GES Cohort 3 AI Search Webinar

Uh, so right, uh, Go-To-Market Engineer School kicks off Monday and we've got some, uh, eager students, uh, that want to kind of get their head in the game. Uh, I'm Jared, I live in Silicon Valley, uh, currently, uh, running this program and doing some operating partner, uh, activity for, for VC firm around go-to-market, go-to-market engineering. Prior to this, I led growth at Auth0 and then campaigns over at Okta. I'm a lifelong learner. I love taking courses, courses like Reforge and Pavilion, Win by Design. So we tried to build this course kind of modeled after some of those other programs where you get to meet people, you get to be hands-on, you get a little theory. This we try to do like, you know, 70% hands-on building and, you know, 30% theory. And yeah, that's the general overview here. We all know that tech is moving super duper fast, and so we have to devote time. We just have to put time aside to stay on top of it. So that's why we built this course. You're gonna get about, I don't know, 20 different mini lessons across, well, actually 6 modules. And you know, you're gonna meet people in the class that are RevOps managers and AEs and SDRs and growth and content managers and PMMs, all kinds of folks. Let's see. So today we're gonna do a little bit of a chat and then a little bit of a Q&A. Yeah, we're gonna get kicked off on Monday. It's gonna be kind of a just what to expect, a little overview of what's going on, go to market engineering briefly, primer on that, and then kind of what to be prepared for in the course. And then we're gonna have— then we're gonna start with Yash from Clay talking about enrichment. Research and scoring, and then you're gonna get a bunch of at-home exercises to do on that with some live help sessions sprinkled throughout the week. And then we're gonna go into messaging and personalization with Zach from Octave. And then we're gonna dig into any, any little bit with Lorenz Ness doing some workflow automation systems there, building some AI email responders. And, and then Philip from Zapier is going to do a module on inbound automation and some structure and theory around that. You get some hands-on work. And, and then we're going to do on March 18th this module around AI search content. LM optimization. And then finally, we're going to have a module with a couple of great guys from Ashby, and we're going to drill into Cloud Code and build a couple of things there, including some analytics. So that's basically the course overview. Then we're going to have a capstone project for anybody that wants to participate in that. Allows you to work on something related to your job or what you want to build, get some feedback on it. And weekly schedule, Wednesdays are the lectures from 8 to 10 Pacific time. There, there is a recording available if you miss it. It'll be in the learning management system. And then, yeah, we're gonna have a handful of live help sessions from the Thursday to the following Tuesday, where you can jump in, share your screen, say, hey, I got to step 4 in the homework exercise, but then I had a question here, or Oh, how do I relate this to a project that I'm doing? So we got practitioners there in the live help sessions that are going to help you. So yeah, hands-on learning. You're going to get premium access and tools in the course Slack community. We've got a bunch of great TAs. We've got a bunch of great students from various companies. I think this cohort we've got students from Gorgeous and Teleport and let's see, Verizon even, Storyblocks, CircleCI, and Writer, and many more. So with that, I am going to stop my screen share and then kick it off and introduce Owen, who's running growth over at AirOps. And Wilson, who's taking the reins on training right now. So take it away, guys.

All right, let's go. Our aim is we have a couple of slides, like Jared, we'll get through these as quickly as possible, and then we want to show you all under the hood in AirOps. As questions come along, I'm going to keep my chat window open. As we go through the slides, keep your questions coming. But as Jared said, I'm Owen, Head of Growth over here at AirOps. This is our third time, I believe, Jared, on the GTM Engineering School cohort. I come from a growth engineering background myself. I love stitching tools together. So if we have time at the end, we'll open it up as well for like kind of a whiteboarding session, which is one of my favorite tools outside of everything in AI today. And then Wilson, do you wanna just give a quick intro on your side?

Yeah, thanks, Owen. So I'm over at the cohort side at AirOps. I lead cohorts for, a number of our builders and clients. So right now we actually have some students from Netflix and BetterHelp, super excited to have them in class with us. Um, but yeah, I essentially help people think about how to architecture workflows and orchestrate it across scale.

I love it. And he'll be putting that into practice in just a few minutes with myself. So before we do that, Let's look at like, why are we here? And I just finished a roundtable. I just spoke for a long time. So I'm a little bit out of breath. But when I think about being in growth today, it's actually like anxiety driving. Here are my 3 points I was listing over the weekend. But it feels like the AI revolution is here. It feels like traditional playbooks are dead. And then every single day you fall further behind. There's this thing like OpenClaw, which I wish I knew a lot more about. I'm only catching up right now on like Claude Code. Every day, like, feels like you're behind. And I think, uh, to give kudos to everyone here, you're probably already in the top 1%. Um, like, for you, AI is not gonna replace you. If anything, it's gonna, like, empower you. And I love this quote by, like, Ross Simmons. He's a big guy in, like, the SEO world. But mediocre marketers will be replaced by AI, and but great marketers will use AI to do great work faster. And I think, like, this is a great call-out. And, um, another Another great quote, um, I was looking at this week from Austin over at Anthropic is that this ability gap is closing. And this is why I think like the GTM engineering school is like so important because like he says, the gap between 'I wish this existed' and 'I can actually build this myself' is just like getting smaller and smaller every single day. Like you do not need to have an engineering team to do a lot of the stuff I would have needed to do on my growth team. 6 and 7 years ago. Now you can just vibe code it or architect it a little bit better, and— but the gap is like well and truly closing, which is why we now have stuff like GTM engineers and content engineers. The, the actual engineering part is becoming easier and easier to do. So kind of opening it back up to the world we live in at AirOps, which is organic growth. Just like everything with AI, it's having an impact on, um, all things search too. And as we mentioned down the bottom right, how you win today honestly isn't all that different, but definitely bars for quality, bars for velocity in different areas are getting higher and higher. So we probably all know what like SEO is, it's search engine optimization. I'll give you the playbook from about 5 years ago. You would look at what's getting ranked in positions 1, 2, and 3. If you were in position 10, you'd be like, hey, let's steal this paragraph and rewrite it. Let's steal their use of images. You'd kind of create a hodgepodge of what the best pages did. You do that on your page, you temporarily get to the top of the listing, and someone else would repeat that work in 2 weeks' time. So it was a constant battle, but it was like manual. Now what works today is— and I think Zach from, um, it's like from Octave was on, a lot of the other, um, teachers within this GTM Engineering School are focused on like, how do you find alpha? How do you find like your hidden knowledge? That again, and how you bring that to life is where you win in organic search today. So we call this frontier knowledge. And frontier knowledge is information that an LLM, unlikely your customer, doesn't know or can't yet access, but it answers a user's question. So Ramp, they're one of our customers. They do a great job at it. They have credit card data, they have trends and industries, and they take all of that and they figure out What are the interesting stories we can tell from it? And they create content. And then kind of like our Mr. Pac-Man here, when someone asks a question in ChatGPT, like ChatGPT is trying to prioritize well-researched, authoritative, well-structured content. They then find Ramp. Ramp have success. So again, we think about across GTM functions, how it works. In yesteryears, you might have had many agencies, and you can only actually get through a limited amount of the work. Now, like, one AI-native IC, like everyone here, can make 10x the impact. So that's what we're trying to— we're trying to literally be an extra plate here, or an extra stick, so that you have an extra, uh, arrow to your bow. And yeah, where does Airops come in? So Organic search is well and truly changing. Uh, winning the battle for attention is harder than ever. Now you need to prioritize stuff like Reddit, and how should you run a YouTube strategy? How should I do third-party offsite? How do I— what content do I create? Like, there's so many questions, and not only is it hard to get into the answers, it's hard to prioritize. Um, so our tool end-to-end, you'll largely, largely live in this action layer, which is our workflow builder, but our tool helps you track your prompts. So what are the questions you care about? You can then build workflows that help you generate content. And then at the far side, you measure the impact of your work. One of the big things under the hood, and it's like what we have that's similar to Octave, is the governance layer. I just came from a session where someone asked, hey, how far can I push the boat out on having AI write my content? We never tell people to go end to end, like have something, write a blog post and then publish it sight unseen. Don't do the same with like automated cadences either. Should always maintain a human in the loop. But these are the 4 parts that you will get some exposure to within the Airopps cohort of tracking prompts, building your workflows, building them on top of unique frontier knowledge, and then tracking impact. And then, does it work? Uh, it definitely does if you're the likes of a Webflow. Um, I need to update my numbers here because they shouldn't all be 6%. As of today, Josh Grant told me that they're at 15% of their signups come from AI search. So you can imagine for Webflow, and those are paying signups, that's a huge revenue driver for them. And again, they prioritize moving quick and generating high-quality content. So bringing it back to some demos, the module, uh, overview for the AirOps session is as follows. We talk a lot about it, and it's what Wilson educates many of our biggest customers, prospects on. We talk a lot about content engineering, and we expose people to our team of experts, our methodology, and as Wilson said, like, how to architect, how to think about building a workflow. That's really what we want to leave you with. And then we have our own community as well of people who are building and architecting content systems. But, um, we have just covered a lot about organic search. You don't need to take anything away from the organic search, like, theme or topic from our lesson at all. Like, the, the principles you learn are multi-purpose. So, like, Joe Roo, who runs an agency, turned a lot of what we taught him into deep research, and he pushes that back to his CMS, and then that facilitates him doing better quality, like, outreach to prospects. Sylvain, who kind of comes at it from like a similar way. He wants to develop brand-aligned messaging. He ships via LinkedIn and email cadences. So he detects a trigger in Clay, runs a process to do deep research in AirOps, generates that brand-aligned messaging that's like personalized, and then sends it out, I think via Lemlist to the prospects. And then Mr. Matteo Tittarelli, who is Jared's partner, and uses AirOps for a lot of his social content generation. And that's the workflow I'm going to demo in a few minutes. So you don't need to think of like, hey, this AirOps session is just— I'm going to create content. We will certainly expose you to that. But the principles we want to teach you about very much are applicable to all aspects of GTM engineering. Four key concepts— I just mentioned them a few times— are the following. Prompting. Like, it is wild to look at what some people are able to generate with actually really poor prompts. So we want to push your skill set further on how to think about a system prompt or a user-assistant pair. Like, that stuff, every time I come to this session, people are like blown away by. Then we look at orchestration. So again, the order of operations in which you do things matters a ton. Like, at what point should you do research? What context should you have before you do that research makes a huge difference. Lots of tools including Airops now have copilots. Use them. Like, there's no point in trying to go to all of the primitives all the time. So what we will teach you as part of our course is how to best work with an assistant, what they're good at, what they're not so good at, and then integrations. Again, MCPs is a new concept, um, but just being able to pull the right data and know where it exists, from the right tool and being able to like have a mental model for when to do that, that's incredibly important. And what you'll actually come away with, and we'll go into, I think, two of these today, and we'll show you how to create programmatic pages. And again, principles under the hood there are, can you pull from frontier knowledge? Can you structure it in the right way? Can you ship it to an extra tool? Then social posts, I'll cover that in like 5 minutes, but optimizing content. Again, whether you're optimizing a, a PMM brief or you're optimizing internal enablement documents, and it's again pulling from rich internal context and trying to structure it a certain way. That's what we're teaching. And then deep, deep research like what Joe Roo or Sylvain did. How can you understand your audience better? Take questions they're asking, and turn it back on them as really high-quality outbound messaging. Those are the 4 buckets we're gonna dive into. So demo time. If— well, sorry, let me pause for a second. Any questions across the group? All right, Wilson, I might have you demo first as I grab a sip of water, if that's good with you.

Sure thing, happy to do so.

Perfect. All right, I'm going to pass it over to you.

Alrighty.

Thank you, Owen. Okay, so what I'm going to be showcasing today is a use case that we discuss in our internal cohorts, which is a content refresh workflow. So essentially taking our existing content, as you see here, I have some blogs from our AirOps website and Updating that, refreshing that, because what we've seen is that refresh content, AI loves it. Your content is 3 times more likely to be cited if it is refreshed. So just wanted to show you what this output looks like here on the grid, and I'll dive a little bit more to what that means, what the grid is. But here I'm taking my existing article and I've applied some updates to it. So this is the final article here, and I also have a comparison view, which shows me exactly what changed. And excuse the formatting here. I, uh, this is supposed to be set up in HTML. I accidentally have this set up in Markdown format. And if I show you this output here, it highlights exactly what changed in purple. But diving under the hood, what does this workflow actually do? So looking at this, starting with my inputs, I'm only taking in two things: article URL, right? Taking the context, the content that already exists internally, and I'm also applying a brand gate here, just maintaining our brand voice. Again, keeping that guardrail in terms of maintaining quality. Just a higher-level overview of what this does. I can't go, like, too, too into depth into the technical details, but just wanted to give you guys, like, an idea of, like, the thinking that goes behind building workflows like this. So this webpage scraper takes our content from our website, from our article URL, I'm also pulling in some AOP data here. So essentially I'm looking at, okay, this is my existing URL. How often is this blog being cited in AI? How often is it showing up in AI search? And then I'm piping that data into an LLM step here, essentially having the model analyze my existing content, analyze the data that I've piped in and giving me recommendations on what to update, what to refresh. And then finally, or sorry, not finally, this, the recommendations get piped into a human review step. And this is where the human in the loop comes into play, how we actually maintain and ensure high quality outputs. So I have the option here, and it's not showing it here as I've already kind of run this workflow, but we can actually go in, we can, um, verify that these recommendations make sense. I'm going in, make sure I have human eyes looking over what it plans on changing, and after I've approved it, it goes ahead and applies those updates. Those updates into my article as this final deliverable here. And then I also have a content comparison step, which allows me to take a side-by-side look on what was changed. And then after I've set up this workflow, after I've codified this process, we take it into the grid. And this is where the scale happens. So in the grid here, this middle column is the workflow that we're just looking at. On the left side are inputs, on the right are our outputs. As you can see, I can run this across a whole host of articles. But that is just a quick overview of this particular use case. Um, but I would encourage you guys to kind of stretch your brains in terms of what existing content you might have, whether that be internal, external, how that can be processed, what data we can pull, what intelligence we can pull in, and how that might inform updates and outputs.

Super. While you're still sharing, Wilson, one question came in from Adam around how do pages enter the workflow? Is it number of days since last updated, or what's the trigger to start the workflow? So in this case, Adam, it's just I think Wilson found pages and then, uh, like manually entered them and then could run the workflow. But I think what you're going to go to, Wilson, is maybe like Page360.

Yeah, so I actually pulled these pages from Page360, which includes, um, some data on citations. And if we connect like GA4, GSC, we can also look at data there. And I'm essentially just Taking a look at like what can use improvements. So let's say these URLs, for example, and I could take action on these and add it to that grid that we saw earlier. I can map it over. As you can see, it pulls in those bugs for us to run.

And one really nice thing, if you want to go back, Wilson, is, um, like on this, if you click on like smart filters, So I guess in this instance we don't have GA4 and GSE all connected, but you can start to sub-filter your pages across a number of interesting dimensions. So in this case, it's— I, I can show in my instance in a moment, uh, if maybe go to opportunities on the left because it's going to be like a similar section. So these are like our out-of-the-box functionality. So you can look at for creation, what are your highest impact opportunities? So popular prompt gaps, that's going to mean and where, where are there prompts that you track or keywords that you care about that you and other competitors don't yet have content for? And it'll like help you create that content. Or maybe you care about something refreshing, like that example Wilson had. So declining citations, almost page one. The system here will detect your top opportunities, essentially helping you prioritize. And then once you have workflows built, like Wilson showed in kind of our spreadsheet or grid functionality. You can bring those pages in with their key context and then start to refresh them as you want. The, the one thing I didn't answer in your question, Adam, was are they like automatically brought in? Not in this example, but you can certainly set it up so that if something appears in one of these buckets, you can like send a webhook and add something to a grid. And very doable and a great idea. Where does AROPS surface the AI search data? I can probably show that on my screen to you, Allison. Let me have a look. Here we go. Insights. Unless you have an instance there, Wilson, I can show Uh, here we go. I can just take over screen share. So similar to like a Peek or Scrunch, you can track all of your metrics in, um, in Airops across AI Search. Like where we get the data from is your own first-party tools like GA4 and GSC. And then we track a series of prompts on your behalf as well. So whether you care about like mention rate, share of voice, um, um, citation rate or position, We're able to give you the metrics that you care about. And the really nice thing is we can feed these into your workflows as well. So if we look at like a citation, what might be an example? Top cited, go to page, wrong one. But under the hood, we can figure out for any given like prompt that you care about and how you're performing today. So this one for us, we care about how can I automate parts of the content refresh process? Very long tail. But here we can see Surfer is doing a really good job in comparison to us, um, but we are actually getting cited by Perplexity but not on ChatGPT. So we can start to ask interesting questions of the data, and in the aggregate then it comes up on those couple of visibility charts we showed earlier. Did that answer your question?

Yeah, it did.

Thank you.

Cool. And the workflow I wanted to show off was more about like social posts. We'll take a slightly more advanced version of it in the cohort, but I use this myself for generating a lot of the posts you'll see on my LinkedIn. And like Wilson, it's easiest to explain from kind of the automation focus layer first. And we can get into the workflow. But here I have a grid and this is where I can like provide one-off inputs as I get social post ideas and it'll help me on the right-hand side run that workflow at the end. So in this bucket, maybe I'll take one further up on my screen. I'm gonna select the post type. And when you're scrolling on LinkedIn, there's a bunch of different buckets into which posts can fall, but there's like social engagement, there's like roundups, there's lists, there's like cheat sheets, there's lots of different versions. So I choose one of these, but under the hood, I'll show in a second, it actually means something. I then provide directions. I think of this just as our CMO telling us like what kind of like tone of voice or what direction to take with the post. If you were to talk about OpenClaw, which is what I was trying to do for my example here, or OpenClaw, you could talk about it from a really technical perspective or a non-technical. I'm going to show you a post that is generated for this audience, which is GTM engineers. And so here I said, make it immediately actionable with an easy entry point to get started. I haven't mentioned anything yet about OpenClaw. And here it's not so obvious in my view, but I can feed it a page. So if we had a page on AROPS that talked all about this, I would just feed that in. I wouldn't tell, um, the agent to go do any other research. I just want it to be grounded in the content I'm providing. But if I don't have that, which in this case I don't, I'm going to ask it to go do research, and that will— should answer the question of how should GTM engineers use OpenClaw within their work, particularly towards running cadences in a more semi-automated way. That's going to be my hook, and then I'm going to set, uh, complete web research to yes. So while I show you the demo, hope this works here, um, under the hood what is happening is— let me get rid of this— and we have a number of steps. This is like where the orchestration point comes in, and we have a number of steps before we go and generate any social post. If you were just to go into Claude or ChatGPT, you might say, hey, I want you to write a post on X, here's a paragraph of extra copy, go write it. It'll definitely generate your text, but it might not be all that good. And it might not be all that good because it doesn't know what great looks like. So that in this demo is what I want you to kind of take away of how you can give it better examples of what great looks like. So my very first step is to give it inputs like you saw there. I gave it the direction, I think, to do a social engagement post, like I want one that gets likes and comments and And then under the hood, I'm feeding it my brand kit. I'm telling it a lot about AirOps, what we care about, who we speak to, the directions. I said make it approachable and provide an easy entry point. And then I can give it a URL or I can have it go do its own independent research. In our case, we picked the latter. Then here, I'm having it find my templates. So this is actually one of the most impactful things. If I do duplicate— as I see a question come in. Um, let me go back. Oops, that's not what I want. Uh, why can't I do this here? Here we go. So I'm gonna go to my knowledge base. And here I have LinkedIn post templates. So happy to share this with everyone after the fact, but I created a spreadsheet as you've seen here that contains a tagged content type. So social engagement, whatever, roundup. It has context to suggest like what this prompt achieves. So it's like, promise people you'll give them something useful in the comments and then give it. And then what is really important is to like give it examples. So this should follow this format. It should be like, X, drop whatever URL below, and I'll reply with 3 ways you're not doing Y. And then it has a full example here of like what great looks like. Again, I can invest a lot more time in giving it more and more examples, but you'll see I kind of have a different way of doing it in the workflow, but dependent on what you care about, the best thing you can do to increase the quality of your work is give the agents really good examples. So that is all I'm doing in this first part. I'm saying go to that knowledge base, pull back up to 5 examples of the post type I'm feeding in. Then I just need to format it in a certain way. Did I write this Python code? Absolutely not. I used our best friend called Copilot over here. And said, hey, take all the inputs of step 6, format it in just like a readable method like we'll see here, and that just lets me pass it into the next section. Um, so we format, and then here I'm just making a division. In short, have I passed in a page URL or not? If I do the page URL, I don't need to go do external research. I want it to be confined to the context I give it. But, um, in that case it will scrape the page. And then you might say, Owen, you're dumb, you're getting it to write the post 3 times. Well, I'll tell you that hopefully I'm not dumb, and I have a reason for doing so. And my reason is that I am using what I call here user assistant pairs. And this is what like ChatGPT will not show you in just like your user interface. It's kind of encoded below the hood. But, um, kind of two different sections. One is your system prompt, um, and again, we give you access to that, but this is where you describe it, the role you want it to take on. So I'm saying here, hey, you're a LinkedIn post writer for AirOps. Here's information on our brand, and you basically generate like the type of content that gets engagement. So independent of whatever I pass in, whether it's talking about education or OpenClaw, that's going to remain true. And then it's objective. And again, did I write this prompt? No, I wrote a good one and then I had Copilot improve on it, which again, you can use ChatGPT or Claude to do the same for you. I give it instructions and I tell it everything I needed to know about my brand. And this is where your governance layer comes into effect. So this is just me telling it something. That's the user message. I also tell it more. I'm saying like, hey, now go write a post on the following idea. But you will notice that this is hardcoded. That is intended. This is where the notion of a user-assistant pair comes in. So within each of these modals, and this will be true if you're in like a Cargo or a Clay, this agent operates in its own vacuum. It's only going to get access to content that you give it. And what we have found when people are writing really long content, and in this case it's simpler because it's shorter, you want to provide it with really good examples in that closed loop bubble. So here I'm saying like, go write a post on this, and then I actually give it what great looks like to me for that example. So this is what I would consider great for its request of generating this post. And then I actually give it another example. I'm like, great, now that you've done that, now let's take another example and now generate a new one. And only after I've kind of fed it two examples where it's like, it already has my tone of voice and knows about my brand, it now knows what its task is, because I've told it in the system prompt. And I've given it two really good examples. Now I'm saying, go write it. And this is where I'm using my inputs, which is from the very top, what it'll take in from our grid. It's taking, um, content from the scraped page, the directions, and the templates, and then it's going to give me my output. Um, so it's going through extra work, it does consume more tokens, but your, your output quality is higher. And maybe what I've done here is overkill, but it gives me quality and it gives me more options to choose from. I essentially repeat the same step. I need to upgrade, um, I do everything the same up to here, and then I say, actually, we've already done one already. What I want you to now do is redo it again but use a different template. So I'm kind of just setting up the same context and then getting it to draft it again, and that does, um, make my second one better. And I just do it a third time to be absolutely sure of my quality, and then I can pass it out. And I know we're coming close to time, but let me just show you this other side of the branch. So if I don't give it a URL, it should go and do research, and I use Perplexity Sonar for that. It's going to be vast and varied with the context it gathers and like aggregates for me. And then on the right-hand side here, I do the exact same thing. I, uh, get it to write the first one, write another variation, and then a final variation. It's going to format them so I can pass it back in JSON format. And then I really should update this to nano banana, but it's going to extract a keyword from what I asked, and it's going to look in stock images to see if there's anything relevant. I don't know what it'll find for OpenClaw. It'll be interesting to see. And then, um, it will provide that as output as well. Uh, Adam has a question: where is the knowledge base stored, i.e. Tone of voice, you reference it in the workflow. Good question. Um, so there's two main kind of, uh, governance layers in AirOps. One is a knowledge base, and that's where we store individual pieces of context. It could be gone calls, could be competitor battle cards. And then we have brand kits, and then this is where— I guess I have an Airbnb example— you can make these as detailed as you want. You just put in your brand URL will auto-generate one out of the box for you. So we'll give you about the brand, ICP, who your competitors are, brand point of view. If you're someone like a customer, uh, like of ours, like Carta, they're gonna have so much context in here, and then they're gonna break it out into more custom variables. How do I pitch X product to Y persona? And then they can pull all of those fragments into their workflow. Um, so this, this is like similar to what Octave does, but it's just all housed within AirOps that you don't need to like pass it between tools. Um, so that is it. If I go back to my example, you'll see here I have a review. So this I think is where I need to select an image, so I'll click open and These are all from our cheapest stock image provider, so I'm just gonna pick this one and accept output. And what it should do is write my posts into the grid. And it takes a second. So here I should be able to get them in markdown format. Just make this over to one side. So this is like a longer form, this like social engagement one. And it's basically got the hook right. Like, that's all that will appear above the fold in LinkedIn. You then need to like click see more to go and see it. And it's giving you like, maybe this is a little bit long, so I might trim it back if I was to use this one, but it's giving you like actual actionable steps to go and get Open Claw set up on your desktop. And if I jump to the third one, Maybe even get something actionable is tough because you need like a dedicated, uh, probably like Mac processor for it. But yeah, it's getting like the hook right. It's doing its best job of like answering the question of like easy access. It's probably like saying build a rocket ship to go to Mars in a social post, which is difficult to achieve, but it's doing its best effort. And then it's making sure that the, the call to action is clear. So different formats, This is what I picked up the get-go. So, uh, input— inputs mean outputs. But let me pause there in case people have questions on this or other contexts that we went through. Maybe not. Cool. I don't know, Jared, if there's anything else from your side you wanted to wrap up with or cover.

Um, you want to talk just a minute or two about what kinds of exercises, what kinds of sort of async homework they're going to be getting in, in the course?

Yes, um, I don't have a slide for it, but what we'll cover is a couple of the questions like Adam probed us on. So setting up your Brand Kit, so you'll have time to go internally, maybe ask your content marketer, like what's your style guide, what's your tone of voice. And if not, you can just generate it out of the box with AirOps or drop into one of our workflows a couple of sample posts. We can improve on it for you. So Brand Kit is one. And the other thing we'll ask you to do is think about the, the type of content you want to generate. So maybe it's emails, maybe you want to help out the SEO team. So think of like what great looks like in your eyes. Maybe it's a post from an existing page you have on your domain. Maybe it's something from like someone else that you want to strive for. So get a couple of really good examples of what great looks like. And then last but not least, um, gather up some, um, like rich context. So maybe it's a couple of Gong calls, maybe it's, I don't know, scraping Reddit if that's like part of another exercise you're doing. So see if there's other parts of courses that come before Airob that get you into your internal context. And then just make sure you have like 3 to 5 examples, um, like 3 sales calls, 3 scrapes from other pages that give, um, a representation of like your internal context that's valuable. So that'll give you great context for your knowledge base and brand kit. And, um, I think we have 1 or 2 sample exercises just to get you used to navigating from the grid, which is what the spreadsheet functionality Wilson and I showed is, as well as, um, in the Workflow Builder. We'll get you to build 2 or 3 very simple 2-step workflows so that you just get used to kind of like dragging, dropping, and interacting with data.

Awesome. Okay, is, uh, creating content, creating social posts, updating, refreshing content, and, uh, and ranking for the LLMs? Is that something people here are interested and keen to learn or struggling with today?

I'll take silences. If they're not struggling, someone on their team is probably struggling. Yes, exactly.

So I mean, we talked here about social posts, which I think is interesting, but for me it's also interesting to see it from a website. Like, how do you get your website pages to rank on LLM and How do you, like, let's say from a thought leadership perspective, let's say you wanted to build your thought leadership in a space. That's kind of the bigger question. And then it's like, well, how do you go about it? How do you know where you're at? Like, that is my broader question that I'd love to learn more about.

Yeah. Um, we, I could probably spend 2 hours speaking about it. But there's many different avenues through which you can answer that question. So one is if you wanted to bring your internal knowledge out as social posts, maybe you're like, CEO cares about that. And it's actually a really good means of getting cited. So when we think about a question in ChatGPT, there's two parts to care about. There's, are you mentioned in the text that is returned? Let's call it a mention. Or is your content scraped? and then has a little link back to your website, that's called a citation. So there's two parts there. But if we think about all of those together, all the mentions, all the sites or citations, and 85% of all that visibility does not live on your website. So when people ask questions about whatever company you're at, about 8 to 9 in 10 of what is What the agent is pulling from is from other websites like Reddit or YouTube or LinkedIn. And many of them are from pages that, and domains that you'll never have heard about. So offsite is a hugely important factor to get mentioned in the answers. So narrowing it back to your question on like LinkedIn, the best thing you can do is again, take your internal context. Push it out through a reputable person on your team, like one of your co-founders, or like, uh, maybe someone who's like known in the industry. And then where you're getting the right content and context into that post that is answering someone's question in ChatGPT or Perplexity, your post is actually going to get cited. And that's like something that we're doing internally a lot more, where all of us on the growth team are posting a lot more around the how to do X or why to do Y style questions, and it's really driving our visibility up. So just from a LinkedIn perspective, that's what you can do. But a lot of the same tactics are true for Wilson's example where it's take that same context. It's rather than the format being a LinkedIn post, you wanna make sure that it comes onto your longer form content. And then there is more like structural depth with which we will teach you like what's important, like heading structure, metadata. And just like explaining concepts like I'm 5 is much better than trying to convey it at a PhD level. So there's a lot of, um, under the hood things that if you want more detail, hit us up. And we might not cover all of it in the GTM Engineering School, but we'll definitely set you up with some best practices.

Awesome, thank you.

Absolutely. Uh, that's a good question from Pierre. Basically, um, I showed that within the workflow I can give examples of great, but his question was, are you then able to look at like metrics and feed that back in? So right now my workflow does not. It is very possible to do that though. Um, I could have a workflow that pulls my last 30 days worth of posts or pulls the last 30 days worth of anyone's posts and then use that to remove examples from my spreadsheet or add additional ones. So that could be like more of an evergreen or updated asset. Actually don't do that, I probably should. So, um, yeah, essentially very doable. Thank you. Cool. Any last questions?

Okay, sounds like we got, we got all the questions answered here.

I actually have, I actually have one last slide here, Jared, because I know people, people do not come to the cohort. We have some stuff as well, but, and where you come you'll get full access to like our platform. That's like $2K a month. We'll give you plenty of tasks so that you have loads of ceiling to go build to your heart's delight. And then you'll learn from the best. Wilson's face isn't here. It should be. He's right up the top. He's teaching people at Kayak and Ramp. And I think you mentioned like Verizon earlier. They're all learning Airops at the moment. And then I just think this is key. And if you're kind of like on the fence of like, do I commit? The, uh, GTM Engineering School or not. Uh, this is from 6 months ago, and, uh, this is in my last deck when I presented here. I thought it was so funny, I wanted to keep it in. And Dharmesh, who's like founder at HubSpot, he was like, 6 months ago, he was like, today it's about like bring your own device, and in the future it's about bring your own agents. People will develop agents at whatever company they're at, and they'll like bring it with them. And that like seemed from this post like actually years away. I almost feel like we're gone past that point now. I have some people joining the team and they're like, hey, like, I have all of these setups, I'm just gonna like install them on day one, we're just gonna like go off to the races. And the velocity progress is insane. The, the more time you can commit to this stuff, the better. And again, like me, you should have like super high anxiety that you're falling behind. I think the fact that you showed up today is just like a credit to yourself, and it means that you're probably ahead of 99 So just, I think that's worth like recognizing.

That, that, that's a great way to, to end this. That post is, uh, mind-blowing. And yes, I, you could see the same thing that is happening. It's incredible.

Yeah.

Gosh, well, I'm gonna lose sleep tonight.

Yeah, you gotta get your sleep. That's all I'll say.

Cool. All right, thanks Owen, thanks Wilson, thanks everyone for your questions, for showing up, and we'll see you in class next week.

All right, happy first week folks, we'll see you later.

Goodbye.

Thank you everyone.