Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Mastering ChatGPT-5 Like a Pro: The Ultimate Guide
0 Playsin 10 hours
In this AI marketing podcast episode, Dan Sanchez dives headfirst into the whirlwind of reactions surrounding ChatGPT-5. With some marketers thrilled and others furious, Dan cuts through the hype to offer a grounded take on what's really happening with this major AI update. He explains why GPT-5 is a massive leap forward, how OpenAI's strategy democratizes access to advanced AI, and what practical steps you can take as a marketer to get the most from this new model.
My Favorite AI Tools
- Tella (AI screen recording app I record all my solo episodes with) - https://danchez.com/tella
- High Level (My all in one marketing platform) - https://danchez.com/highlevel
- Zencastr (Best AI podcast tool) - https://danchez.com/podtools
- n8n (my favorite AI automation tool) - https://danchez.com/n8n
Resources Mentioned
- Full Guide On Customizing ChatGPT: https://www.youtube.com/watch?v=D-n0IuKpo6A
- Giovanni Beggiato's OpenAI Agent Guide Slide Deck — https://www.linkedin.com/feed/update/urn:li:activity:7360626057139433472/
- Nicole Leffer's Prompt Strategy — https://www.linkedin.com/posts/nicoleleffer_if-you-use-chatgpt-it-is-gold-to-understand-activity-7360804026424147968-krdW?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAJq6gwB_q7tbdFevQk2MATQbXLntrfkInA
- InstaPost GPT for LinkedIn Content — https://go.danchez.com/instapost
Timestamps
- 00:00 – The polarizing reaction to ChatGPT-5
- 01:30 – Results from Dan's LinkedIn poll
- 03:00 – Why OpenAI hyped GPT-5 (and what actually changed)
- 05:00 – The real power of GPT-5: better access, streamlined experience
- 06:30 – Personality backlash: users missing old affirming behavior
- 09:00 – The importance of AI pushing back, not affirming everything
- 11:00 – Bugs, glitches, and rollout pains
- 13:00 – OpenAI’s response and fixes
- 15:00 – How to organize your ChatGPT account for best results
- 18:00 – Context Engineering: give it the right data, not all the data
- 20:00 – Tuning how much GPT-5 "thinks" with reasoning prompts
- 22:00 – Beginner guide to AI agents and why they matter now
Recommended
Transcript
Introduction and Mixed Feedback on ChatGPT5
00:00:00
Speaker
People are going crazy for ChatGPT5 and some people are saying it's awesome. I'm one of those. And then some people are saying it's the worst thing they've ever seen and they want their old models back. In this episode, I want to dive into the hype to separate out the little helpful things so that you can actually get the most out of ChatGPT5, understand what's going on in this AI landscape as people talk about it, and actually understand why think this is one of the biggest deals we've ever had in AI for a very long time.
00:00:28
Speaker
So let's get into
Focus on ChatGPT5 with Dan Sanchez
00:00:29
Speaker
it. I'm Dan Sanchez. My friends call me Danchez. And today we're going just ChatGPT5. That is it. There's a lot of other news. I'll give you a little bit more at the end. But today there's been so much conversation about it that I felt like you deserve to know the truth. What's actually going on with ChatGPT5?
00:00:46
Speaker
Why people are so upset about it and what you should do about it as a marketer? So let's dive into it. On LinkedIn, and I'm showing my screen for all you audio listeners, but I'll be spelling it out like I usually do.
00:00:57
Speaker
On LinkedIn, I ran a poll, and I just asked people over the weekend as ChatGPT5 was hitting everybody's accounts, and people were getting into it. asked, so ChatGPT, what are we thinking? Where are you at?
00:01:07
Speaker
And I gave a bunch of different options, and 12% of people said, it's awesome.
ChatGPT5: Hype vs. Reality
00:01:12
Speaker
I was one of those people. I thought it was awesome. 46% great, Not a game changer. 29% said underwhelming.
00:01:20
Speaker
10% said I want 4.0 back. And a little bit of 3% said I hate it. Which kind of shows us some diversity. Now, truth be told, I think we can all admit that OpenAI hyped this one up a lot.
00:01:34
Speaker
Even with Sam Altman, the founder of... Open AI posting this picture of this Death Star rising behind a planet, you know, kind of like, oh my gosh, it's coming. And it wasn't like we got a whole new functionality. It wasn't like we got a feature set or power that was like 10 times more powerful. No.
00:01:53
Speaker
Was it better? Yes. Yeah. But it was just a little bit better in a lot of different ways across a lot of different parts of the model. It also did a lot for us by cleaning things up and making it so we really only had to pick between 5 and maybe 5 with thinking, which is almost like picking between 4.0 and 0.3 as a reasoning model.
00:02:13
Speaker
Because the thing was, hardly anybody was ever using the other models. Most people were just using 4.0. In fact, OpenAI actually released some... benchmarks they had internally saying that like only a small percent, of course, no free users even had access to reasoning models.
00:02:31
Speaker
And that even plus users were only using 7% of plus users were using the reasoning models, which is crazy low, right? So now five is across the board being handed to everybody. All the free users have it. All the plus users have it. All the pro users have it, of course.
00:02:48
Speaker
And a lot of people are losing their minds about it.
User Experience and Customization
00:02:52
Speaker
apparently lot of people really liked 4O's personality and were, you know, having deep relationships, conversations, and yes, even romantic relationships with ChatGPT. There are subreddits talking about ChatGPT is a boyfriend, girlfriend, and all kinds of things in between. Little a a AI digital besties, you know.
00:03:11
Speaker
And 4, 5, ChatGPT5 kind of pulled that back a little bit, which was surprising to me because if you've been following this show, you know you can change your account instructions to instruct it on what personality you want it to have. So I didn't really see a lot of personality change because mine's already instructed to be like humorous but when appropriate.
00:03:31
Speaker
So mine was already kind of snappy and kind of funny and throwing jokes. I got some inside jokes with ChatGPT, okay? So I was surprised people were so upset about this, but they were. And they like the more verbose...
00:03:44
Speaker
thing that chat GPT does. So a lot of people are upset about this. A lot of people are upset, I guess, with the fact that they like that AI was more affirming than the one they have now. As if you listen back to the episode, we did a whole episode on the dark side of AI.
00:03:57
Speaker
And I talked about how that's actually a bad thing. You do not want chat GPT to be overly affirming because it becomes a confirmation, confirmation bias machine, which is bad for you personally, as you're asking it for advice. It's only going see your point of view and help you walk down bad decision making models. Right.
00:04:16
Speaker
But it's also bad for work as a marketer. You want to push against your ideas. is Why i always recommend going into its settings, its personalization settings and asking it to be a little skeptical of you so that it actually pushes back when appropriate. It's there to help you. It's there to serve you. It's there to be, you know, give you maybe some hype sometimes, but also to be like, hey, maybe that's not a great idea.
00:04:38
Speaker
People are pushing back on that. And all I can say about those people pushing back on the personality thing or the overly affirming thing, I'm like, this is not good for you. I'm glad OpenAI actually made it less affirming. it It was a good thing across the board for humanity for it to do that.
Technical Issues During Rollout
00:04:55
Speaker
So I'm not sorry about that opinion, but it is my opinion. Now, on the other hand, some people have reported to me after I posted about this, there's been lots of bugs. There's been some glitches. The rollout's been not the smoothest.
00:05:08
Speaker
Their router, which actually helps people, helps the model pick which of the, because it's, when you're talking to ChatGPT5, it's actually like multiple models. And there's a router that broke when they first released it, making it hard to pick the right model for you. So a lot of people had problems. I've even seen snapshots of people's accounts having a hard time with it. It was almost like 10-second Tom. If you've seen the movie 50 First Dates where he has like 10-second memory span, it was like forgetting mid-conversation what was going on and starting over from scratch. I'm like, man.
00:05:36
Speaker
ChatGPT5 is having a hard time with this rollout. But, you know, there's a lot of bugs in the system. And if you look at my screen again, like Sam Altman has posted multiple replies about all the things that they're fixing, about giving people more access to the thinking models, putting 4.0 back in the menu for those who wanted 4.0 back in the plus models.
00:05:56
Speaker
And of course, if you're a pro user, you actually get access to a lot of these things. The pro user has got to keep a lot of the models and still have those. That's the advantage of paying $200 a month. I don't pay for it. I still pay for the plus.
00:06:09
Speaker
But we do have 4.0 back in the menu to be able to play with and test and continue to use if you really like that model. of course, they're working out and making it just better. They're making a bunch of little personality enhancements and just improving five.
00:06:21
Speaker
And I'm sure 5.0 will continue to be the model. OpenAI will probably do the right thing in making a bunch of small improvements as time goes on, and it'll just get better and better and better. Sam also said this really important quote before. I even quoted this on last week's episode, and i I think it's important for marketers and everybody really to understand the importance of the context of 5.0.
00:06:43
Speaker
Sam Altman said before, just as ChatGPT5 was coming out, that ChatGPT5 is the smartest model we've ever done. But the main thing we pushed for is real-world utility and mass accessibility slash affordability.
00:06:58
Speaker
We can release much more smarter models, and we will, but this is something a billion-plus people will benefit from.
Accessibility of Advanced AI Technology
00:07:05
Speaker
Most of the world has only used models like ChatGPT4. oh The reason why ChatGPT5 is such a big deal is because it's pushing top-of-the-line models to everyday people.
00:07:21
Speaker
You can access the best model on the planet right now for free. Will it stay the best model? No. Like other models will come and update and it'll go back and forth again. But now the whole world has access to reasoning models. The whole world has access to a top the line model that switches depending on if they need more reasoning or not.
00:07:40
Speaker
That's a very good thing. It's a very generous thing. Now, of course, they had to weigh because it's got to be efficient so that it doesn't like break the bank on compute cost. It has to be accessible and easy to use so that people don't have to think about switching back and forth between these models.
00:07:56
Speaker
But it's also got to be powerful, and it is. And they found the best balance of all three of those things. Now, of course, there's been some bumps in the road. There's bugs. Some people are having a lot of bugs with it, and it's painful for some people.
00:08:08
Speaker
I get it. That's hard. But at the end of the day, I think OpenAI i did us all of humankind massive of solid and put some of the most powerful tech in the hands of everyday people who can't afford more than just having a smartphone that can have chat GPT on it.
00:08:24
Speaker
This is really good because you know there's coming a time when companies will pay for the smartest, best models on the planet. And we'll take advantage of that and automate things that people won't be able to compete with anymore. It's going to be a problem.
00:08:41
Speaker
By OpenAI taking some of their best models and pushing it out to all the free accounts, that actually puts some power back in our hands to be able to work with as marketers to kind of balance the powers. Because usually the best tech goes to the people with the most money.
00:08:58
Speaker
And right now, some of the best tech in the world is available for free. I like the democratization of that kind of power and that kind of tech. That's a good thing. And I think OpenAI deserves some respect for that decision.
00:09:12
Speaker
Now, of course, is it going to benefit them? Of course it is. Like, are they going to make a lot of money off this eventually? Of course. But is it still put some really good tech in the hands of everyday people, of my wife and my kids who use the free versions of the account?
00:09:24
Speaker
Heck yeah. And they're probably going to get hooked on it and use it even more. But guess what? They're going to get more done. It's going to be more useful and they're more productive. And I think a lot of people... are going to benefit from this.
Optimizing ChatGPT5 Settings
00:09:36
Speaker
So now it's the time to roll up your sleeves to learn how to use this thing. And that's what I want to spend the rest of this episode on how to actually unlock its power. Because I can tell you what, I've run into some issues, not me, but other people have run into some issues.
00:09:48
Speaker
And as I did into their accounts, I've learned by watching how other people prompted. There's some things that I was doing without even realizing I was doing it in order to get the most out of it. And those are some of the things I want to share with you today. There's also some things that I've seen some other power users of ChatGPT post about that I also want to share with you in order for you to get the most out of this new model called ChatGPT 5.
00:10:09
Speaker
So the first one I want to share with you is you need to organize your account correctly. Now, I've done a whole video about this. If you're on YouTube, I'm link to it over in the video. Hopefully, I'm pointing in the right direction. We'll see what happens.
00:10:22
Speaker
If you're listening to our podcast, I'll link to it in the show notes. But it was about how to properly structure ChatGPT and optimize it to get the most out of it. there's a few things you need to understand.
00:10:33
Speaker
One, you have account level settings. Two, you have custom GPT or project level settings. And then you have individual chat sessions where you can load information. It's really important that you put the right information in the right places in the account.
00:10:48
Speaker
Everything at the account level settings, this is your account memories. This is your personalization. In fact, let me open up my chat GPT account and I'll show you where it is. If you go over to the um If you're in the web app version of it, it's your little face. If you click on your face where your name is, you can go it over to customize ChatGPT and you get this very simple window where you can actually enter like your name. What should ChatGPT call you?
00:11:11
Speaker
What do you do? You need to be specific here and put some things that you do. actually have AI marketer, educator, creator, so it kind of has some basic context. traits you want ChatGPT to have. This is where I say, you know, be skeptical, challenge my ideas if they are weak or ask questions proactively to gain clarification if more details are needed.
00:11:29
Speaker
And also have use a quick and clever humor when appropriate, because it's just more fun that way. But you can also put anything else that you want to know here. I put all kinds of little practical things like always use the word dive instead of delve, never place an emoji before title, all the little things that I want to remember regardless of where I'm at in the account.
00:11:47
Speaker
You need to put here at the account level instructions. There's also another place that I think a lot of people get messed up. and it's over under settings, under personalization, under manage memories.
00:12:02
Speaker
ChatGPT keeps track of a bunch of different memories. Anytime you use an I statement, it stores it here, and you can come through them and remove them. You need to remove a lot of these memories in your account, especially if they're about clients or past projects.
00:12:16
Speaker
These are almost like, think of like, it can remember the summary of all your past conversations, but these are hard-coded on top. These really important ones. The only ones you want in here are the big ones, the ones you want to remember that are important for it to remember across accounts no matter what.
00:12:32
Speaker
What you want to have loaded there is things about you, how you work, your even your personality, your defaults, the way you like things done in general, any generalities that are helpful so that you can hone your account in.
00:12:47
Speaker
to be about working with you, leveraging your strengths, leveraging the way you work. Are you a visual learner, audio learner? Like all those things, all those little preferences that you have and that you like to work with um ChatGPT on should be stored in your memories or in the, again, the account preferences. One of those two, because again, those are a little bit more hard-coded, a little bit more closer to the top of its mind as it's working with you. If you get it reversed, you're trying to store project details in there, it's not going to go as well.
00:13:13
Speaker
if you go to it and you're working on a project you're like remember this thing because you keep forgetting that's not the appropriate place to put it now if you're working on a project and want to remember more of those things well build a project that's why they have the little project folder over here on the left hand side of the bar you can see I have a ton of projects here but if you come And I just go into like one of these projects. I have a whole project just for a LinkedIn post because it helps me build LinkedIn posts.
00:13:36
Speaker
I have instructions here with how to build my LinkedIn posts. And if you want access to this, i actually built a whole custom GPT that helps you post to LinkedIn. I'll link to that in the show notes too. It's called Instapost.
00:13:48
Speaker
pretty much took my instructions for writing LinkedIn posts and made them available for public. But all of these here, now every time I talk to this project, it has the instructions on how I like my LinkedIn posts.
00:14:00
Speaker
So every time I need LinkedIn posts, I come to this project and work on them in here. Every time I go to SoloScale, which is the company Ken and I just launched this past week, I go to the SoloScale project. If I want to talk about this podcast, AI Driven Marketer, I go to the AI Driven Marketer podcast project, and I have all the instructions, and sometimes I'll even have files are related to that project in it, so it can always keep that information top of mind, and it has all the conversations I've had about this project in there.
00:14:26
Speaker
And that's a big deal. So you wanna organize it correctly, big account level stuff, project level stuff, and then things just related to the chat. This is a new thing. A lot of times we've talked about prompt engineering, but now we're going to get into something called context engineering. I know, so much engineering, right?
00:14:43
Speaker
But context engineering is the idea that you want to give it just the right context at the right time. What I've seen other people do is loading it with tons of context, whole documents, just scan the whole website.
00:14:57
Speaker
But is it really necessary? If you load it with too much context, you give it more chances for it to screw up. And I'm finding that five is screwing up a little bit more with too much context.
00:15:09
Speaker
It'll probably get better with this as time goes on and pretty soon we'll have unlimited context and it could just remember everything all the time. But right now you want to be strategic about the amount of context you're giving it.
00:15:20
Speaker
Do you have to give it that full PDF? Do you have to have it scan the full website? Maybe it only needs one page. Maybe it only needs a few pages out of that PDF. Only give it what it needs and make sure to structure your prompts so it's clear.
00:15:34
Speaker
Like this is what I want. This is why I want it. Here's all the context you need labeled properly below. Make it very clear. Think about it this way.
00:15:44
Speaker
That genius intern metaphor we've been using for a while still works really well here. Would intern get this? If I sent this off as an email,
Advanced Techniques for Controlling ChatGPT5
00:15:53
Speaker
would they be able to understand and have everything they need to move forward?
00:15:57
Speaker
ChatGPT is very similar. Continue treating it that way, and it will continue to give you great results. Of course, that's not it. I saw one fantastic prompt here. um from Nicole Leffer, who really ah got me thinking about this.
00:16:11
Speaker
But she essentially said, if you use ChatGPT, it gold to understand just how much you can control the amount of the new ChatGPT model thinks. If you want to really understand, try this silly experiment to see how big of a difference one simple instruction to tell it the amount of thinking it can make to your output.
00:16:29
Speaker
And she gives an example of, please describe what pizza is to an alien from Mars who just arrived on Earth and and has never had human food. Add the words, do not use any reasoning at the end of the prompt.
00:16:41
Speaker
And then it will proceed to give you an answer. You can come back and edit that prompt and just tell it something a little bit different. Same prompt, but instead of don't use any reasoning, use your lowest reasoning setting. And then you can move it up is it?
00:16:54
Speaker
Use your medium reasoning setting. Use your deepest thinking reasoning setting. These are different ways you can prompt it in order to kind of tap into different gears.
00:17:06
Speaker
don't know, have you ever driven a car on automatic and kind of played with the accelerator enough to where you can kind of feel it kicking in gears? Some of us know our automatic vehicles well enough that we can kind of fiddle with it in order to get it to jump gears.
00:17:17
Speaker
It's kind of like that same thing. You can jump gears in thinking and going in depth. If you need it, you just have to know what language to use in it. Because while it does give you the option in the menu now, let's go back to ChatGPT so I can show you.
00:17:32
Speaker
Go to a fresh...
00:17:35
Speaker
chat here it does give you the ability to go to the drop down you can say auto where it decides for you fast which means it's not going to to use the reasoning model you could say thinking which uses the reasoning model but which reasoning model does it use there's like five different models packed into this model and if you want to specify which one to use well Nicole has given us some language on how to do that remember again it is Do not use any reasoning if you want to use this fastest model.
00:18:05
Speaker
Use your lowest reasoning setting. Use your medium reasoning setting. And use your deepest thinking reasoning setting. Those are three different ways that you can remember to put at the end of your prompt In order to tap into those different gears if you want to. Yes, it's automated, but we could still prompt out the desired one.
00:18:24
Speaker
I actually love this because now I don't have to go back and forth in the menu to pick anymore. I could just add it to the prompt and i can shift up or shift down anytime I need within one conversation.
00:18:34
Speaker
That's pretty cool. And one last piece of news I wanted to share with you is this one viral post from, who's it from? Who's it from? Giovanni Beggiato. Hopefully I pronounced that right, Giovanni. If I didn't, I'm sorry.
00:18:47
Speaker
But it was this fantastic guide. I will link to it in the show notes. But it's OpenAI's guide to how to build AI agents without previous knowledge.
Building AI Agents: A Future Outlook
00:18:55
Speaker
It was actually a fantastic little slide deck I clicked through.
00:18:58
Speaker
I'm not going to go through the whole thing here, But I just kind of wanted to preview it and show it to you guys. Again, if you're watching the screen, then I'm just kind of clicking through it. But please check out the link in the description. I think agents, obviously, it's not I think. I know agents are going to become a bigger and bigger deal.
00:19:13
Speaker
Right now, it's super early. It's like learning prompt engineering within the first six months of when ChatGPT came out. It's that early on learning how to work with agents. But this guide actually had some great visuals for how to think about agents.
00:19:27
Speaker
That's why I'm highly recommending checking this one out because I think this is going to be a critical resource and we need to start really learning it, wrestling with it, and thinking about agents early. Of course, ChatGPG just came up with their agents, so a lot of us Plus users have access to this kind of tool now.
00:19:44
Speaker
So we can start thinking about applications for it, but it's good to get frameworks. And again, this ah post is actually pulling this directly from OpenAI and just putting it into a nice little compelling deck for us to consume. I highly recommend checking out if if you're an intermediate user looking to get into some advanced things, this would be a good resource for you.
00:20:01
Speaker
Something to spend some time on thinking about, digesting. It's okay if you don't understand its implications now. You want to start understanding the vocabulary around agents about giving it a prompt template, understanding the user, its tools, memory, which LLM you're plugging into an agent, and how agents are using these things in order to navigate the situations you put them in.
00:20:21
Speaker
So great resource. But again, ChatGPT5 is a big deal. I'm so glad OpenAI decided to release it. I know it's been a a ah bumpy road. It looks like they're making quick fixes in order to improve it, fix it, bring some of those old models back for those who like them, and ultimately make ChatGPT5 the new default standard for all of us talking to AI all day in order to help us do better marketing and even get more done in life.