Ai Agents: The Future Marketers You Can't Afford to Ignore

AI agents are going to transform marketing. We're going to tell you how, we're going to show you how they're already transforming search and you need to know about this to keep your business afloat. But we're going to show you other examples of how AI agents are going to start to transform marketing overnight. This is the show if you want to understand how to start to implement AI agents for your business. I'm your co host Kieran Flanagan here as always with my co host, Kip Bodner. We are going to get into today's show, AI agents. So, Kip, I was, you know, I like to like be obsessed about one thing at any point in time. And so I was really in the blend of advanced prompting and how you construct prompts. And then I realized when I was creating this custom GPT that did like a variety of things, basically had a menu of six things that it would been far better if I had created agents to autonomously do these tasks and were specialized on those tasks. And this is what we are going to talk to all of our listeners about today. Because I say to you, Kip, that in the future, not too distant future, there is a new layer that sits on top of the Internet and it is the AI agent layer and it is going to transform how we do everything inclusive of marketing.

Before we even get into that, can you break down what an agent is like, give people the, the basic lay of the land and then we're going to go into it.

All right. I think the best way to actually describe the differences or what an AI agent is is actually to describe the difference, the difference of an AI agent versus a co pilot. Because I think we've become accustomed to co pilots. ChatGPT is a co pilot, Gemini is a co pilot, Claude is a co pilot. All of these kind of AI assistant applications are copilot. So let me walk through a real use case that I think will exist in the future for marketers around content marketing of you have a copilot. The way this works is if you're using AI to do content and you have to use AI, Kip and I are telling you to be in the 5%. You have to start using these tools. If you're not using these tools, probably not the show for you, this is the show for the 5%. And so the copilot helps you figure out what research to do. Right? So this is me constructing a blog post or some content and then does the re. It can do the research for you. If you prompt them, you can like look at the research, you can figure out how do I create an interesting and post. You can ask and prod the co pilot to suggest some titles, to suggest introductions. You use all those suggestions, put together the content and then you write that first draft, you get some feedback from the copilot and then you actually post that content. Then you probably look at your data and say, okay, well how did this perform? Look to see what went well, what didn't and incorporate those learnings into your next batch of content. And that's kind of how you use copilots as like an assistant, how an AI agent would do that job as it would do it much more autonomously, right? You would say to it, hey agent, I actually want to grow traffic from blog by x percent. So the agent does not need constant guidance. Now what it does is it goes and does research. It actually looks to see how I can take that research and craft it into a post that will get traction based upon historical data. Then it creates the post, then it posts the post, then it actually goes back and starts to look at the data over time and improve itself without you really needed to give it much prodding or assistance. And what is actually really interesting in the AI agent space at the moment from people that I talk to that are really bullish on agents is we don't really know what that experience should be. Right. We're going to get into the Google example they give where the agent booked the entire flight. And what I mean by that is does the user look at the agent like look at the screens traversing until the agent does its job? Does the person just get a notification to say the job is done and they trust the agent does? In this scenario, do we just get a report each and every week from the work the agent has done and how that's performing over time? I don't think the UX experience for agents is a well known thing right now. But that is the difference, right, between a copilot and an agent.

Okay, so now that we've, we've level set as to what an agent actually is, what a copilot actually is, most people who are using AI right now are using copilots. And Kieran, I think what you and I are here to say is that the next 12 months of AI is all about agents and you're going to see more and more agents, especially for business use cases. You know, we're all about helping businesses grow, helping marketers get better here on the show. And so you're watching this show, you're going to see a ton of innovation happening in agents. And the reason for that is because agents is the next stage of automation. Right, Kieran? In the most simplest way to think about it, businesses have been obsessed with automation. Agents unlock a new layer of automation. It's almost like you had a skilled intern who can go do a bunch of things for you and then come back and you can, you can kind of review, approve and ship out to the world. I think that's probably going to be the initial agent, like user experience until agents get far, far more sophisticated. But I think this is going to be the year of agents. We're going to keep calling that out. And Kieran, we had Google I O recently and on Google I O, one of the big headlines was all around agents. And if Google's bringing agents to the masses, that will be a huge step forward in agent adoption.

Maybe.

Give everybody. I know you're sharing your screen here. Give everybody a little breakdown there.

Yeah, we should get in. So Google definitely veered towards agents and showed the different types of agents that can work with all of their different tools. Where we're going to start is search because they made some pretty big announcements in search that are 100% going to impact all of us marketers who acquire organic traffic. And if you are a really good marketer, a really great company, the likelihood is you get a bunch of your traffic from organic. The first we have to show because where is the. I actually lost the. Is it here? Up here? So this is the release post. This is something called AI Overview. And like within the PR launch there was this line, I think I slacked it to you, which is, hey, we're going to continue to be on the side of the publishers and actually ensure that the publishers get more and more traffic. Then I look at the AI Overview experience and I'm like, liars, liars. All right, so what we're seeing here is Google Overview. Someone searches. So look at the entire top of the organic page, right? If you actually look at the organic click through rates on organic listings. Like they go from the first three get the 80% of all clicks and then the next seven get the majority of the rest. Right now I struggle to see Google's like we're on the side of the publisher. Because what's really happening is this is an evolution of feature snippets. But now we actually do the search and, and the entire answer is generated and it's on the top part of the page. Now they are put in some citations in there. But if you have Your answer, if you have the question answered, why do you ever need to go and actually click on those links? Because the users are always going to veer towards quicker, easier, get the job done as fast as possible. And so if featured snippets, the results of featured snippets is anything to go by and we talked about it on this show before, which is the data that we had looked at showed that they had a huge impact on the amount of traffic you would get from the first 10 links on Google when feature snippets appeared in those pages. I can only imagine the decrease in click through rate on organic listings when the entire top half of the page is going to be these kind of AI generated answers. And so that is the very first announcement that they made and I think is a monumental, will have monumental impact on the ability to acquire organic traffic from Google.

I think we don't even understand how much impact it's going to have. It's going to be huge and probably a negative way in the short term, but I think it's going to get more positive in the long term. Kieran, I also think why we're starting with search in this conversation about agents, Kieran, is because robots are going to be the majority of searchers in the future. Right? Like we're talking about giving humans good AI results in the future. Agents are going to be out doing this research, doing this searching for you and they're going to bring back an even more kind of curated answer back than even a search engine. Right. And how we basically research and discover knowledge is in one of the most rapid periods of transformation since the Internet started.

All right, you're getting onto the agent part. So the first part, AI overview, is hugely, I think it's going to be monumentally impactful for organic search. As you said, we don't really know the results right now. We actually have a little bit of data coming up from a really good account on Twitter. I want to watch this little 58 second clip from Google IO and we're going to let it run and let Sundar do his thing because this is really, really a big change to search going forward and really speaks to the agents, AI agents and how Google thinks about search. And then I want to come back in and talk a little bit about why we were right, why you have to listen to this show. Because we call this, no, we call this in November 2023. So here we go.

Another example that's a bit more complex. Say you just moved to Chicago. You can imagine Gemini and Chrome Working together to help you do a number of things to get ready. Organizing, reasoning, synthesizing on your behalf. For example, you will want to explore the city and find services nearby, from dry cleaners to dog walkers. You'll have to update your new address across dozens of websites. Gemini can work across these tasks and will prompt you for more information when needed. So you're always in control. That part is really important. As we prototype these experiences, we are thinking hard about how to do it in a way that's private, secure and works for everyone. These are simple use cases, but they give you a good sense of the types of problems you want to solve by building intelligent systems that think ahead, reason and plan all on your behalf.

So what's happening here? So Google is moving from a search engine to an action engine. So what do we mean by that? Right, the historically what search has been is I search for something and then I complete the task. What Google, I think believes the future of search is, and we called this out when we saw ChatGPT and they launched the App Store and we said, well, ChatGPT would not just do the knowledge part for you, it will then complete the task. Well, Google demoed on stage is the ability for you to search for something and then the actual AI agent to like acquire the information and then complete the task. They showed an AI agent booking a flight on your behalf. They showed an AI agent ordering DoorDash on your behalf. They showed an AI agent working with all of these different apps that are plugged into the Gemini system and it could just do them on your behalf. This is the, I think the land grab or the war that is on between Google OpenAI really those two at the moment, which is like who builds the kind of action engine, the personal assistant that does the thing for you. And basically what we said that search is transitioning from is there is a layer on top of search that pushes the search back, that pushes the search into the background so you never really have to see the search again because you have AI agents doing that and taking the action on your behalf. And people will say, well, what does that mean for me as a marketer? I will be honest and say I'm not sure it is a such a monumental change to like human behavior and how we complete things on the Internet. Which is why I talked about having that AI agent layer that it's going to take a little bit of time to figure out. Like, what are the, what's the new buyer behaviors, what's the new consumer behaviors when we start to use and have access to these tools.

Hey guys, real quick. We've gotten a ton of requests in the comments and they're all about prompts. You all want the prompts we're talking about on that show. We see you, we hear you, and now we are ready to share the goods. Click the link in the description below to access our new AI prompt library. This resource features real AI prompts that Kieran and I have covered on this show. We broke them down by category and templatized them so you can fill in the blanks and make them your own. Now you have the frameworks you need to use AI, analyze ads, create brand campaigns, develop product stories, just to name a few of our favorites. Keep telling us what you need in the comments and we'll see what we can do to whip up more for you in the future. Now back to the show. All right, Kieran, I want to, I want to pause at some answers to your question right there because if you're watching today's show, you're like, all right, this is coming. What does this really mean for me? And like, what should I do to prepare and what should I consider doing? The first thing you need to understand is that access to knowledge is becoming commoditized. Google doesn't want to leave the search engine business. They just see that AI is going to commoditize and make it so easy and way past, way cheaper than free to actually get access to this information that they need to move to the next step, which is what you do with that information to take action on it. And there are a few things that's going to happen as Google, OpenAI and others do, that consumer and customer expectations are going to be reset. And what that means is that everybody is going to be expecting this like very action oriented, agent oriented experience. And that has a bunch of pile on ramifications. One that means that most companies are going to in some ways become technology companies because they're going to need to make it easy for an agent instead of a human to do something. And so what you're really talking about is solving for the era of robots. And anything that a human does to take basic action, whether for consumer or business purposes, assume that's going to get moved to robots over the next few years. And so what you have to do is say, hey, if a robot was solving this problem, what do they need access to? What data do they need? What systems need to be integrated together? What do we need to make publicly available right when a human's doing it? And you Might have a human into human interaction, then your customer service agent can access a bunch of stuff on the back end that's very proprietary. In a world of robots, you can't. That stuff has to be public facing on the Internet for a robot to interact with it. And those companies that are really slow to move towards serving robots in addition to humans are going to become obsolete because they're just going to be delivering a 10x worse experience than the more modern companies.

A lot of why we create the content is for Google, for Google to index, for Google to publish. And people are saying, well, like in the future kind of this breaks the relationship with the publisher. Why would you create content for Google or for these bots? Well, you'll still need to create the content, but the weird thing is maybe you're creating a lot of that content and it's structured in a certain way to train agents to pick up on that content so it can answer questions and do work on your behalf. So I don't like, maybe it changes a little bit, but you're still creating content with some sort of relationship with the kind of bot agent. The other thing I will say, Kip, that I wonder if you picked up on from the, the change. So one thing that actually was pretty interesting to me is I think Google may have inadvertently disrupted its golden cow, which is YouTube. And so why do I think that, right? What is the video search platform of choice today? It is YouTube, right? You can go into YouTube, you can look for a bunch of things. One of the big biggest use cases on YouTube is like how do I fix things? Right? I don't know if you've had that experience. I've been on YouTube, like how do I fix this? How do I fix that? How do I do this? How do I do that? One of the things Google show demoed as the ability to search from your phone and have the AI answer your question. So this was an example of someone who bought a record player. Record player did not work. They took a video of it and they were able to just have Google answer that through the new kind of AI overview screen. For me, that kind of cannibalizes YouTube in some ways because that search used to go to YouTube, now it will go to its own platform. And I do wonder about how that impacts your search strategy for even YouTube if a lot of those queries starts to get cannibalized by the AI overview. Because one of the things I believe is, or I had believed is YouTube is much more defensible, defensible than traditional Search. But then the kind of ability to do this does make me think that maybe even YouTube might be a little bit cannibalized by some of the AI improvements.

I think you're right. I think, I think we are unclear about how much kind of omni modal artificial intelligence AI that can answer text, video, audio is just going to be able to deliver very personalized experiences. So if you're out there and you're thinking, oh, what's safe in this AI world? I think one of the things that's not safe is kind of generic advice, right? Like, I think gone are the days of like somebody writing an article or making a video just to answer a question that is applicable to hundreds, thousands of people. Now it's going to be like, oh, what is this personalized answer just for me? And 80% of that might be the answer for everybody else, but the last 20% may be very different than what everybody else needed. And AI is uniquely positioned to automate and customize that answer in a way that just humans are not capable of really understanding right now. And I think we're going to see this be One of the core reasons agents exist is to kind of deliver more personalized automation and tasks versus very kind of broad based rule automation. Friend of the POD, Greg Eisenberg has talked a lot about one of his companies, Boring Marketer. So the Boring Marketer account I think has one of the better threads on examples of agents. We're talking about agents today. We want to make it really tangible. So what are some examples that they kind of glean from the Google I O announcements? And I think one of the very obvious one is that you're going to have email agents that are going to be virtual assistants. We've talked about the show already, booking, travel, arranging meetings, things like that, managing your files. And then one of the things they're pushing on is AI teammates or employees that agents also really do kind of become a teammate that can do collaborative work and get feedback. Search agents, which we just went over how Google's moving from a search engine to an action engine. AI advertising agents, how AI can generate and agents can generate ads for you and personalized ads. One of the things for marketers that I think super important that you need to know is that if you have a few variants of your ad creative today, when you're running ads, that's going to move to hundreds or thousands of ad and creative variants over the next few years, because AI is going to create highly personalized variants of your advertising and you're Going to have way more creative to manage, but you're going to get better conversion rates from it. So it's going to be worth managing. And so I completely agree on the advertising side of things. Analytics agents. So agents who help you find insights in your data. I think that is a great example. Project management. None of us want to do the menial project management tasks. I think agents are going to be able to automate that. Visual content agents. I thought this was an interesting one, which is like how do you help tell a story and visualize a story and use AI to do that instead of relying on canva or your PowerPoint skills or Google Slides skills? Content agents creating agents and then personalized agents who can help you kind of do cold outreach and prospecting and a lot of the connection work that you would need to do from a research and crafting of a message to go out to one other person. So I think these are some very compelling use cases. Kieran, when you were reading this thread, did you have a use case here that like really jumped out to you for agents?

I have like three for marketers that I think will exist in the not so distant future. The AI, the first of all, the AI, the ads one makes a ton of sense. I built a very low level version of that, which is basically what he's saying is the AI agent will be able to scan the website and from the website create the right titles, ad descriptions and imagery. And one of the things I was messing around with or actually built was a custom GPT that would scan a website and craft ads for it. Like you can actually do that pretty easily with ChatGPT and it works really well. And then I actually added styles of known copywriters that you could choose a style and then create the ad. That one a hundred percent, I think that is going to exist. And actually these platforms are getting so good at telling you where to spend your dollars, how to spend your dollars and craft the ad for you. Like pmax I think actually just chooses the title or creates a title for you. So they're starting to really actually craft the ad for you. The ones that I think are going to be really interesting. He mentioned a version of it, but I think the next step agent, which is today marketers generate demand and then kind of hand it off and coax it to do something and maybe it's handed a BDR takes it over or someone takes it over. I think the AI agent is going to be able to figure out how to provide that person with the next step, whether that Next step is they want to get into the product and demo something. The next step is they have a question. The next step is they need some information. The next step is, whatever it may be, they're going to be able to learn like what is the best next step and then craft and create that next step for you. And I think that is why I believe over time more and more of the customer journey collapses into marketing and marketing owns more of this because they'll be able to automate that and will not have to pass it across for a longer, you know, for time, a long, until way, way further in that journey. The other one I talked about a little bit like analytics is definitely one he covered that I don't know that we need to add to that. All of the data analytics recommendations should be done by AI over time. That's a mixture of like data science, mll, gen, AI. But that, that one should exist. The other one I do think will exist. Even though I've been pretty lukewarm on content marketing, there is something within content that I think over time AI agents will be able to get much better. Now maybe what the AI agent does is, is ingest data, figure out how to get better and provide you with first drafts. Right. So if you combine data from keyword research, you provide data from internally internal data sources. If you combine data from external things that you want to replicate, the agent should be able to learn enough and do some work and learn from that work how to craft better content over time. So I think something in content marketing will likely exist, but I don't think it will ever be. I'm still on the, on the side of it will never ship a fully finished piece of content other than the one that you said would be commoditized. That I agree. Which is if you create a one to many educational answer, AI will do that much better than you and it will do it at scale.

Maybe the simplest way to think about this, Kieran, is AI helps either you as somebody who's trying to solve a problem get unstuck or it helps your customer prospect get unstuck. Right. Like if somebody's rooting around your website trying to find an answer, AI is just going to help them find it way, way quicker, candidly. Right. Like you're going to have agents and copilots that are going to help do that. If you are stuck finding a story for this perfect piece of content you're writing or this video you're making, AI is going to help you find it and it's going to help you integrate it in a way that like, would have just taken you hours and hours and hours previously. Right. And that's the beauty of what we're talking about here. We're talking about automating the next higher order level of things that are actually like very hard and very complex. And what I suspect is going to happen, Kieran, is you're going to see a boom of agents over the next year. And a lot of those agents are going to not be that great when they first come out. They're going to do a couple of tasks, okay? And they're going to fail at a bunch of tasks and people are going to be like, oh, what are you talking about? But as the models underlying those agents get a lot better and the access to data they have gets a lot better, you're going to see agents doing very complex tasks that are going to unlock a ton of productivity as well as like huge gains for marketers. Because agents are better at guessing at a one to one prospect level than a marketer can ever be. When you're a marketer, you guess at like what your entire audience cares about. Agents are going to be able to take your guess on what the audience cares about and personalize it to that specific person. And that is a big, big unlock.

Right. I think your point there, which is really true, is agents will get better the more context they get. Today a lot of that context has come from browsers. You see a lot of these AI agent platforms released in browser extensions to be able to capture the information in the browser, provide the agent with context, and that context allows the agent to do work on your behalf. The reason I believe OpenAI released a desktop application is because they want to move towards the hardware. And if they can actually see. Now they did say that the desktop app today is not seeing what you're doing on the app, but maybe in the future you can have an option where you just turn that on because you're okay with the app learning what you do on a day to day basis. And then the app can start replicating that and have agents that do that on your behalf. That's where I think we're going. Which is why I believe one of these AI companies could actually build hardware like a phone, like a laptop, because then they can capture that data and have the model embedded into that hardware. Which is again, comes back to like, what would have been the most obvious company to have done that? Apple. So Apple, if you want to ping me, I'm happy to like come on board and give you some consultant to do that.

At some point maybe this year when we're doing shows I think of other shows maybe we should do like a bag fumbler episode of we just where we talked about all the companies that just like fumbled the the transition to AI who insists who had a huge head start and Apple is certainly on that list where Siri's been in everybody's life for years and years and years. And they're probably going to have to replace Siri with OpenAI on the back end because their infrastructure is. Is it good enough? Right. And I think that that's pretty crazy. All right, we're kind of getting to the end of today's show. We wanted to give you the overview of agents because at Google I o that was like a big theme was agents and really the changing evolution of search and Google moving from a search engine to an action engine. Karen, for people watching today's show.

What.

Are the couple of things you would tell them to do on the agent and AI search front to like start preparing themselves for the new world we're going to be working in?

I would map out your customer journey, break it into jobs to be done and start to see see what jobs to be done you think can be automatable by an agent and then actually have those stack, rank, prioritize and start to work on them. As the models get better, so do the ones that are doable today and then over time start to do the ones that the models where where the models get better. They you know you'll be able to do much more and more. But I think it starts with like the map of the job to be done and knowing what AI is good at today and should be good at tomorrow and then starting to prioritize the automation of those jobs to be done so you can spend your time on much higher value tasks. That is like the number one thing I would start doing today if I want to incorporate this into my business. And I think you will be best served to be ahead of the curve than behind the curve.

And then on the search front what do you think they should do?

I think if I was in search I am still optimizing and trying to improve my search strategy because it's still the best channel we have access to in terms of organic and figure out what my figure and be early to the LLM strategy. Right. I'm still not sure how the LLM strategy plays out because we don't know what the click through rate is going to be on the results it season there but Ethan's episode. Give a bunch of information on how you can start to prioritize, how you can start to optimize your search for LLM. So basically, how do you appear in the answers that a LLM gives when you're, when it's, when it's asked about queries related to your brand? I would start to do that today, right? I would start to look to see where I appear in those different search engines or those different LLM engines and I would start to figure out how I can do things today that can influence those LLMs in the future to have my business be surfaced in the same way it is surfaced in search. Because even if someone doesn't click on the link, it's still a great reinforcement of brand that you are appearing for all of the queries and questions related to your product.

I agree. The one other thing I'd add is that Google is essentially giving most businesses a split test, right? Because they're only rolling out AI search in the US initially. So most companies have some somewhat international search traffic, right. Traffic outside the U.S. i, what I, what I would do is I would daily, multiple times a week, look at my search traffic trends and especially cut it by us vs non us to understand what the behavior differences were for my business in terms of what the effect the AI driven search had versus the traditional search. And based on those learnings, I would start putting plays together to hopefully better optimize for that AI search engine as you start getting early data and the fact that it's just going to be us for a while does help you kind of isolate and figure that out a little bit more and compare to kind of the legacy way of search people.

Yeah, that's a good way to think about it. Split test, be in the data, see what's happening, be interested, be curious. That is the best way to, to, to be on trend.

All right, look, we're going to have a bunch of, a bunch of shows on agents over the next six months. We've got friends, friends and coworkers and folks who are deeply into it. We'll bring some of those folks on. So if you want to hear more about agents, hit subscribe like today's show and we will be back with you very soon on marketing against the Green. 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 three days ago. I totally missed HubSpot grow better.