Transcription
Hey everyone, nobody's talking about Gemini 2.5, the brand new model from Google, because everybody's over building anime images in ChatGPT. I'm here to tell you it has blown my mind with how smart and how good it is, and I'm going to show you how to use it for your business today. I'm going to actually show you my behind-the-scenes professional workflow that I use to do amazing research and deep research, then craft an incredible prompt to build a marketing strategy, and then put that in Gemini 2.5. It's going to craft not only the best marketing strategy I've ever seen from an AI model, but it's going to actually make the different components of that strategy into actual assets that I could go and put in a tool like HubSpot and get out to my customers immediately. It's taking work that would have normally taken weeks and condensing it into a day or two. You're going to want to wait to the very end to see how smart it really is. Let's get right into it.
[Music]
Hey everyone, I've got a big show for you today. We're talking about Gemini 2.5; it's the smartest model out there. It has gotten way overshadowed by OpenAI's image generation release, but I'm going to break down for you exactly how smart it is. I'm going to show you how to build the best marketing strategy possible with it, for free, and I'm going to show you a really good workflow of how to actually do that. So first, let's talk all about Google Gemini 2.5. It is a very, very important release that no one is talking about. Why is nobody talking about it? Well, nobody's talking about it because OpenAI image generation basically blew up the internet, and it's all everybody was talking about. But what Gemini is doing is adding far better and more advanced reasoning capabilities, and especially if you are writing code to build software, people are loving it. People were begging for Gemini 2.5 to be in Cursor and the tools that they use to code. And you can kind of see here that from some of the core kind of benchmarks that happen and and that are used to kind of grade these models, Gemini 2.5 Pro is way better than a lot of the other models that we all think about and talk about on a regular basis, and it's got really advanced reasoning and really advanced coding.
Now, look, if you're watching this show, you're probably not a hardcore coder, because I'm not a hardcore coder. We're here to talk about marketing; we're talking about growing your business. And I wanted to test how good Gemini's reasoning really was, and for non-coding reasoning, a great task that I think is building a strategy: Can I give you some information, and can you build me a deep and thoughtful strategy? And if you're a fan of the show, and if you you've already subscribed—if you haven't, hit subscribe—you've seen me do this throughout the course of the last year or two, where we are continuing to test these models to see how good they are at generating strategies. And so I want to give you a little bit of tips and tricks here right now, and I want to show you how to build a really great prompt for Gemini.
Hey, I I hope you all are enjoying the show. I I've been pushing the team to go all-in on AI, looking for ways to work smarter, move faster, and drive better results. One of the first big wins that proved just how AI can transform our workflows was when we boosted our email conversion rates by 82%, and yeah, AI made that happen. We took our standard nurturing emails and turned them into an AI-driven conversion machine. We tested a ton of ideas—some flopped, some worked out okay—but one AI-powered tactic completely changed the game. If you want the exact playbook we used to make it happen, you can grab it right here; scan the code or click the link in the description below. Now let's get back to today's show.
I just happen to have my ChatGPT window open, so I used it to write the prompt. You can use Gemini; you can use whatever you want to write the prompt, but I use ChatGPT 03 Mini because that's what I had. And I asked it to write a prompt for a marketing strategy for HubSpot's Customer Agent. It's our brand new AI agent that is amazing at handling customer support and queries. It's a great product to flex a bunch of support requests from human-based sales reps; people are loving it. And so what I had is I had to do a deep research project on Customer Agent because I was being lazy. I know I have all this collateral about Customer Agent; I just wanted to see what it would go and find and package itself. And quite frankly, you could do this for any of your products of your own, or even a competitor's product, or a just a product you're interested in; have it do a deep research. Then I asked it to build a prompt to use an LLM, so it did a bunch of thinking. You can see the thinking it did over on the side; does a lot of thinking. It took like 10 minutes or so; it came back with a really in-depth overview of our Customer Agent, broke down what its key features are: on-brand, natural conversation, seamless handoffs to human, integration with CRM for personalization, uh, typical ways you would use it, high customer satisfaction or retention—all these things people want, right—and talks about its integration with our other tools at HubSpot, a very, very in-depth piece of research, right? Then I said, "Great, take all of this, and because it forgot to do the marketing campaign," I said, "I wanted a marketing campaign for email, advertising my website, those things." I said, "Hey, can you do that prompt uh that I had asked you for?" And it wrote me this really awesome prompt. And a killer workflow is: deep research; use that deep research to write a prompt for you; and then use that prompt in the best model appropriate for that prompt. In today's case, we're going to use it in Gemini 2.5.
Deep research. So I'm going to take this prompt, which really breaks down—and if you want, I we can share share this prompt with all of you—as you can see, way better, way better than if I were just going to come up with something on my end. And so I'm going to come over to Gemini, going to paste in this prompt, and uh that I think is pretty awesome. I'm also at the end of this; I want to I want to just I want to give Gemini all the information possible, so I'm going to actually come back up here, and I'm going to share the actual raw deep research, because it summarized this research in my prompt, but I I find this is another tip: the more context you provide, the better. And so I'm going to copy and paste this back over to Gemini. No magic tricks here. And then you can see I'm in gemini.google.com, and I have picked 2.5 Pro experimental, the brand new new, the best model that Google has, and now we're going to see, and we're going to see it together, and it's going to crunch all the information I give it, and now it's thinking. And a reasoning model, you will see what it's doing here; it will always show you what it is thinking, and you can look at its thinking, so you can look and see, hey, here are the different things it needs to do: I need to understand the requests the user is asking of me, deconstruct the product information, structure a marketing plan, so it's it's telling you like that's the work it's doing, and I'm keeping a timer on all of this. This is pretty quick; it's we're doing this in real time. This is like I think a minute or so, so very fast time to come up with a strategy here. And so now, if you've watched the show before, and again if you're first time, hit like and subscribe. If you like what you're seeing, hit like and subscribe. What you're going to want to do is now do a review, and I'm going to give you my honest breakdown of like, as a marketer, do I think this is good? Do I think it's okay? Do I think, "Oh man, I would just could totally redo this?" And then I'm going to give you a summary of how I think this compares to other strategies that I've done in all the other models, other versions: Gemini, Claude, ChatGPT, everything historically. I've really liked ChatGPT 4.5 or 01 Pro or Claude 3.7 for marketing strategies; I felt like those have been the best models for building marketing strategies. So let's see. We've got HubSpot Breeze Customer Agent, B2B marketing strategy campaign. It's giving up me an executive summary: This marketing campaign aims to establish HubSpot Breeze customer agents as a leading AI-powered support solution for mid-market companies and enterprise solutions. It's going to leverage our seamless integration, multi-channel strategy, blah blah blah, great. Okay, it gets basically it gets the assignment. Now, target audience and segment: mid-market, one to 2,000 employees, got that right; enterprise, 2,000 plus employees. I didn't tell it that. It, through the deep research and through all the information it has about HubSpot, it's able to come up with that, showing key industries. Yeah, these are these are looking like some really good key industries for us. Key decision makers, personas. I didn't tell it to break it out, out personas. This is a good example of wow, it's doing some really, really great reasoning. Customer service, uh, VP of Customer Service, Support Managers, CTO, CIO, Head of IT, CEO, Business Leaders—these are all the right personas; these are all the type of people who be making a decision about the the product. So so far, very good work in terms of setting the stage for a campaign.
Now, value proposition and core messaging, uh, the core value proposition is: Effortlessly scale your B2B customer support with Breeze agent, an intelligent, seamless, integrated chatbot that delivers trusted, on-brand, 24/7 automated service using your existing knowledge. Very good. I I'm I'm very much aligned with that. Key messaging pillars: effortless, setup, trusted. So at HubSpot, we build products that are easy, fast, and unified. And so what I'd want to, and that's kind of the positioning on our website; that's what you would understand. And so I want to see if it's getting that. Oh, integration, unified, yep, ease of use, yep, trusted and accurate, scale, efficiency, reduce cost, enhance satisfaction. It kind of got the fast, but it definitely got easy and unified. So like, generally pretty good. Like I would probably go in and edit a few of these value props, but like it's not like, oh gosh, it totally missed the mark. So far, I'd say this is better than most of the experiences I've had in building a strategy in an AI product. And now, well, I like this; it's doing tailored messaging snippets for the individual kind of levels of decision makers: for CTOs, for managers, for VPs. I like that; I've never seen an AI model do that, and I like that a lot.
Channel-specific strategy. Now this is where the rubber meets the road. I love a good channel-specific strategy, and so I want to understand is it actually good at this. So website: Oh, we're gonna have a dedicated landing page, and it's giving me the content, which is compelling headline, and what's great is I could go and highlight any of these and ask it to like give me the whole landing page; say, give me all of the text and info for the landing page. And so while it is making that, let's go back through what is a very detailed strategy, right? And so it's showing you blog articles: Benefits of AI chatbots for B2B support, How to scale support without scaling cost, Breeze versus competitor. These are smart, reasonable things. What they're not is like very thought leadership, high point of view things, and I think if I had dropped a section of like, this is what we believe, this is why we're different, here are our provocative takes on the market, I think all of this would be much better, and that's something if I was iterating on this strategy, one thing I would do. So we've got case studies and testimonials, social media strategy. Interesting; it's it's giving us an opinion, primarily LinkedIn, and it's showing you what the and it's saying short-form videos on LinkedIn. That is very savvy. No one has done that yet. Like, as a marketer who's in the game, I can tell you that there is arbitrage in short-form videos on LinkedIn because LinkedIn's pushing its video product; we're doing a lot more with video. So to have an AI model pick up on this and the importance of this, very, very good. Uh, LinkedIn ads. I I would say traditionally these models are probably worst when it comes to ads in terms of where should they place them for effectiveness, so it's saying some LinkedIn ads here. We we have hit-or-miss experience with LinkedIn, uh, just because sometimes the costs are really high on LinkedIn versus maybe Google or Meta or other platforms, but it's giving you like, if you were not an expert marketer, I will tell you right now this is very good. Like if you were running a small business and you're like, I don't know what to do, I just need to do this; if you literally followed this advice, you would be better; you'd have a better marketing campaign than like 80, 85% of companies out there, and that's like that's a really good; that's a really good place to be at. It's even covering retargeting, giving me more sophistication here, which I like. It's giving us our content pillars: Thought leadership is, which is interesting; it has an interesting angle. Again, I think if I had given it an interesting angle, it would have been better. Uh, use cases and ROI so far, uh, measurement, optimization, and KPIs. I love that. How to optimize: So running A/B tests, refine audience targeting. I didn't give it a budget intentionally because I just wanted to see what it would do, and it gave me an allocation of where we should spend our money and time, and like it's not wrong. Would I would spend a little bit more on content and social, less on ads, but it's it's like directionally pretty good. It's giving us a timeline and project plan. Overall, overall, very, very good. Let's see what it did on the landing page. Okay, it then gave me the entire landing page that I could then go and edit, and it's it's holding place; it's telling me what type of visuals it needs; it is giving me a product solution section. So I can literally go through each of the sections on this plan and just say, build it for me, and I'll have one massive thread of an entire marketing campaign. It's not just a strategy; it's actually a campaign builder, and I think that's one of the differences with these new really advanced models is the older models were good at like helping you organize, structure, give you like an outline; then they got a little smarter, and they could actually like do a pretty good strategy for you. Now they could do the strategy deep enough, and then they could build out the different assets at a good level of quality for each part of that strategy. Like this is a very in-depth breakdown of this landing page and exactly what you would want to do: social proof and trust section, use case snippets. Like this is basically like a prime-time product page. Like if you're doing good product research and product marketing, this is the type page you would build.
So what do I think? I think the Gemini 2.5 has built the best marketing strategy that an AI that I have worked with has done to date, and I think unequivocally this is the best model if you were going to take a really good prompt and really good research—like we started with, remember those are key—and turn it into not just a full-on marketing strategy, but a full-on campaign. I think Gemini 2.5 can be your entire marketing campaign builder. You can take all of this, go build it in a tool like HubSpot, and get it out the door very quickly. Like if this was just me, within one to two days could have everything built for this entire campaign, other than the visuals, and I'd probably go over to ChatGPT 4.0 image and do the visuals. And maybe next time that's what we'll do on a future show is we'll take like a core asset and show you how you actually build it real time. Drop a comment down below if you if you'd want to see a show covering that. Uh, it's pretty awesome. So I would highly recommend if you haven't, try Gemini 2.5. It is free, which is mind-blowing, because Google's trying to pick up some market share; they want to be a leader in this AI race. And right now, if you are trying to grow your business, there's a huge arbitrage, and that arbitrage is that you can go use Gemini 2.5 for free to build great marketing strategies and to build the V1 of all those assets that you can then go edit, put in a tool like HubSpot, and get out to your customers, which is really, really remarkable. So check out 2.5. Hit like, subscribe. If there's any questions you have, leave those in the comments. We'll see you really soon on Marketing Against the Grain. This data is wrong every freaking time. Have you heard of HubSpot? HubSpot is a CRM platform where everything is fully integrated. Whoa, I can see the client's whole history: calls, support tickets, emails, and here's a task from 3 days ago I totally missed. HubSpot: Grow better.