integrating AI features into company technology and emphasizes the importance of transitioning from experimenting with AI to effectively utilizing it for professional purposes.

00:00 Welcome and Introduction

01:36 Exploring ChatGPT's New Image Generation Feature

07:45 Deep Dive into ChatGPT's Deep Research Feature

10:48 Streamlining Workflows with ChatGPT Projects

11:58 Leveraging Custom GPTs for Repetitive Tasks

13:45 The Future of AI: GPT Operator and Scheduled Tasks

16:52 Final Thoughts and Future Directions

Transcript

0:14

Hello everyone and welcome back to the podcast. Welcome to the Vera Shafiq podcast, and we are already in the midst of April, 2025, already in Q2. The year is flying by. And of course one of the things on people's minds as this year progresses is ai, generative ai and all of the advancements that are happening in that arena. And when I started this season's podcast, I promised that I would be talking. About AI tools and really how AI can help to supercharge your efforts as a franchise marketer or a marketer in general. And I've talked about it a little bit, but not every single podcast that I've released has been about ai. Because as we know, there are a lot of other things to talk about, but today I'm gonna focus purely on the experiences that I've had in chat GPT. So today's gonna be all about chat, GPT, the features that it currently offers as of, I'm recording this on Wednesday, April the ninth of 2025, and I wanna talk about. The wonderful things that Chat GPT's really been able to help me with. And, looking forward into what potentially it's gonna be able to offer in the future. But I wanna get started with a feature of chat GPT that was released pretty recently. Probably a couple weeks ago, which is the new image generation feature. So chat, GPT moved away from the DALLE image generation. It still has DALLE as an image generation option, but the 4o image generator is just hand over foot, head over heels, head over, whatever you wanna call it. Way better than the DALLE image generation, in my opinion. You've probably already tried it out and you've probably already realized that the photorealism is amazing. So whereas before DALLE used to give, images that were realistic, but they really didn't look like photographic images. Now the photo realism is there with chat, GPT-4o image. And the great thing about it is, the consistency, the image consistency that we are getting now with this feature. So for example, I'll give you an example of how I've used it recently is I've given it an example image of a brand's marketing creative, right? So I took a Facebook. ad for a specific brand that I'm working with and I told Chat GPT, here you go. Here's an image. Please pay attention to the brand guidelines. And I called out some of the brand guidelines, the colors, the fonts, the logo, the style of the images, and then I asked it to give me some similar assets that had different scenes going on in there, but with the same brand guidelines, and it did a phenomenal job. Of generating alternate versions of the image that I gave it. So you know, what this allows us to do now is to. Be able to give a seed image to the AI and then have it generate images that follow the same style or even use the same person. So say you have a spokesperson or a person that features heavily in your creative, you can ask it to create images of that same person in different poses, different situations, different scenarios. And it does a really good job of it. The other great thing is it follows directions really well in terms of, for example, if you have a uniform, so I'm working with a home services brand, right? They have a specific uniform that their team members wear specific colors, specific styling, specific pieces of apparel that they wear. You tell chat GPT that you want. The character in the image to be wearing that specific uniform and it follows it to a tee. So now you can create advertising assets, marketing assets that have people uniformed in your specific brand apparel. So that is a game changer. The other thing is, if you have an image that has something in it that you need to correct. You can tell GPT-4o hey, take this piece of the image. You can circle it. You can actually select a piece of an image and say, please change this piece of the image, but leave everything else as is. It can correct your colors and your fonts. What I've found is that the logo can sometimes be hit or miss. So if you ask it to replicate. An overlay of a logo on top of an image, it sometimes gets the logo slightly wrong. It's never spot on. So what I recommend is have it generate the image and then overlay the logo yourself in Canva. It's really easy just to upload that image and take your logo and slap it on there in Canva. The other great thing is that it's now really awesome about text overlays. So whereas before it used to mess up with text, it couldn't spell things right, it used to look like Russian or some weird language on there. Now if you ask it to overlay a caption or a slogan or a tagline on the image, it does an excellent job. That is, the really exciting thing for me is while there are other image generation AI tools out there, and I've talked about Ideogram. Ideogram is a really great image generation tool. They have a free option as well as a paid option, and it's really great of photographic images. It gets all the fingers doesn't generate. People with six or eight fingers on their hands, it gets all of that, right? It doesn't have the ability for you to upload a seed image and then get the output, which is. As accurate as what chat GPT's doing right now, in my opinion. Other honorable mentions for image generation in ai, I would say imagen 3 which is Google's model is a great one, so you can get that for free. If you just go on Google, do a search for imagen 3 and you will. Get sent over to that specific model. That's a great one. It also does generate really good photographic realistic images. There are certain things it won't do. It won't create things in likenesses of other people. It won't create images of children or babies. So there were some kind of. Caveats there, but right now I'm having a lot of fun using Chat GPTs 4o model. Granted, when you ask it to generate an image, it does take a little bit of time to do it. It's not instantaneous and it can take up to a few minutes to generate the image. I don't know if it's because I'm using the plus version of chat, GPT maybe. Pro users are getting a better experience on that. But I would say if you haven't tried it yet, definitely try it because it is going to really drive efficiencies for your advertising and marketing, asset creation system and processes. All right, let's move on to another feature of chat GPT that I'm really enjoying. And it's an older one, it's been around a bit longer, but if you haven't tried it again, this is a really great one. It's the deep research feature. So you can go in there, select on the deep research tab, and then ask chat GPT to do any kind of deep research that you want. And what it will do is it will go back. Do a bunch of research. It'll take, 10 to 15 minutes searching the web, doing all kinds of research and reasoning, and coming back with a really solid, substantial, comprehensive output. I'll give you some examples of what I've used deep research for. So I've done it to research marketing agency, so I've asked it to provide me a full report. On the offerings of a couple of different marketing agencies, I've asked it to give me the strengths and weaknesses of those agencies, provide pricing plans dig deep into reviews of those agencies and client feedback, and then provide me comparison tables. So it went away, took about 10 minutes, came back and gave me a really awesome outline of, what these agencies do, pros and cons, cost benefit analysis what I should be aware of, what I should, be looking at in terms of if I'm gonna ask them follow up questions. And the key to deep research is that you wanna give it as detailed of a prompt as you can. So as with all ai, generative ai the prompting is really important. Give it a really detailed prompt and then it will still come back and ask you more qualifying questions. So that's the beauty of deep research is after you prompt it, it will say, okay, just to check. I wanna make sure you want this, and this. And then. Ask some more questions just to make sure it's really understanding what you want out of the research. And then as I said, it goes away, does the research and comes back. I've also done this for deep research on competitive analysis for a brand gone in and asked it to go in and do a really deep dive on not just the national competitors for the brand, but also. The local mom and pop competitors at the local franchise level, and it's done a really great job of digging deep into all of the data that's out there on the web. Something that, you know, if I tried to do that myself, would take me several hours. It goes into the Reddit posts, it goes into forums. It goes to G two crowd, all the B2B kind of platforms, depending on what the search is for, right? Or the research is for, it goes everywhere and then brings back all of the data and it even provides you the links to where it found the data. So you can fact check it and make sure that everything it's come back with is actually legit. All right, so that's deep research. Let's move on to projects. Okay, so chat. GPT has a feature called projects. This has been around for a while. I've used this a lot, really just to streamline my workflows when I know that I'm gonna be working on something on a long-term basis, and I want chat GPT to remember everything about this project so that I can build on it. Over time, I create a project. Then I ask it all the questions and prompts that I need. And the beauty of the project's feature is that you can provide a set of instructions in there to give it a foundation of what you want it to do going forward in that project. So you can give it a background on, say for example you're working with a specific brand, give it the background on the brand, on the project, the brief, if there's a brief involved. And it keeps a memory of everything that's being held in that project so that anything you do in that project is going to be additive and build on the previous history. So that is an awesome feature. Again, if you haven't used it, I highly recommend you give it a shot. The next feature of chat, GPT is custom gpt. These have been around since day one, and again. It's not a new feature, but I definitely wanna call it out that it's something that can really provide value if you are looking to do something again on a repetitive basis. For example, ad creation, right? I have created a custom GPT, for example, on meta ads. So if I want to create a set of meta ads. On an ongoing basis and, for a specific brand, it could be for a specific brand or another, a bunch of brands. But what you can do is you can create this custom GPT. It's similar to a project, but it's slightly different in the sense that you give it a set of instructions, you give it the guidelines that you want it to do, so you give it a very clear outline of how you want it to output the results. And then you can go back and use that custom GPT again and again for different results, right? So I can say, Hey, I have a spring meta campaign that I need to run for X, Y, Z. Here's the target audience, here's the geo targeting that I wanna do. Here's the website that I wanna send traffic to. Please provide images. Copy for meta ads that are conversion, objective, et cetera, et cetera. So you give it the background and it will spit out some really great, ideas for meta ads in general. So custom gpt definitely use that for things that are, things that in your workflow that you do a lot of and require a specific format in the output. Moving on to the next feature. This is a feature I haven't really been able to test because it's only available on the$200 a month pro plan. And I'm too cheap to purchase that. I, this can't justify spending$200 a month right now. I probably should give it a shot for at least a month, but, I haven't tested it out, but I have friends and contacts that have and are using it. And that's GPT operator. So that's the feature that. Essentially goes out and surfs the web just like a human being and can go and do things, fill in forms, make appointments, make book appointments, make phone calls, et cetera, do things on the web just like a human right. It's called operator. It even fills in the CAPTCHA field of online forms. So you can tell it to go out and book, an appointment or go out and. Do some research and fill out some forms based on your specific preferences. Again, this is a feature that I haven't tested, but it's the beginnings of what is happening everywhere right now in generative ai operator like functionalities now coming out of Google. I believe it's coming out of Anthropic as well. So this type of agent really that acts on your behalf and goes on the web is something that's becoming a staple part of generative ai, but definitely exciting and I'm sure it's gonna get better over time. But that's another feature chat, GPT that I'm excited about. And then finally, the last feature that it has been around again for a while. I've been using it. I'm not as stoked about it as some of these other features. I'm not as excited about it, but I'm sure it's also gonna get better. And that's scheduled tasks. So you can go into chat, GPT and ask it to do something on a regular recurring basis. And this is helpful for things like if you wanted to on, give you daily updates on ai, for example. So I have a daily update where it will go out. And look for franchise marketing related AI developments that it's been able to find and come back with the top five developments for that day. So every day at 7:00 AM I get a notification from chat GPT that it's done this schedule task. And gone out and found the latest and greatest developments in marketing AI for franchises. I had to fine tune it a little bit'cause it was going out and finding me things on a global scale and I was more interested in things related to the US and Canada. So I had to fine tune that. It does have a tendency to come back with the same thing day after day, it doesn't necessarily come back with unique output. So I had to tell it, Hey, please give me unique output. It still keeps giving me the same thing sometimes, so it's a little bit finicky, but I can see how it could be really useful for just scheduling things and getting things on an automated kind of calendar or a schedule. So give that a shot if you haven't already. All right, so that's pretty much all I wanted to share today. What I wanted to end this episode with is, this is great, right? All of these features that chat, GPT alone is now offering us. And we might be on the forefront of the leading edge of this'cause, I can guarantee you that we are, if we are using these features, we are probably. Some of the very few people that are I know marketers are gonna be in that top echelon of AI users right now, we might have an edge over our competitors. Because we are using these features right now. But what happens when everybody starts doing it right? This is gonna get democratized and it's gonna become more and more of a staple tool that people use. So what do we need to think ahead for in terms of using AI right now? You just using it is gonna give you that really great competitive edge, get you ahead of the competition. And really create these efficiencies in your workflows, but eventually it's gonna be time to bake these types of AI features into your tech, right? And that's the thing that's gonna be really exciting is once everyone has it and it's no longer a differentiator anymore, where you're really going to have to make those moves is to start putting AI into your technology, into your underlying. Company technology. Start thinking about that as well, especially if you want to get ahead as a brand and as a business. And then I'll leave you with a final thought, and that is there's a time to test and tinker with ai and there's a time to use AI to get your job done better than your competition. So if you are still in that testing and tinkering phase where you're not. Really using it to do any meaningful work or getting anything productive done, then think about taking that leap and moving away from the testing and playing around phase And putting a plan in place so that you can start using AI to get you places, get you some productivity, get some things achieved, and actually put it to work for you in your professional day-to-day life. That's it for today. Thanks for listening. Tune into the next episode and please subscribe to and review this podcast if you liked it. Catch you next time.