How to Win at AEO and SEO Using AI Agents in 2027

Grok bot and Jev. Okay, so that's what we're going to cover first. So what we have on my screen over here is my bot, and it's literally just called AEO SEO bot. Okay, and this is something that we were trying to do internally. So this is the bot that runs a content system that we have. So each weekday it pulls what buyers said on calls, what the team decided in meetings, and what is already working on YouTube and X. And it puts that in one table. And we have Jev. Jev is the classifier, right? Which we'll talk about in a second. It'll pass a row before anyone writes. And then if a page already covers the job, so is this duplicate content or not, then it will basically give a confidence score on whether it's duplicate content or not, right? And then we have a couple more lanes with this bot. So on Saturday, it'll check what competitors are ranking for. So what are competitors ranking for from an SEO and AEO standpoint that we should be ranking for? Let's say based on the products and services that we sell. Okay, so that's one piece of it, right? We also use ClickFlow so we can see what content is decaying and what's losing clicks and should we be drafting an update or should we be consolidating content? Should we be updating it, right? And then we have a GEO audit bot that sends citation gaps and then they send it over here to this bot. So that's what it is ultimately. And then we have a measurement bot over here as well. So basically we can track over time, hey, What's being published every single week and what's actually getting indexed by Google. That's still important, right? You also want to make sure, hey, what's actually generating impressions? How many impressions are you generating from this new content? How many clicks are you generating? But more importantly for us, how many free consultations are we generating, right? And so this is still early days for us, but I'll show you how we've gone about measuring this because I've basically started to roll this out like last week or so. So this is what the bot does. And then I'm going to show you, this is like a table that it kicks out over here, okay? So just so you guys know, it pulls from my content. So it can pull from like my X account, it can pull from YouTube, it can pull from internal calls that I'm having. It's looking for content spikes. Okay. What it says over here is like, hey, this is clear to publish because from— is this already live? On a scale of 0 to 1, it's 0.32, right? So no. Does it seem like it's— are we inventing statistics in this one? No, right? And so we have a couple of gates over here, like criteria that we set when it comes to JEVs. So you can say, hey, from an AEO SEO standpoint, have we actually published this piece of content before? Okay, great. Because you don't want to publish more duplicate content. That's the scary thing with AI, right? 2, do you actually have real statistics to back this up or are you just making stuff up? 3, we also have other gates where it's like, hey, if Eric actually talked about this and this is in a YouTube video, why don't we just embed that part of the YouTube video inside of the blog post? Also, are we putting in images in there as well, which I'll show you an example blog post in a moment. Here's the other thing. I made a video yesterday because OpenAI Dev Day happened, right? So what OpenAI Dev Day changed for marketers? Okay, so it said it's already live and AITell is 0.72, but I don't actually believe that. And so I might just reply over here and say, hey, so for number 5 on your list over here for OpenAI Dev Day, like we published that to YouTube, but where do we have that on our blog from an AEO SEO standpoint? And where are you calling out that there's an AI tell here? Are you looking at the YouTube video or what are you looking at? Because we need to tighten this up. So it's just saying, hey, these are potential ideas. Okay. Programmatic SEO plus search visibility, AI visibility, AI-powered CRO, buying gate. Okay. From LLM SEO to managed marketing agents. So it's suggesting that there's a gap from an AEO standpoint here that we should be targeting. But notice, Jev is saying these are all holds over here. So 1 of, what is this? 1 of 8. Made it through. Okay, so is this live already? On a confidence score level, it's 0 to 1. Okay, so it's 0.38. So it's, you know, it's, it's not that high, but it's, you know, it's kind of getting up there. Um, are we inventing statistics? You know, 0 to 1. Um, is there AI tell? Okay, so if their AI tell is like above 0.5, or honestly above like 0.40, um, then we probably want to redo that, that, that piece, right? But the AI tell is too high. We've already published it and we're inventing stats. We need to go ahead and redo that, right? And then you can just have this thing redo it. So look, this table over here, this was pulled— oh, I guess this is today. So I had a YouTube video go out yesterday. So keep in mind, it's constantly scanning, okay? So 11,000 ads in a month. I literally had a video that said that where Cody Schneider last week when we did this webinar, he's talking about how he makes 11,000 ads in a month, right? His process for doing it. Okay? This is on YouTube. It's saying it's live on YouTube, right? Or Opus versus Astra thumbnail task or grading a LinkedIn sequence or forecasting the SOW. So it's saying these are skip, okay? And it's saying these are draft over here. So keep in mind, am I going to say this is perfect? No, but look, it's pulling this from Granola over here. So it knows when I say something interesting, it'll pull from my transcripts or my Granola, for example. So I actually think my most interesting content is when I'm talking with my team because I'm not trying to create content. I'm not actually thinking about it. It just kind of comes out. So is that something I might want to draft? I don't know, but that definitely is a spike, right? Probably not something I want to draft, but because we're focused on driving more customers that want marketing, like AI marketing implementation. So that's us, right? So like this one over here, okay, hey, what OpenAI Dev Day change for marketers? Okay, it says draft this one. So let me show you an example because I can just keep going through these tables, but hopefully you've seen enough here where you understand that it will basically present things to you and you still need a human in the loop. So humans aren't going anywhere, guys. Like you still need a human in the beginning and then let the agent do the work in the middle and you still need a human at the end. It's not, oh, I've replaced all my people. No, no, no, no, no. We need to hire more people that know how to use this stuff. That is the right mindset. Not I'm looking to cut headcount. That's not how you're going to win long-term, okay? So anyway, let me show you an example. I'm going to show you 2 examples. And by the way, shout out to WPvibe. So we're still using WordPress today. I know it's dated, but WordPress, WPvibe is an MCP with WordPress. So it allows, Grokbot here and all my other connectors to talk to my WordPress and then draft the post out there, okay? So let me show you an example of what this looks like. So this post over here says, okay, how to stop your content agent from quietly killing the posts that already rank, okay? So what does this do? Okay, should this brief open a blank draft or touch a page that already ranks? Okay, so it will cite the video that I made over here, right? Sometimes it'll embed the video. It'll say, hey, what do we merge? What do we update? What's net new? Okay? It'll also create tables in here. It'll also cite another discussion I have around Jev over here. Here's a table of contents down here, which would probably honestly go higher. What this also does too is we, whenever we can, we use the last 30 days skill, which is a great skill from Matt Van Horn. And it's supposed to pull statistics and put the statistics into the article itself. So this one's like how to embed a 4-deployed marketer who ships agents. Because what we hire at my company is 4-deployed marketers, right? So how do you embed a 4-deployed marketer who ships agents? Okay, here's what your marketing CRM does. Or sorry, your marketing team has a CRM full of stalled deals, a backlog of automation ideas, and nobody responsible. So does that kind of read like AI? A little bit, right? This is why you still need human editors, guys. Okay, so this one over here. Here's a table of contents. Okay, great. It's higher. Okay. Strategy decks over here. Okay, great. You have numbers, right? Eric's discussion of agent adoption shows 17 of 100 companies using AI agents, da da da. 3.5 covering marketing staff. Okay, fine. Internal links to our services pages, right? Or our answer engine optimization pages. Those are always helpful. And then let's see what else we got. We have some visual diagrams over here, which has been upgraded, right? After I started seeing these, which look like they're AI generated, I'm just like, hey, let's go with the pastels, right? So the idea is you're constantly improving these over and over. Okay, here we go. So for supporting evidence, a 2011, maybe that's a little too old, right? Maybe I need to refine it a little bit and say, hey, we only want stuff after 2020, okay? And so you still need solid SEO, you still need You know, good statistics, good content. You need to have, you know, all the double E-A-T, like all the algorithm stuff to think about. Like links still matter too, right? And now it's just about, okay, how do you make sure that you format things where it's easy for agents to pick up? How do you make sure that you have good technical documentation when it comes to, you know, making it easy for agents to find that documentation? The LLMs have learned what Google has known for a long time. You need other people to link to you, right? Or to cite you. And so citations are the name of the game. And, you know, having a brand being omnichannel where you're creating content across the board. And I'm going to show you the measurement piece too. So let me show you this because again, I just started rolling this out last week, but my whole thing right now is if I go back to Grokbot, so let's look at this. If you think about this, okay, if AI actually gives you the opportunity to do more, you're going to want to have more output because look at the, The only bottleneck before when it came to content generation was how much content could you get out? How much could you research? How much could you write? And how much could you review? Well, okay. Now that you can research very quickly because the bot can research for you, now that you can write very quickly and then you could have evals for the writing, okay, well, that's good because then that kind of does the editor's work as well. And then the editor kind of just does the final checks. So Let me move over to the measurement piece. Okay. So what has happened so far with these publishing? Okay. If I scroll up over here. So first when we publish, okay. So I'm like, okay, let's just get 20 out in a week. Let's see what happens. Okay. So we get 21 posts live, but we had an indexation issue. So none of the posts were getting indexed. Okay. Now we kind of have that being fixed. And now it's like, okay, here's how many impressions we got from it. Here's how many clicks we got from it. This is as of just recently, let's just call it that. And so what's going to happen now is I'm going to continue to look at this, but the key here is this. It's not that these results are very small right now, it's more so as I'm testing this, are the numbers getting better, right? Are the numbers getting better? Because we had tested this with the earlier SEO AEO flow I was talking about, and that one's actually generating clicks now. We saw like 4 weeks later that those posts are generating clicks. And it's starting to compound, right? And so this stuff takes time, but how do you index these things faster? How do you make sure that you have the right internal linking pointing to this stuff too? Because again, SEO and AEO kind of go hand in hand. And so, but you do need a measurement bot here too. So we have the AEO SEO bot that helps with the content that we ingest or again, competitor content or decaying content, GEO audit. It'll figure the content out there and then we'll draft it up. And then the next step is we want to make sure we measure it because if we can't measure it, we can't manage it ultimately. So that's how I think about this right now. By the way, you can see on the right side over here, the good thing about GrokBot is you can have multiple routines for each bot, right? So measurement daily cohort card. Okay. So this is weekdays at 8:05 AM. And then there's a density index watch and then there's a decay fix. Okay. So that's run at 14 days and 28 days, right? Now, Now when you look at this AEO SEO bot, we have this weekly content ingester which happens at 8:35 AM. So you can change up these routines if you want. But with these other bots, sometimes you get lost with all the routines that you have. The cool thing is it's kind of organized here. So by the way, earlier as I was doing this podcast, podcast, I was doing this webinar with you guys. I was like, hey, this seems to be an error over here. Why are you saying that we covered this already? When we don't have this on the blog, so it says row five held because Jev thought a page had already covered it. I'm opening that score now to see which URLs use. There's no dev day post on the blog. Idea came from the YouTube video. Okay, what's the highest leverage thing to do? Is this high leverage? Should we even be doing this? Right, and a good question I like to ask quite a bit is now that you built this, how would you rebuild it? Because after they built it, the bot's like, oh, you know, here's what I would do differently next time. So I ask that question quite a bit. So take that for what it is. So literally all I did was I made a leveling up content engine and I said, hey, take the Marketing School episode from last week, or sorry, I did this last week. I did this on Friday. I said, hey, I want you to package the best moments from this Marketing School episode and I want to test this on my YouTube channel. I want to see how it goes. So it will basically edit the video. It'll come up with thumbnails as well. And I'll probably spend 5 to 10 minutes on the thumbnails. And here's the thing, the thumbnails become a lot more interesting because you as a creator of that video, you know what You know what resonates. You have taste, and you're going to get the best results that way. Okay, just straight up. And if it doesn't work, you can test it quickly. And then by the way, who's scheduling this? ChatGPT is scheduling it. Who's making the repurposing videos? Guess what? We put in Jev there and say, "Hey, using the Jev gates," I just call them Jev gating, right? Let's Jev gate this, and let's figure out what should we repurpose. What deserves to be repurposed from this video over here to to our other channels such as X, such as LinkedIn, whatever it is, okay? And sometimes they'll say, hey, this video should probably just go to X and LinkedIn. We should leave the other ones alone. That's fair. We didn't have that layer of judgment before. So people were just blindly following a process saying, oh yeah, we got to hit one per day, one per day, one per day, right? On these. And so, but now it's just like, hey, if it's actually good and Jeb says it's good and you're continually refining that eval, it's going to get better, right? And then also as I continue to tweak the thumbnails here, It learns what my preferences are with the thumbnails. But I use Codex for heavier lift work. And I found that it's just been incredible for design type work and also getting new landing pages out too. So I was like, guys, why don't we— we're trying to float out these new offers. Why don't we just crank out some new landing pages? And I was just like, hey, why don't you just look at the singlegrain.com aesthetic on the website? And this is actually what it looks like. Look, I was like, hey, why don't we just offer people a free managed marketing agent, right? As an offer. And reality is we can probably only do like one of these, but let's just get these out there and then target account-based marketing, like larger companies, right? And apply for a free pilot over here. Here's who we work with. Here's the agent. And then here's the application. Here's how it works. Here's the FAQ. How soon can we start executing? All these things over here. So not bad. Now we're going to move over to talk a little bit about organic social and a little bit of Meta ads too. Okay. My my Muse bot here named Doodoo. What do I say? I say, you know what? I just dictate using Whisper Flow. I say, hey, we have these new offers coming out. I want you to scan because you're connected to my Instagram. I want you to scan the top performing organic reels that I've put out there and only the reels that are good for repurposing to my ICP. What did it do? It found six of these, like this tier list one, like you know the the. S-tier through D-tier AI tools for $20 million plus businesses. Okay? Or the CFO bot that we have for 9 and 10-figure revenue companies. This is a little dated now. Just keep in mind because this stuff is changing so quickly, but at least it identified 6 and it said, hey, I need you to record a call to action at the end of it. You know what I did? I dragged that call to action video in here. I made like 4 of them and I said, hey, you can use these variations. I want you to edit it for me. It did it. Not only that, I said, hey, now I need you to spin up different landing page variations. Okay? So let's see what it said over here. Okay, so here's a design handoff. I said, hey, Muse, I don't trust you for the design, so give me the handoff document and I'm going to pass it over to Codex, which is what you just saw, the designs that you just saw. Okay? So I'm kind of just going back and forth here and it's like, okay, here's a landing page. Thank you. Qualified almost. I dictated to Doodoo first, and then I took the handoff document and I passed it over to Codex. So it wasn't as simple as me just saying I dictated the last one. So I stand corrected, or I'll correct myself. But all the concepts over here, like it showed what the HTML looked like. It didn't look that good, right? It kind of looked like just regular AI generic stuff. I'm like, okay, well, one, I don't want to edit the videos. Two, I don't want to post it. Three, I don't want to log into Meta Ads. 4, I don't want to find what's actually performed for me. And so, can you just draft it for me? It literally drafted it for me inside of Meta Ads. It's sitting there. It's sitting there with— and now that we have these ads that are edited, I just need to review them. But guess what? I'm going to iterate. And the key thing, guys, is that we're building, we're measuring, we're iterating. We're building, we're measuring, we're iterating. And most teams, not most, a lot of teams I should say, including ourselves, sometimes we have this disease called build— it's a building disease, build, build, build, but it should be build, measure, iterate, build, measure, iterate.