A (0:00)

Hey, everyone. O3, the new model from OpenAI is out and it's awesome. It's making me a better marketer instantly. And I'm going to show you some ways that can make you better at marketing. I'm going to show you how you can use it as your personal leadership coach, how you can use it to take your favorite business book or business framework and instantly apply it to your business. And then I'm going to show you how you can use it to be your creative director. And actually going to show you how I made a full ad campaign for HubSpot with it. It's going to be awesome. I'm going to do all of that in less than 30 minutes. Let's get to today's show.

B (0:32)

We'll be back to the POD in just a minute. But first I want to tell you about something very exciting happening at HubSpot. It's no secret in business that the faster you can pivot, the more successful you'll be. And with how fast AI is changing everything we do, you need tools that actually deliver for you in record time. Enter HubSpot Spring Spotlight, where we just dropped hundreds of updates that are completely changing the game. We're talking Breeze agents that use AI to do in minutes, while what used to take days, workspaces that bring everything you need into one view and marketing Hub features that use AI to find your perfect audience. What used to take weeks now happens in seconds and that changes everything. This isn't just about moving fast. It's about moving fast in the right direction. Visit HubSpot.comspotlight and transform how your business grows starting today.

A (1:40)

All right, so OpenAI just launched two new models, actually, 03 and 04. I'm going to talk to you all about 03 because it's the model that I've been playing around with and really loving. Why have I been loving it? Cause it's really smart and it's really fast. And that is pretty incredible. I had been using 01 Pro from OpenAI as well as Gemini 2.5 and what I have found using O3 is that it is very, very smart. O3 is available to all OpenAI ChatGPT Pro and Plus users and it is really, really good. And it's really good at not just thinking and reasoning, but finding really interesting connections in kind of uncertain subject matter. And I don't know how to explain that other than just show you some really cool stuff to take through. I've got a couple examples from friend of the show, Dan Schipper. He is awesome. And this was one of the early things about the O3 model that really blew my mind. Dan showed an example of. He basically gave this really kind of far away picture of this keyboard and he asked can you read the title of the song on the notebook? And what O3 did was actually crop and zoom in multiple times to get close enough to say it looks like the title says Moon River. I've never seen any AI model been able to say, hey, I can't see this. So let me crop and zoom. Right up to this point, it's just been like, well, I can't read it. Good luck. And so that is like a very, very big departure in terms of just how it's thinking about interacting with the queries and the prompts that you're giving it. Another really cool example that I thought was awesome from Dan, and this is a really great example for all of us, is if you use an AI note taker. Dan uses Micronola. There's Fireflies, there's fellow. There's a bunch of great note takers out there. What he did is he took all of his transcripts from a week of meetings and put them into O3 and basically asked it to do a comprehensive review and be a leadership coach for it. And what he found is that he really avoided conflict and meetings in really subtle ways. And that was like pretty interesting that you can just take these transcripts and get really advanced leadership coaching. And I feel like in a lot of management case studies you get these really like obvious examples. This is something that clearly it's picked up on like very subtle and fine tune movements. Now the other thing that's interesting, and I think we'll probably do a follow up entire show on this use case and just walk through how to do it is you can use O3 to create really good mini courses. And so he used O3 to create a design course and then he uses the reminders tool in ChatGPT to give him a lesson every day. And he basically has a daily course on design which is really remarkable. And it's combining the brilliance and complex thinking of O3 with the practical need to learn something as well as the reminders tool in ChatGPT. So these are all really cool examples and all things that make O3 really fascinating and showcase its smarts. But I wanted to show you some of the things that I've been doing that I think really, really showcase what it can do. So the first thing I had it do here is I'm reading a book and that Book is called Range, and it's by a gentleman named David Epstein. It's a few years old, and it's all about. He's making an argument that generalists are better than specialists and that we've overrated specialization. One of the things I believe is the AI era, the generalists are going to win. You even had Sam Altman a couple days ago say, I think the AI revolution is going to look a lot more like the Renaissance than the Industrial Revolution. And I thought that was a very interesting thing. The Renaissance was a period of generalists spending time in creative pursuits, creating new and amazing and different things, where the Industrial Revolution was a period of intense specialization. And so what's interesting is Epstein in the book Range, basically has this framework, and it's called the kind versus Wicked problems framework. And he's basically like, specialization is really good for kind problems. Things that have a very limited scope and clear sets of rules and outcomes. And wicked problems are like, very complex, very gray. There's not a defined set of rules. Those rules change. And one of my hypotheses, right, is that AI is really good at kind problems. AI is not as good at wicked problems. Humans are really good at wicked problems. But we've all been here, we've all been reading a book, or we have a CEO that has a book that they really love, right? And you're like, okay, cool, so now what? What do I do with this information? And so what I said is, At HubSpot, we're building software for marketing, sales, and customer service professionals. I basically just asked, I was like, can you write me a deep research prompt that gets a deep understanding of the concept of kind versus wicked problems, uses it to research and chart the core strategies and tactics of marketing, sales, and customer success that are wicked versus kind problems. And so what it does is it went through and it gave me a prompt. One of the things that's interesting about O3, O3 prompts in like, very different markup prompt versus like, just your traditional text output. It is really thinking of prompts as a format of code, which I think is very smart, very good. Gave me this amazing, amazing prompt. And you could use the same prompt, right? Because I have kind versus Wicked problems described by David Epstein. The book Range, you could really replace this first line and a couple of the other lines and use this prompt for almost anything. And so it gave me this incredible prompt. And my instructions were pretty brief and not that good. And I reviewed it and I was like, this is great prompt. Could you just run it for Me. And so it was smart enough to ask a few follow up questions. Hey, Is this for B2B? Is this modern or traditional? Is this global or regional? These are kind of starting to be the standard questions that a lot of these models are asking. And what it ended up giving me is first of all a great interview. So let's say your boss just walks in, it's like, hey, I've been reading this book, it's really good. You should read it, think about how it applies to our company without even reading the book. Quite frankly, you can get a great summary here of what the core theses of the book in this case like entire idea of the book is kind versus wicked learning problems. And then what I thought was really really good is it created diagnostic criteria. I didn't ask it to do this, to outline when something would be kind versus a wicked environment and say rule clarity and stability, pattern, repeatability, feedback, speed and accuracy, effect of experience, like does experience matter? And so it gave itself a very clear criteria. This alone is exceptionally valuable. Then gave us examples like golf and chess. Those are kind environments, wicked environments, natural disasters, people management. And so we got all this and then it's like, all right, now do I apply this to my problem? So at HubSpot our problem is like marketing, sales, customer service and helping other go to market functions be successful. And so what did it do? It gave me an entire table of the kind, hybrid and wicked problems. And first of all, I didn't ask it to do hybrid. It created this other category of parts of the go to market that have some kind and some wicked environment aspects which is pretty interesting. And so for marketing it's like hey, performance marketing, SEO SEM, that's a pretty kind environment. It gives you like why it is wicked or kind. It's rapid, abundant feedback which is really good for a kind environment. Clear rules from the platform, patterns of what works. These are all components of what makes something a kind environment which is really, really cool. And so it walks through literally everything, all the aspects of marketing, sales, customer success. And it can then give me a really detailed summary here of how those things impact organizations, their talent, their hiring. And now what I will do next is I'll take all of this and apply this to HubSpot and our AI tools to understand where we can continue to play, to win and really help companies be successful and lead with AI and where AI needs to be more of a co pilot in those wicked environments. One of the great things in the book range is that he talks about, you know, Even in those like kind of kind environments, like chess, the best outcome is when a human and a machine work together to play chess and they reach heights that neither a machine or a human could get to on their own. So I think this is a really good example of how smooth human smart O3Is. That was a very logical. Like if I had to assign this to a management consultant or a grad student, this feels like that's what I would have gotten from them. And that is really, really remarkable. So what we did here is we took a very basic concept and framework from a book, kind versus wicked environments framework. That's literally it. O3 was able to go and do deep research on that. Then it was able to tell me and make a heuristic of how it would rate different things for kind versus Wicked. And then I was able to give it my domain. But you could give it to any domain. Let's say you were a manufacturer. You could have it do the same thing for your manufacturing business, right? It doesn't matter your e commerce business, what have you, right? You could apply anything to this framework and have O3 really think through it for you. And by the way, I probably would have come up with a similar evaluation criteria, right? But it would have taken me forever to go through and rank and score and think about which aspects of marketing, sales, customer service and success would get put in which of the two buckets. And O3 was able to do this for me in like a couple minutes. And that is a huge, huge time savings. Like imagine working with somebody at your company talking about a framework or a favorite book and be able to come back with a treasure map of this is exactly how this book applies or this framework applies to the business we are running. That is really, really magical. And that's part of the power that you get from O3. And I find it really good at building tables and charts. And you'll notice in my initial instructions, I wanted it to build tables and charts because I think it is really good at clear and structured thinking. I'll give you another example of this. I gave O3 a prompt and a bunch of information about the future of marketing. And I was basically like, hey, you know, we've had inbound marketing for a long time. AI is disrupting marketing. I'm a marketing dork. What do we think marketing is going to be called in the future? What is going to be kind of the next dominant parts? And I put a bunch of context here as to what I think some of the changes in marketing is and it took some of the core themes, emotion, resonance, taste and curation, action and immediacy, scaling one to one interactions. And it came up with a bunch of suggestions. And then what I like is that did a quick heat check map again. I think it's really good at tables, maps, frameworks. If you're doing anything with that, you're definitely going to want to be involved and be using O3. And one of the things I love is it just gave me a short list recommendation. It's like, hey, Resonance marketing, signature marketing, Pulse marketing, none of these I'm a huge fan of, but the logic and everything was very, very interesting. And I was like, hey, everybody's talking about vibe marketing. Do you like vibe marketing better than these? And it's like, hey, vibe marketing is catchy but it carries a soft drink commercial fuzziness that can feel ephemeral or even vague. And then it talks about of its recommendations, why it likes them better than vibe marketing, which I thought was really interesting. And then it gave me like a gut check. Like even the header names in O3 feels like a conversation you would have with another person, right? It feels like if you were doing a big research project or big strategy consulting project, this is the type of conversation you would have with another person or a small team of people on the topic. And instead I get to have that in real time much, much faster. We were able to work through all of this in a few minutes versus like what probably would have taken days or weeks with a full team of people, which is really, really different and really interesting. The last thing I want to show you with O3 is I think one of the kind of master class parts of using AI right now is using different models to accomplish a task. And so what I wanted to do and I wanted to test O3 smarts was would it be great at prompting OpenAI's image generation model, model 4.0. So you can see here, I'm in ChatGPT model 03. I said, hey, using the following creative guide. Cause I had already done a deep research. We talked about it on a previous show. A creative guide of all of HubSpot's publicly available, you know, fonts, colors, everything, using the following creative guide. Please write a prompt for an LLM that can generate images as if you were the best creative director in the world working for HubSpot. The output should be a summary of a creative brief. Then that creative brief should be used to generate an image of a campaign mood board. Then a brief of the mood board should be used for Billboard digital ads. This campaign is going to be targeted to growth professionals and marketing sales customer service. The campaign is going to focus about HubSpot's customer agent. So please have the prompt research customer agent and then so I put in all of the creative brief and all of that information. Again, pretty big context window of our creative style guide. And then again in its markup, more technical writing. Basically wrote an amazing, amazing prompt and did a good job saying hey, first you're going to need to research, then you're going to write a concise creative brief. Generate a campaign mood board image, billboard concept, a square digital ad gave the dimensions and then here are the output formats. And so I said great. I copied that and I moved over to model 4.0. So I pasted that prompt from O3 over into ChatGPT 4.0 which is the model from OpenAI that can generate images for you. And as you can see it provides it with all of the HubSpot information. And 4.0then generates the creative brief which is really good by the way. Single minded proposition. Empower your team with 24.7ai support that delights customers and drives efficiency. Key benefits seamless AI driven customer support that operates around the clock. Mood board, one sheet billboard digital ad and so goes through makes all of this. It's really good, builds the copy and everything. And I said great, please run these prompts to generate images. I was like I just want to see some stuff. It gave me the best image results for creative that I have gotten so far. Is it perfect? No. But this is a really good billboard. It gets the core value prop support smarter, not harder. It has the company strap line, it has the call to action URL and it has illustrations and colors and everything within the HubSpot style. Very good. And one of the things I've noticed with 4.0 and the image generation is that if you ask it to generate multiple things, it often doesn't and you have to go back and prompt it. So I was like, hey, can you do the mood board? And so it gets me the primary colors type of photography gives me a solid mood board. The Queens Lex and Deca, the fonts we use, which is really good. Now I can say can you generate the digital ad which it also hadn't generated. So you see amazing prompt, really good adherence to the style guide instructions. But 4.0 which is good at generating images, you know it's not OpenAI's smartest model and image generation is very expensive. So I think it is a little metered in that One of the things you can't do is just go create like a whole host of images from one prompt. It's kind of a one image at a time situation. But I still like to go through and put them all in the prompts because that way I can just type a quick sentence just like that and call back like, hey, I gave you all the instructions for this digital ad. Now I want you to go and generate that for me. And so what I suspect we're going to see here is a version of the digital ad that is really good and able to be in line with everything else. All right, and we're here. And you know what? AI supports real results. Resolve 50% of tickets automatically. That's pretty good. I'm not gonna lie. Like, that's a good and solid ad. It's in line with the creative. It made smart decisions. Orange, the background tends to work really well because it's high contrast on a lot of ad units compared to the rest of the ui. So it's all very, very good. And what you're gonna see here is that if you're like me, I think O3 is quickly gonna become one of your favorite prompt writers and kind of high skilled peer thought partners. Those are the two things that I think O3 really, really excels at. And I find it to be much smoother, much more humanistic than a lot of the other reasoning models today. I think part of that's the speed, but I think part of it's also just the way it frames ideas and makes them feel much more straightforward and in human language versus kind of like robotic. And so those are three really cool use cases that you could all go do in O3 if you want those prompts. We have an amazing prompt library where I've taken some of those prompts and put them in. You can scan the QR code, you can click the link in the description below and get those prompts and play with O3 yourself. Please drop a comment with your review of O3 and again, always hit like and subscribe. We'll see you real soon on marketing against the grain.