[00:00:00] Speaker A: Hi, welcome. In this podcast, we talk B2B marketing and what it takes to know your customer innovate and profit. We're glad you made it. This is the campaign by 97th Floor.
[00:00:20] Speaker B: Hello. Happy Friday. I am Paxton Gray, CEO of 97th Floor. We are a performance marketing agency focused on generating revenue and leads for mid level and enterprise organizations through both organic and paid channels. Thank you for joining us today. This is The Campaign, a B2B marketing podcast about better knowing your audience, innovating beyond best practice, and converting visitors into customers. The campaign is a weekly conversation with B2B marketing leaders designed to fit as much value as possible into 30 minutes. You can catch the campaign live on Fridays at 2pm Eastern and you can find episodes on YouTube, iTunes and Spotify. And at 97th floor.com today we're joined for a second time by AI and SEO expert consultant Dr. Marie Haynes. Marie is known for her deep expertise in Google's research algorithms, machine learning systems, including the intricacies of eat. Marie is a fearless experimenter who thrives on exploring emerging tech. Now she's laser focused on mastering Google's AI mode and harnessing the power of LLMs like ChatGPT and Gemini to prepare her clients for a future where AI agents revolutionize. Revolutionize the web. Marie was with us at the end of March actually, and this is part two of our series when we talked about before in part one about AI in search, covering LLMs, AI mode, and how to adjust your strategy to show up in generative search. If you haven't listened to the episode, it'll be in the comments if for those of you listening live, you can also find
[email protected] today what we're doing is we're taking a step beyond search and exploring how AI is going to change basically everything else. There's a lot coming and it's coming very fast. And by the end of this episode, Marie is going to lay out a framework that we can use to accelerate our own company's adoption of AI to start getting ahead of what's necessary to take to stay competitive. So with that, Marie, thank you for joining us again today.
[00:02:10] Speaker C: Yeah, it's great to be here again, Paxton. This is the talk I'm most excited about. I think there's so much coming that we have no idea how to prepare for. So we need to start talking about it. So let's get right into it. 100 rambling. Let's just go.
[00:02:24] Speaker B: Yeah. And I'm calling it right now. We're gonna have a hard time keeping this to 30 minutes, I can say that right now. So to dive right in, I think it's going to be important for us to start talking about agents. And I think that's a word that everybody is hurt in some capacities, or at least to AI. But I think few people really understand or completely understand what that means, what that looks like. In your research, how would you describe agentic AI? What is an agent in the realm of AI?
[00:02:53] Speaker C: So an agent is an AI tool that can do things for you, that can take action for you. I think there's a bit of confusion about agents. They don't necessarily have to work autonomously in that you set them free and they just do stuff, although they can. But a lot of the time they do stuff on your behalf and agents can have memory about you. We saw just recently Gemini in Google released a version of Gemini that draws on your search history, which could be a little bit frightening, perhaps, but also could be very valuable. Then ChatGPT announced that now they're keeping memory across all your chats. I actually find it incredibly helpful and even GROK this week said that they're introducing memory, so agents can draw on your memory of what it knows about you in order to help you do things. One of the things that's important about agents is that they can use tools and so they can connect with APIs. You know, an agent might connect with your Google search console API and instead of me writing a program to say, you know, find me the keywords that we lost this month, I can just converse with my agent and my agent will go off and do it for me and maybe provide for me a monthly report or something like that. And then the other thing about agents is that they will be. I mean, agents can now, but we haven't seen a whole bunch of this yet. We'll be able to communicate with other agents. And that's the thing that I think we need to really talk about. Since we talked last time. Google announced this new agent to agent protocol, which is the thing that I think is the beginning of how the entirety of the web changes. Like, maybe I'm being overly dramatic about this, but I think that that's probably the main thing that we should talk about is what happens with that.
[00:04:38] Speaker B: Yeah, I do. I do want to for a second, spend just a little bit more time on agents. So Sundar Pichai, CEO of Google, said that within two to three years, Agentix Solutions will be deeply embedded into our workflow. And Demis Hasabis, who is CEO of DeepMind, the person really in charge of Gemini said that agents are going to fundamentally change how we use the Internet within the next couple of years. So I think that the term agent, you know, there's agent AI, which is Dharmesh, I believe his project and the way that they frame it is an agent is AI powered and it's specific, kind of relegated to specific job that this can do.
[00:05:27] Speaker C: Yep, that makes sense. Yeah.
[00:05:29] Speaker B: Now I think what's perhaps confusing for people to understand is why relegate AI to a specific job when theoretically you can say, well, this can do anything. Why do I need multiple agents instead of just AI, just do all these things for me?
[00:05:47] Speaker C: That's a really good point. I think some of us are already using agents without knowing it. So if you've used deep research, either ChatGPT, Gemini or Grok, deep research, those are agentic and they're specialized in knowing how to go out, find information. When I'm using, if I'm doing deep research on a project, it's not just doing a deep search on things. It goes to a website, browses the website, then says, oh, actually, you know, we need to learn more about this. And it goes to another website and surfs the web sort of like a human would. And I think the reason why we need multiple agents, although you have a good point, I think eventually we'll kind of, it'll all morph together, is that this is how the way AI works. When deep research first came out, let's say. So I think Gemini was the first to come out with it. It was okay. Like it was kind of mind blowing at the moment. When you. But when you compare it to what it is today, it's not as good. So these agents, the way that machine learning systems work is that they're continually learning to get better. And there's an excellent. If you have not seen, there's a documentary about Demis Asabas and DeepMind called AlphaGo and the game of how DeepMind learned to play the game Go. And it learned by playing billions of games against itself. And it had one goal, and that goal was to win the game of Go. And so that's kind of how these agents will improve. So let's say I have a research agent. Well, my goal is to provide me with better and better research. Not. I mean, you could say, well, yeah, the goal is to do all these things amazingly well, but if we hone in on one thing, it's really good. So I picture eventually I'll have multiple agents. One's really good at looking at traffic drops. One's really good at understanding new AI technology and how we can apply it in our lives. One's really good at helping me develop prompts. And each of these agents are kind of like employees that we would hire and then train them. I mean, Paxton, your employees, you probably have very few that are like, we do absolutely everything in the company. They're specialized in what they're learning to be good at. And that's the same way with agents, is they keep learning and they keep getting better and better at what they're doing. They make some mistakes, and they learn from those mistakes and then become better. So I think that the reason why we have multiple agents is that each of these will be trained to do one specific role and do that role well, and. And then eventually these agents will talk to each other and combine, get things done by working with each other, just like a company.
[00:08:26] Speaker B: Yeah, I'd imagine, too, that separation of agents would be based off of sets of data.
So you may have an agent that has access to all your medical records, and you want that to have specific instructions and duties versus just like a general agent has access to everything.
And then, yeah, tasks. Like, what is. What is the goal of this? You know, I was thinking about at a personal commerce level, theoretically, every person may have an agent specifically looking for gifts for a spouse. And that's all that agent is doing all year, is just looking and scouring for gifts. And you wouldn't from a resource perspective. I think that's another reason why the division of labor is important, because to a certain degree, I believe in how these models work. It becomes inefficient if it's doing a lot of different things versus becoming hyper specialized in one goal. Is that accurate from your research?
[00:09:23] Speaker C: Yeah. Yeah, that makes sense. I think the goal is, the important thing is that we're training these systems with all of our feedback. I don't really know all the details of how we'll train them, but they will learn with reinforcement learning. I don't know if you saw the new models that OpenAI released this week. O3 is just mind blowing. I. I was using it just before we got on the call here, and they said that they had, you know, advancements in reinforcement learning. Every way that, that I know of that AI models improve or AI systems improve is by teaching them, you know, either via human reinforcement to say, like, yeah, this was a great answer, or you accomplished your goal is one thing, and then also unsupervised in that the AI model itself says, like, did I reach my goal. Or maybe another AI model judges it to say, like, did you actually succeed? And if you did succeed, well, then let's prioritize those thinking patterns that got you there. If you didn't succeed, well, let's kind of think a different way the next time. And so. So, yeah, it kind of makes sense that allocating resources as they are important to answering specific goals is going to be important.
[00:10:32] Speaker B: Yeah, yeah. I love that what you're describing too, is not dissimilar to just the human brain. Hey, that thing worked. I'm going to reinforce that pattern. This thing doesn't work. I'm going to change that pattern and find.
[00:10:43] Speaker C: I always laugh at how Demisa Sabis talks about Atari games. That's how DeepMind first learned was by playing Atari games. And all that they gave the system was the score and the rules of the game, and then it had a goal of maximizing the score. One of the first games they gave it was, I don't know if you remember the game Breakout, where there's a little paddle that goes on the bottom and you have to hit the ball. And then the system learned that if you made a path up the side and hit the ball and on the top and. And broke all the things like they got the score increased. And so it did that all the time. And I'm like, well, that's how my brain was trained as well, because I played Atari games, you know, through my whole childhood for many, many hours, repeatedly. And every single time you failed at a level, you'd be like, okay, I'm going to change this little thing the next time. Or, you know, I had maybe had to work a hundred times to get past this one level. And then I trained my brain to, to think in that right way. So that's AI works so very much like our br. Kind of wild.
[00:11:43] Speaker B: So I think that's actually kind of a great segue into these agents. So we've got a fleet of agents, and this, I think, is going to apply at the organization level, and it's going to apply at the personal level. But the. These agents become more powerful when they have the ability to interact with different tools, sets of data and other agents. So Google, was it this week or last week they came out? It was this week, I believe is this week.
[00:12:09] Speaker C: Yeah.
[00:12:10] Speaker B: Yeah, it was.
Yeah, it does.
So tell us kind of what Google released and what that means.
[00:12:19] Speaker C: Yeah, I think if for anybody who's watching this, if you have the time, find Google's Google Cloud next event. They announced something like 200 things in this event. I've covered them all in newsletter that's coming out Monday. But most important here is the things related to agents. And one of the things that they're releasing is agent space already out. It's kind of hard to wrap your head around, but the video that they show in this event really makes it worthwhile to understand every agency. So now, right now it's limited, I think, to agents, to companies that have 100 seats or more.
And every person in your agency has their own little personal suite of agents. So let's say one person in your agency, Paxton, is responsible for, I don't know, optimizing content. And so they'll have their own little suite of agents that they've built with their thoughts on, you know, how they should incorporate keywords, what entities they're going to use. You know what, like all of those things, they'll have their own little suite of agents that help there. But also the company will have agents. And so you will have, let's talk, not necessarily SEO, but policy you'll have when you onboard an employee. You have to teach them, you know, this is how the company operates. This is when we, you know, this is your budget, you have for this. This is how you can take vacation time off policy. Like all of those things become a part, all those documents become a part of an agent that anybody in the company can have access to. And so instead of me asking like, can I have two weeks off this year, I just ask the agent. My agent asks your agent and has this communication. And what Google showed that really kind of made it click for, for me was they had one agent that was an onboarding agent and it had a list of tasks basically that you would do when you hire a new employee. And one of those tasks was to connect with the HR agent. And the HR agent would walk the person through all of the things that HR normally would. One might be a payroll agent which walks the person through setting up their banking data and all that stuff. And so what happens is a lot of the administrative tasks in a company become much quicker to do because of agents. You have to do less paperwork. You, you don't need people necessarily there to answer questions. And then all of these agents have the capability that the agent should know when it should be elevated to an actual human. Because obviously that's the big concern is that AI is not perfect. And, you know, we don't want to be making big mistakes with, with our, with our people, basically.
So agent space is one thing that every company should look into like right now it's just for big enterprises. But eventually Google said in one of their blog posts this week that every business will run with multiple agents. That's kind of hard to wrap your head around. Right? Right now we're having a hard time even understanding what agents are and Google's saying that eventually we're all going to use multiple agents. Yeah, it's kind of wild.
[00:15:31] Speaker B: I think it's as people look at this, you know, the HR example there was like zapier. I think people wrap, they understand zapier and I think they are commonly looking at agents as zapier, you know. So okay, employee onboards, therefore fire this eight like not fire, but begin working with the wrong word and run through these things. But I think it's more advanced than that because in the HR example the HR agent isn't going to then engage with the employee and say okay, employee, fill out this. It's that the HR agent is going to engage with the employee agent. So the employee merely onboards and then everything is done. It's not that they're walked through this list and if there is a problem with their bank account or something out of the ordinary, then something is flagged. But let's, yeah, I think it's important for people to understand this doesn't. It's not the automation of work. It is.
[00:16:39] Speaker C: It'S the automation with the knowledge domain behind it. Right. So it's one thing to automate like okay, next we're going to fill out your banking info. My personal agent will have all of my banking info. Yes, right.
[00:16:52] Speaker B: And access to your bank account.
[00:16:54] Speaker C: Yes, exactly. And let's say there's a question about my banking info. I can't think of an exact example but then, and it might be a question that nobody's ever asked before, then the knowledge should be able to be put together by this agent to answer that question. So really it's the knowledge that matters. And your company's knowledge is what makes us different than just zap here. So zapier is one part. Automation is a part of being an agent. But the most important part I think is communication. So if I have my specific, you know, let's say I'm, I'm booking a vacation and I've just started working for you and I don't know, like does the, does your policy cover that I can take a two week vacation a month into my, into my employment? Well, my agent has access to my calendar agent and all of them couldn't converse basically to Say, well, here's the policy and here's what needs to shift or whatever and answer all of my questions on that.
[00:17:57] Speaker B: Yeah, I think too.
I mean, this is why it's, it's landscape changing. Because my wife will have an agent and my employer does, and I do. And it knows my calendar. It also knows my wife's calendar. It knows my, my employer policy, and it also knows my interests, and it knows weather in different places and events in different places. And so right now it rely like we rely on me saying, you know what? I'd really like to go to Bali. Can I make that happen? And then I need to trigger some things. This, we can live in a world where it will say, hey, you know what? In three weeks, you don't have anything going on, Your wife doesn't have anything going on, your employer, you don't have any big meetings, and this is in compliance with your policy. The weather in Bali is really great. We know you like Bali and this festival is happening and flights are cheap. Click this button and you're, you're going to Bali in three weeks. I don't know. I mean, that's all possible.
[00:18:55] Speaker C: Yeah, you know, it is. And it, it's wild, but also kind of scary, right? Because there's a level of control there that, like, how do you know that your agent recommending you go to Bali is not influenced somehow?
[00:19:08] Speaker B: Yes.
[00:19:09] Speaker C: You know, I, I actually just did a consultation with somebody who does tourism in, in Bali, and we, you know, you could create an agent that if I was interested in traveling there, you know, connects with my agent to tell me all the things that I would love because maybe I don't want to do, like all the hiking and things that people like generally do. Maybe I want to focus on this one particular area. And so there's commercial incentive for people to create agents. And then maybe I'm jumping ahead here, but Google's agent to agent protocol kind of makes it sound like my agent will go out and instead of searching the web like we do now, will find other agents that are most likely to provide me with an answer. And that's where there's opportunity right now, is if you have a business that's providing people with information, if you can create an agent so that when this starts to happen that my personal agent is seeking out deep information, then you can embed your personal experience and expertise into your agent and make it look attractive to my agent and therefore make more money. There's all sorts of ethical implications we could talk about there. And I think that that's part of why it seems like it's taking Google a while to roll out some of this because it like really is world changing in the way that we access information and make decisions in our lives.
[00:20:31] Speaker B: It is world changing and it's, it's world changing. For Google, their entire business model is reliant on these ads appearing in certain places, both not just within search. And I mean that's a huge chunk but Google display ads. So like they still want people going to websites today, but all of this changes. So I think we've been using some personal examples and this is a marketing podcast, right? So how do we. But I think it's important those personal examples are going to mean like your market is going to be using the Internet in a different way than they're using it now. And so agent AI and agent to agent frameworks are just going to absolutely change the face of how people are accessing information. So let's talk about from a high level standpoint and then let's get, I do want to at some point get to a very practical standpoint, but high level. How do we as organizations prepare for this shift into agents, talking to agents, what should we be thinking about today?
[00:21:34] Speaker C: It's so overwhelming right now. You and I were just saying that to stay on top of all the things that happen even since like we just talked a couple of weeks ago and like we could talk for five hours on the things that happened since then.
So one of the things that I would say, because right now I think it's too early to go out building, I mean maybe you can build some agents, but I don't think they're going to be used really right now. And I think that the technology to build the agents. So Google came out with again this week the Agent Development Kit. And this is if you're at all into programming, even if you're not into coding, I would recommend that you try it. That you, they have a really good quick start guide. You can put that quick Start guide into Gemini or ChatGPT and say like, even if you're brand new, just say like, I don't even know how to start coding. Guide me through it one step at a time and see if you can build something with it. So if you're a business and you're like, what the heck do we do? How do we prepare? The first thing I would say is encourage all of your employees to be using AI in some capacity at least once, once or more times a day. And you may fail. Expect that you're going to fail. Lots of times expect that AI is not, it's not like you're going to plug it in and all of a sudden it automates everything that you do. You're going to get, you're going to learn. Communicating with AI is like speaking another language. And so I would encourage all of your team to learn how to do that, to learn to find new ways to do things and then share that with the rest of the company. I personally, if I was running an agency, I would have at least one person who is in charge of staying on top of what's chang in AI. I'll give another shout out to my newsletter. Like I eventually one of the agents I want to create is an agent that takes my newsletter on like here's what happened in AI and in search this week and then it goes with your agent and pulls out like what's important for you and it knows like, oh, you're working on this particular type of e commerce client. And you know, there was this new change in AI overviews for E commerce and puts that all together. So I would have one person at least who's responsible for just finding opportunities for the company and then I would just remain open to what's happening. One of the things that came out with the agent to agent protocol is that even though Google created this protocol, it's meant to work with any language model. I think there's like 200 different language models. So if you're like, oh, I haven't really been using Gemini. I mean I really think we should use Gemini for a lot of things because it's Google's. But if you love ChatGPT or you love Claude, you know, that's fine. Just continue to do what you can do with these tools and just experiment with things because eventually as you're creating stuff that actually works for you, then those become your agents and we'll be able to work together with other people on your team. So I don't, it's difficult because right now all these people are like, I can see there's potential for AI to do amazing things and I can see that there's potential for those who know how to work with AI. But I can't grab anything yet. You know, there's not a lot of people who are directly making money off of using AI at this point, but that will switch. And I think what's going to happen, I, I personally think it's going to happen over the next year that it will get easier and easier to create agents and then People will start to communicate with agents and then there's going to be this shift that people who know how to do that will have incredible opportunities. So stay on top of things, be open to learning new things and experiment is what I'd recommend.
[00:25:14] Speaker B: Yeah, I love that and I agree. You know, if you go out and build an agent today, yeah, I think you're gonna have a hard time getting people to actually use it. It's still very early days and whatever you build is gonna quickly get outdated. But I think maybe a comparison would be social media and pretending you're running a company. Back in 2005 when Facebook launches and people are saying, hey, this social media thing is going to be really big for businesses. There's not a lot that you can do in 2005 to get benefit from social media, but there is stuff you can do to prepare for when it gets big. You know, getting like branding on point, messaging on point, getting your tone of voice on point, all those become extremely important so that when this launches, you are ready to go. So an investment in understanding AI, its applications and building agents I think is going to be one that pays off. Even if it doesn't pay off today, it will absolutely pay off in the future for sure.
[00:26:09] Speaker C: Exactly. Yeah.
[00:26:11] Speaker B: Okay, I love that. Now let's talk about so if the world become so one example is voice search was one that like it was talked about so much six, seven years ago and it threatened to like totally disrupt our entire industry. And it just didn't. And I think part of the reason it didn't is because the limitations of what it could come and return to you. And I believe that this agent to agent framework is going to bring voice back, you know, because I don't need to see anything anymore. Whereas before it's like I really still needed to see something to fully get what I wanted out of voice search. So what other implications would we be looking at in this new world of like agent to agent communication? That's how most of the Internet or a lot of the Internet operates.
[00:27:06] Speaker C: I think we need to really be paying close attention to Google's Project Astra, which right now seems like a little toy. So Project Astra is. And you should have access to it. I think most people who have the Gemini app on the phone and you can use like live video to, you know, I could hold it up to the screen and say, who am I talking to now? And although it doesn't always recognize people, but I can ask questions about my plants, I can ask, you know, so let's say in the future I'm, I'm gardening and, and I see some weird bug on my plants or something. You know, I can hold up Project Astra and say what is this thing? And right now what it'll do is it'll use Google Lens so it'll show me results from search. Which means, so let's say you're a website that actually answers that question for you need to really be focusing on images because Lens is going to be looking for other images that look similar to this problem that I suggested.
But then I think it will eventually go much further that like let's say, you know, oh, that's aphids. It's horrible. You need to like treat it with something. And then my, my, you know, my, I'm saying my agent really Gemini, which I think will become most people's agents will talk about, you know, do you want to treat it naturally? Do you want to get a spray for it? And I'm like, you know, I just want to get rid of it quickly as possible. And then Gemini will say, well you know, you need this certain type of spray and then connect with agents of businesses near me that sell that spray and then say like, oh, it's actually on really cheap at Walmart. Do you want me to order it for you? And my agent has my credit card info and goes ahead and orders it. So that whole process that we went through, I never touched a computer to do that. And I think that's how things are going to change. And then to take it a step further, the next thing we're going to see is wearables that just this week there was a TED talk where Google's new version of glasses. So I'm sure you remember when Google Glass came out and like it was really not very good. And that's because it was like ahead of its time. Like they didn't have AI the new glasses look like very much like the glasses I'm wearing, just a little bit thicker rimmed and so you'll hardly be able to know that somebody's wearing them and they have a heads up display that you can see. So you know, I could be fixing something in my home and you know, Gemini can be telling me like make sure the shelf is the right angle and you know, and like actually coaching me as I'm doing stuff, which is pretty wild, right? So that's, I think when we first pictured voice search, like I don't like talking to my phone at all.
But we're going to do that I think because And I already find myself doing it with, I'm using Project Astra all the time and it's incredibly helpful for so many things.
[00:29:55] Speaker B: Yeah, yeah, that's, that's why I'd love to have an AI just listening to me all day and saying, hey, you remember you said you were going to do this thing.
I've already taken care of the first couple steps for you. Here's where I need you to step in. Or, you know, you could maybe consider altering your message in this way or these things, these thoughts don't align. Like, how would you analyze that? You know, just like something listening to and like helping me through my day. Oh, I absolutely love that.
But yeah, the implications for commerce is going to be fascinating when everybody in the world has agents operating for them and shopping for them. What's that going to do to E commerce when you're not necessarily doing a display ad anymore, but it's, you're trying to get your agent to appeal to other agents and, you know, advertising is going to change wildly.
[00:30:49] Speaker C: Yes, definitely. I think we're going to see a point where if you're searching for, let's say you're searching for new shoes that you want to buy, and my agent will connect with, directly with all of the different companies, you know, if it knows I'm interested in Nike or like certain brands, it'll connect with those agents to get me the, the details of the shoes, what the, what options are there. But let's say I want a comparison and I want to know, like, oh, I'm just getting into running. I'm not. But let's say I am just getting into running and I want particular shoes. Well, people who right now run affiliate websites, there could be some type of revenue model where you could actually pay for ads to say, you know, and speak to our running shoe expert agent. And then, you know, that agent recommends things to me and draws from there. Now, it would only be worthwhile if you really had, like, you weren't just comparing the specs of products. You really had insight that could only come from people who are runners or, you know, have experience. And that's why I think Google brought out. Well, I know it's why Google introduced this whole idea of experience in EEAT is that informational content that just collates all the information that's known in the world. Like, it's not really as needed now because AI can do that. But what AI will always need is new, fresh experiences, new fresh outlooks on things that are happening in the world. So I think that we're going to see that ads really morph from just, you know, here's a banner ad, here's a text ad to here's a conversational ad. And you know, I might run an ad if I was doing traffic drop assessments where you could have a conversation with an agent that's trained on all of my writings and my thoughts that I've put into it. And then if you want to upgrade to talk to me, it costs more money. And so I think that that's kind of one of the things that we'll see especially with SEO. I think a lot's going to change with SEO that like a lot of the tasks that we currently charge for people are, businesses are just going to be able to get from their agents. One of the agents that Google announced in or that Google gave as a demo, they have an agent garden is actually an SEO agent and it has multiple things that go out and do keyword research, competitor research compares your competitors title tags to your title tags and finds opportunities for you to optimize. Like it's nowhere. I'm sure it's nowhere near as good as what your team would do yet. But like it's going to continue to improve as it goes. So everybody who has expertise in an area, it's going to push us to have better expertise to not just be like the collaters of information, but rather to like be something that can be better than AI, which I think there'll be a window of time where we can do that and then you know, maybe we get to, I don't know how many years in the future. Bill Gates said recently that he says within 10 years AI will be able to do pretty much everything that humans can do on a cognitive level. Which is pretty scary, right?
[00:33:57] Speaker B: Yeah, it's wild. I am fascinated at the idea of analytics with agents because we could, you know, Kickstarters perhaps closer to this where they say hey, here's a product idea. If you like it, give us your money and then we'll build it.
If, if there is some technology to amass preferences from all of these agents around the world of buyers, you know, it may be that they'll be able to say hey you know What, I've got 25,000 buyers right now that are authorized to purchase self lacing shoes. So Nike, if you build this, I've already got 25,000 buyers. As soon as you build it, they're going to click purchase and it'll be sent, you know, Amazing.
[00:34:41] Speaker C: Yeah.
[00:34:41] Speaker B: And so it can be that the market we can more quickly drive product innovation and new products. You know, it may be like on the B2B side I would, you know, if I had an agent out shopping, I would be looking for product management systems that are less expensive than what I'm paying for right now. And if there was some team that said, hey, you know what, I've got, you know, $50 million worth of money, set it like that. I know if I build something cheaper, they're all going to switch away from these more expensive ones if I can just build what they need and that would drive so much innovation much faster.
[00:35:21] Speaker C: It will temporarily because eventually you'll be able to say I wish I had a product management suite that did this and then I'll just build it myself. Exactly. I don't know if you've played with in Gemini Advanced there's Canvas right now which I didn't understand Canvas at first but this week I've been playing with it. I saw Ethan Mollick on LinkedIn had a post where he said if you're trying to understand a paper, put it into Gemini 2.5 in advanced and ask Gemini to make a game out of it. So I did this with the paper I was reading on Topic Authority and then Gemini within 30 seconds wrote this whole entire game in HTML and JavaScript and then played the game and it opened up and had me categorize topics and it wasn't perfect but that was within 30 seconds. And and then the new O models from Chat GPT are just wild this morning. I, I'm always trying to brainstorm with these models and I was like, I want to do more with my newsletter. I'm telling you the conversation that I had with O3 this morning brought forth like patentable ideas of, you know, you could do this. I'm currently using ConvertKit, you could use the Automations in this way. It's and then it made this plan for me where I'm going to like do this, this and this and said okay, you're going to start on Tuesday with this step. Do you want me to send you a reminder? And then chatgpt tasks set up to sent me a reminder to get this set up. So basically you're right, innovation is going to be driven by the people who know how to use these tools.
[00:36:58] Speaker B: This has been such a good conversation. As predicted, we could not stay within our 30 minute time limit. So great. And I think overall we're talking about a lot of, of big philosophical ideas but the main takeaway, I think that's important for People listening today is dive into agents, learn how they work, start learning how to build them and how exactly you're going to employ them in the future. You know, we don't know for sure yet, but the best way to prepare for that is to start getting your feet wet and embrace the this awesome new world that that's starting to unfold.
Marie, thank you so much for joining for joining us twice and diving in this. It's a pleasure talking with you. How would you like for people to get in touch with you?
[00:37:48] Speaker C: Yeah, you can find me at my
[email protected] we didn't get to talk about the brain stuff that I was focused. But you know what, that could be a whole talk on its own. I recently just put out a podcast with a bunch of ideas on where AI is headed in the future and it won't be long, I think before we have not necessarily implants.
One study that I put in my newsletter this week, not a study, a company actually has an implant that just goes under the skin in your scalp and reads your brain activity that you can communicate with machines. So like it's just wild what we're thinking of. So yeah, community.mariehanes.com is one thing. My newsletter
[email protected] newsletter I am on socials. You can find me at Marie Haines and and I'm always open to to emails as well. You can helparee haynes.com but although I'm a bit late to get back to to most emails. But yeah, happy to talk.
[00:38:45] Speaker B: Okay great. Thank you so much for joining us.
Thank you for listening. We're going to take all the tips and action items Marie gave us today and write them up for you so that you're registered.
And if you're watching, if you're registered watching here on LinkedIn and that's going to hit your inbox early next week. For everyone else watching, listening after the fact, you can find all those resources and listen to past episodes at 97th Floor.com.
[00:39:09] Speaker A: Thanks for listening. The campaign is produced by 97th Floor, a 20 year old marketing agency that helps companies like McKinsey, Pluralsight and Check Point know their customers execute innovative campaigns and drive profitable growth. If you have an allocated growth budget and product market fit, we'd love to do research and build a proposal for you. Visit
[email protected] and if you enjoyed this episode, make sure to subscribe. See you next time.
