How I Build that [Agent] - 7 Marketers Show AI Workflows That Save Hours

Hey, it's me Dave. Today's episode is brought to you by Optimizely. Look, I hear about AI agents every single day on the internet in the X5 community, but it's confusing. It's a little bit overwhelming, which platform should I be using? Which agents are actually worth building? How do I even set this up? There's a lot of noise out there. It's kind of hard to tell what's actually going to make the job easier, versus what just can be another thing to manage. Or is this just a cool little thing some guy wrote about on LinkedIn? Well, Optimizely's agent platform is built specifically for B2B marketers. You can build your own agents or pull from a directory of 50 plus pre-built ones made just for marketing use cases all in one place. This is really smart. They also just launch virtual teammates. These are digital members of the team that can learn over time and can join meetings, complete tasks, support campaigns, and keep your whole website Optimizely. Humans, that's us, of course. We still get the final say over these robots, but this seems really useful. This is worth leaning into right now, right? Agents have so much potential and using AI in a platform like Optimizely for the stuff that each year week, so you can get more time for the strategic thinking and actual fun parts of your job on into that. Let's make marketing about the craft of marketing again. I think there's something there. So go and check out Optimizely right now. See how their AI agents can help you with your B2B marketing operations. They purposely built this for B2B marketing. This is really cool. Optimizely.com/exit5. That's Optimizely.com/exit5. Go and check out their agents. Hey, it's me, Dave. Today's episode is brought to you by Webflow. Webflow is a website platform for the agentic web built for modern marketing teams to build fully custom sites that perform in AI search. No developer needed. Your website isn't just your home page anymore. It's actually the first thing an AI agent is going to read about your company and from there, decide whether or not to recommend you to a potential customer. Before a demo, before a sales call, before anyone on your team even knows that buyer exists, which means if your site isn't structured to show up in AI search, you're not in the buying conversation. It's as simple as that. Webflow is the platform marketing teams used to fix that. You can build a fully custom site without a developer, optimize it for AI search, govern your content, and actually see how agents are reading and interpreting it all in one place. Make your website your biggest growth engine right now with Webflow. Check them out at Webflow.com/4/exit5. That's Webflow.com/FOR/exit5. Welcome everybody. This is another one of our Ex-A-Five live sessions. We do these twice a month. Our whole thing here with Ex-A-Five is to help you separate the signal from the noise and marketing. There's a lot more noise than ever right now with AI and everyone talking about what's really happening and what they're doing. But we bring the real because we focus on actual real live human marketers who have to do this for a living for their jobs. Today, this topic is all about, I love this title, How I Built That Agent. We have seven marketers here today. Every one of them has built an AI agent that they're actually using. This is not a hypothetical. These are actual in-use agents that are doing things that these companies, not a demo, not a concept. This is a thing that actually runs. I'm joined by Dasha here as I mentioned, my co-hosts, and it's going to be an awesome session action-packed. I hope you're going to fill your notebook with a bunch of ideas that you can actually take back with you today. But before we get into this session, I want to give a quick shout out to our sponsor, the crew at Optimizely. Great people. I know they just had an event of their own. I hear about AI agents every single day on LinkedIn and in the Ex-A-Five community, which platform should I be using? How do I do this? How do I set this up? There's a lot of noise out there and it's hard to tell who's actually going to make your job easier versus just another tool that you have to manage. This is why Optimizely has their agent platform. It's built specifically for B2B marketers. You can build your own agents or pull from a directory of 50 plus pre-built ones made just for marketing use cases. I love that all in one place. There's a lot of nuance in what we do as a job. These are digital members of the team that learn over time, can join your meeting as complete your tasks, support campaigns, keep your website optimized. Heck, where's my guy Richard? Maybe do drop off for the kids before this? Humans, of course, will still have the final say on everything that ships, but that's how virtual teammates works. This is something worth leaning into right now using AI in a platform like Optimizely for the stuff that each year weeks. So you get more time for strategic thinking and the actual fun parts of marketing, which is I think why we all got into this job in the first place. Check it out. What's the format for today? Give everybody the rundown please. I'm so glad you asked. This is exactly how today is going to work. Today is a real battle. It is a competition of humans and the agents they have built. You are going to see two groups with a timer. Okay, this is going to get really intense. The first group is going to have four marketers and four agents. Then we're going to vote, which marketer an agent did you like best, which agent would be most helpful for your team. Then we're going to go into group two, two minutes. Again, each presenter gets two minutes to demo their agent vote for the one that you want to see more of. Because in the finals, you are going to see the two marketers that won on their two-minute pitch plus a third wild card. Then we're going to vote for the ultimate winner. If you have questions as we go, which you probably will, please use the chat and someone likely the human presenter is going to answer them in the chat or in the community later. Good job, Dasha. You're going to take this. Give us a little definition. Before let's set the table, please. What is an AI agent? I don't know. Yeah, what is an agent? It's a good question. First, I'm going to review this definition that we pulled from Google Cloud. Then I'm going to give you my own human definition. AI agents are software systems that use AI to pursue goals and complete tasks on behalf of users. What does that mean? A metaphor that I like to use is if chatting with AI like cloud or chat GPT or GROC whatever you're using is like chatting with a smart consultant, then an agent is like handing the consultant your computer access to all of your systems and then just asking that consultant to go do work on your behalf. So that's two definitions. Take it or leave it. That's how I would define an agent. Yeah, you know, I'm going to make a boomer joke. Even though I'm not a boomer, I'm at back in my day, you know, we call this Dasha. We call this an intern. It's a smart intern. Yes, but the intern never sleeps. It never goes on vacation. It does not need washroom breaks and you probably pay it a lot less, but it consumes way more water. Yeah. It's always at the water cooler. And now, and look, we, so we did something cool. We surveyed. Thanks to our guy Ben Reed and the growth team here at exit five. We did a survey because I think it's important to get a real look at what's going on here. Real quick, you know, we got a thousand people to respond to this and here's the headline for me. Despite what you read on LinkedIn, 58% of you actually have either not built an agent or have tried but gotten stuck. Okay. So if you're feeling like you're behind, you're not. You're not alone in this. That's exactly why we're doing this. I thought this was interesting. And then we got one more about a question here about what stopped you from using agents more? Dasha. Yeah. What the data showed is that 75% of marketers just didn't know where to start or didn't have time to learn. Which great thing you're here. You're in the room. We can see you. This is an hour where you're going to learn all about agents. So my recommendation, close lock, close your email, actually commit yourself to this hour so that you can have that notebook of notes and go build your first agent or maybe your second ethic agent. I love that. I say that at our at our events. Someone goes to our event. They're on their phone their whole time that that's the first person who's going to go to link. I went to exit five's event and it sucked. I'm like, well, if you're here and you're paying attention, you're going to get some value out of this. All right. Let's get into the lineup. We got a bunch of different examples and AI chief staff agents that find your next big customer. How to do one-on-ones. The slides are going everywhere right now. And, you know, after we go through all this lineup, this is not just for fun and games. Let me get the prizes slide. Let's hit the next one here because this is for cloud, right? Not only are you getting your name, we physically mail you a hand made real trophy with your name on it. But you could feature it in my newsletter, which is a big deal. Ask Shalom O'Humany LinkedIn messages. He got this week for being in that as part of one of these. And I'm ready. Enough of the small talk. Enough of the housekeeping. Please. Can we bring up Sam and can we get this thing off to the races here? Let's get it. So, I'm Sam. Let me walk you all through the agent that I built. I want to give a quick shout out to Hillary Gridley who built this website called 5Couch the 5K for AI. And this is where I took the inspiration. But the problem that I'm really solving here is that every time I start my day, I have to go through all my email, my previous slacks, think about all the meetings that I've been in. And I'm sort of sitting there for a good, sometimes 45 minutes, sometimes an hour, usually about half an hour every single day thinking about, okay, what are my priorities for that day? What do I need to take care of? So, what I built using the couch to 5K's inspiration is a chief of staff. The chief of staff learns about everybody I work with. It looks at my calendar, it looks at all my slacks, my emails, it also listens and reads all the transcripts from all my meetings. And it says, Sam, this is what you need to focus on. This is what you need to work on. Now, it's not perfect. It doesn't always catch everything. So, I do talk to it. I talk to exactly the same way I would an assistant. And then it also helps me with prioritization. So, it's reminding me, hey, you've got a couple things that are due today. And I actually talk to this agent. But for the sake of the demo, I'm not going to talk to the agent. I'm going to just say like, hey, give me a few times to put my head down today. What it's going to do is it's going to look at my calendar. It's going to tell me this is what I recommend. And then I'm through this agent. I can actually book out my calendar. So, got 30 seconds left. What I want to use you all with is this is a smart system. It's actually learning behind the scenes. If you vote for me, I will share with you how I built this with cloud code, how it learns, and some of the systems that I put behind it. Nice. I love a personal use case. How can I use these things to make my day in life better? There it is. The tune officially branded zoom outro tune. All right. It's good. I'm doing a lot right here. I'm going to get array. I'm going to give myself a raise after this. I'm doing the timer, man. I'm in the chat. There's a lot happening right now. Is that Alan? I can see next. It's a promoted day of this time to be our timer guy. Congratulations, Dave. You're doing such a great job. Our next presenter is Alan. He's going to show us an agent that's going to find your next big customer before your competitors do. It's up. I am Alan. I am director of marketing at Recharge, former head of marketing at SKIO, which was acquired by Recharge, and today I'm going to talk to you about agentic cool hunting. If you're unfamiliar with the term cool hunting as an idea coined in the late 90s, where brands would pay scouts with great taste to go into a city and look at what cool kids were doing and wearing and then give that information to corporate overlords so they could sell it to people in now dead mall stores and MTV. I love this idea and I wanted to replicate it at work with robots. So I kind of did. We're unfamiliar with the Shopify ecosystem. There's like three million stores on Shopify. About 52,000 of them do more than a million annual revenue and everyone who sells software in this space cares about the top 10% of that 52,000. Everyone selling software is using the same tools for signals, usually something like store leads and store leads is fine. It is directly accurate for established brands, but it sucks at uncovering interesting new brands like grooms in 2023 store leads would not have told you anything about grooms and a few months ago they sold for $1.2 billion. How do we find these brands before anybody else? The old shitty version was a Slack channel where people would dump Twitter links and they get a lot of rocket fire that thumbs up emoji and then no action was taken. The new way is my cool hunting agent. Every Tuesday, it scours newsletters and next accounts and podcasts and CPG directories. It looks for cool new brands. It writes a dossier on why they could pop off. It serves it to my BDRs. This week it found 46. We are actively working 29 of them. And if you vote me through, I will show you how to find your next big customer too. Yeah, dude. I love that. Great hook. I'm not going to buy us anyone, but great hook there. I think we need something like that like a news jacking agent. Okay. We're off to a good start. Vibes are high. Next, Maddie's senior content marketing manager at digital ocean, 13 clawed skills that can edit a draft before I do. I'm into that. Let's go, Maddie. All right. So unlike some of the other agents you're going to see today, this one is really bare bones. It's not very flashy. It's really about what it actually does at the end of the day. And so I built some agents in Zapier and they then, you know, there's a lot of different ways that you can approach the process. But this one is one that you can build yourself and clawed without being a technical marketer or coder. So it's a series of clawed skills. They're chained together to work as my content editor. And just to clear up the air, I definitely also edit things after running this. That's an important human touch that just like can't go away no matter how many ways that you try to work around it. So when an editor turns in a draft, when I write a draft, this workflow runs it through everything that I used to check by hand is like the first step. So it's like, did it cover the brief? Does it match our style guide? Does it, I'm brand? Is it built to actually get found? And so the all-envisibility part matters a lot to us right now at digital ocean. And most of what I'm personally writing are like competitor comparisons and alternatives pages. And so it's a type of content where a wrong claim is a real risk. And the proof that it's working, these are some selected results from some of those alternatives pages that I've been creating. And using this workflow is like that first editing chain. So yeah, top profound citations in the track prompts and top organic results within 24 to 48 hours of publishing. So it definitely works. But I haven't told you the hard part yet, which is how you get a dozen plus editor skills to work together instead of stepping on each other and the messy middle that I work through to get there. So that's round two. If you vote me through, then I'll show you how that works. All right, we've got our final presenter in this cohort. And then we're going to vote. So get your thumb or index finger or whatever you're going to do to vote because it's almost that time. Hey, everybody. It's really nice to meet you. I'm Richard. I am just so excited to be here. It's my second time on. And the first time was just so fun that I'm really glad to be here. I don't have any funny anecdotes this time. But I will say is that I'm very, very excited to show you a suite of LinkedIn ad agents that takes all the busy and guesswork out of managing LinkedIn ads. I spent over $2 million managing LinkedIn ads. And I'll tell you what, I've wasted more ad spend that would care to admit. I think even the best media buyers that can run on LinkedIn would tell you that it's so easy to waste them. It's easy to do it in not making manual bit adjustments quick enough. It's easy to get drift in your audience targeting. One of the most prolific ways to waste spend, however, is in my opinion, having poor and declining creative. Just add channels to tell you that up to 80% of your ad performance is determined by the quality of the creative that you run. It doesn't take as much into consideration in terms of targeting perfection or bid perfection. But rather, what are you doing to speak to your target audience? So my agent runs both headlessly and also through a design system on a living web page. It will go through and detect automatically creative reviews that fall within a declining cycle. If a creative shows declining within a time period of four to six of those reviews, it will flag as being needed for view. So that way, you're not jumping to fix something. It's not actually broken. And that way, you're also not actually waiting to long to correct something. Then these are fake ads. It's a devastating environment, but this is live for actual clients of mine. You can go ahead and get insights as to what's going poorly with the creative, give it instructions as to pre-approved messaging guidelines as to how you can go ahead and fix it, give additional context to guidance, and generate a creative prompt that's ready to drop into any LLM of your choice. Vote for me. Good job, Richard. Just being a dad, battling the elements out there, still able to come in and do a good job. All right. So my runner show says fill right now because we're going to roll the poll somehow we have to vote already. And look, I saw the question in the chat, aren't we going to see the AI agent? Did they got two minutes, man, it's like two slides are going to give you an overview. And then we're going to vote which one you wanted to learn from now. This is not who's was the best. This is which workflow are you actually going to try because then we're going to bring them back up here. Now, and Dasha, obviously, we can't get to all these. So we do have a more a good call to action here, which is what? Which is you should join the community. If you are like, wow, these speakers are friggin smart. Well, guess what? They hang out in the exit back community. So you should join the community and then you can ask them more in depth questions. I have been able to reach out to so many people through the exit five community and just grab one on one time. It's like if you graduate and like you're all the matter supports you, it's the same kind of feeling in there. Like you're like all part of exit five. What would that be like? I have no idea. So do we have a winner here? Looks like a mat. I'm not a big math guy, but it looks like Maddie 112 of the 332 possible votes. That's pretty awesome. Maddie 13 clawed skills to edit your first draft. We're going to see more, but it's close. Alan Finn chugging along. We don't know. There is a wild card. So we got one from this round, one from the next round and then the third amount of votes. So keep voting. This does matter. And that's it. Let's end the poll. We're back up. Play number five. We're back on the wagon here. This is. I'm sweating. I got the air on here. This is a tight old vintage shirt. Smells really weird, but I'm wearing it today because I think it's cool and. You got some compliments on the shirt in the chat. Did you see that? I know. That's okay. Maybe feel good. Thank you. All right. I'm about to start the timer. Little drum roll. Okay. So. Hi. I'm sure on. I lead marketing. And so unlike a lot of other agents or skills you're probably seeing online, whatever is. My team or I build actually has to work and it has to work in production because at the end of the day, I'm responsible for revenues. So I care about getting shit done. One big problem that we have as an organization is generally that I am right now a bottleneck, right? Was a bottleneck between sales customer success and also our other teams. So basically, if you can imagine, we have a horizontal platform. There's a million personas use cases industries. There's a lot of context to have to sort through. And basically, this informs a lot of the different things that we're doing, whether it's like marketing campaigns, outbound sequences, partnership, collateral, and so on. And so what I wanted to do was unstick myself, get out of it. So I don't have to be the bottleneck across all of this, but also allow this to be something reliable. So I've been building my own internal tools and systems for years. I've been working at automation platforms. And I know that like one of the biggest reasons that things are not working in production is primarily because the data is not right. And a lot of people don't know how to put those things into production. So what I essentially built was a way to capture every single sales call from Fathom, which is where all of our data was living. So I was the only one going into Fathom and I cleaned and structured it within our own software database and then built an agent within Cloud that actually is able to extract clear and precise quotes, actual trends, actual reliable data. And then we have built agents that actually work on top of that. So this is landing page agents, thought leadership ads, and also outbound messages. and so on. And what's super cool is we have 12 members using it. Adults it in less than a week. I think one important thing that I also forgot to mention is that we pre-extract before we actually have it going into an agent. So if you vote for me, I'll show you how it works. Nice job. Customer research is hard. Doing this in two minutes is also hard. Sure, I'm good job. We're happy to have you. These are also good. I want like an in-depth newsletter. Chris. Chris is up next. Look, I don't know. This is such a niche thing, but Chris is high off of a big, big reshare from Darmash earlier last week that I saw. I must have just skyrocketed his LinkedIn. Now he's going to show us a hundred buyers reacted to this homepage in 97 seconds. All right. Good hook. Show me. We all know that customer research is hard. And even if AI gives us limited options, we still use to do our customer research. Some of the challenges that I'm seeing with my clients and B2B teams are limited resources or limited access to customers or limited traffic or sometimes research takes ages. If you do with a consultancy, sometimes we are not able to go as deep as we want or sometimes it's one and done and it can be consistent or regular or it's also hard to scale well and easily. So how do we get that outside perspective that we need to actually be able to launch messaging that works? This is the agent workflow that I built. So first we give the agent a brain which contains all our research data, anything that we already have, positioning messaging strategy, any decisions that we made, any gaps that we know we need to fill. Second is the agent access the brain and uses it to actually either produce copy if it doesn't have an existing copy that we feed. If we have existing copy, we can feed it to the agent. Third step is we test our messaging with AI personas. So the way that I define AI personas is not just asking such GPT to A, B, my SAP. It's I augment AI personas with real human data and we give this AI personas the same message, the same questions we divided into fit personas, influencer personas and even non-fit buyers so that we can get as many perspectives as possible. The final step is we turn these responses into actual usable insights. So the agent gets the exact responses, groups them into patterns. We find what work, what was confusing, anything that we can test after and the human in the loop actually makes a decision. One thing that I always say is when people say AI writes slop copy, I always tell them first use research, use AI to understand your customers, then write copy with AI. If you vote for me, I'll show an example on exit five homepage. Good idea. I got beat with the AI stick this week trying to work on my slides for for drive. I'm going to tell Dash all about it in the future, but there's there's a learning there Chris. All right. We got one more. Got McKeele. Yeah, let's do it. Okay, so I run optimizel.com basically that's the easiest way to think about it. We have CMS experimentation, a bunch of different things. We are B2B company. We work with thousands of other companies and one of the hardest things that we wanted to do was account based marketing. And if you're familiar with doing a account based marketing, you might do like one to many, one to few one to one program, but it costs a lot of time and resources to do like true one to one personalization. And it's hard to scale. So I basically created these couple agents that are running off of our ICP database and Salesforce. You can see here, I built a little search engine for it. You can see it's running for like 2600 companies right now. And you can see the agent actually created the landing page base on a bunch of different data from Salesforce built with six cents, all the call transcripts that we've had with this company. And it's essentially built out this like custom landing page experience kind of talking about what they care about and interestingly enough, like companies will literally put what they're investing in and their open job wrecks on their site, which is very fun. And yeah, you can see here it's actually citing and sourcing stuff from our CMS as well, taking a screenshot of their site. And then my favorite part is because I know who I'm reaching out to. This is outbound, right? I don't even need to show them a form. So this is my favorite part I just want to click on the lasers. But yeah, we bring in a bunch of lasers essentially to in immediately paying a person on teams internally letting them know that this company's interested in outreach. And then we can immediately book a meeting and as you can see, we immediately know who the owner is internally. So that makes it super powerful. And yeah, we've run this I think over 14,000 times at this point. And the agent has only had like, I think it's 120 was last I counted. So I'd love to show you like how we how we kind of pulled all of this off. But yeah, it's kind of a cool experience that we built. Hey, it's me Dave. This episode is sponsored by Zoom webinars. Zoom webinars lets you host polished engaging webinars that actually drive pipeline. We recently switched to Zoom webinars for exit five live. I don't know if you've seen this. It's been everywhere. I think our conversations about webinars, everyone's talking about webinars right now. Let's do great webinars. That's what we're talking about, right? I've been super happy since we moved over to Zoom just really the high quality. It just some it feels different. Our webinars feel professional and polished. Our media team shout out to Marissa and Rhea. They've been putting a ton of work in behind the scenes to get this thing right. And it really helps us run these exit five live sessions and put on a performance. This is not a webinar. It actually feels like our performance and the results have been there are attendees and our team has been really happy. The platform gives you everything you need to run a real event from branded registration built in promotional tools, live engagement features and post event follow up that helps you convert attendees into pipeline. That's the difference between hosting a meeting and running a great event. If you're a BDB marketer with a webinar program that needs a reboot or you're thinking about building one, you should go and check out Zoom webinars, right? You know Zoom but go and check out the webinar product and come to one of our upcoming exit five live sessions every other Friday. It's two Fridays a month. You can see it in action. Check them out. Check our sessions out. I think it really gives a good example of what you can do. Go and check them out. Zoom.com/exitfive. That's zoom.com/exitfive. Hey, it's Dave. This episode is brought to you by compound growth marketing. They're a full, funneled, demand-gen agency. Look, I personally hired them twice. I think I was the very first customer before I was even a thought leader just when I was a VP of marketing. A real marketer back in the day, right? CGM was one of the best agencies that I've ever hired. I think that's because they understand the job. The founder John came from this background, knows what it's like, knows the life of a marketing leader. They help high growth, cybersecurity, and enterprise software companies show up earlier in the buying journey where potential customers are forming opinions about which product to use. They offer the combination of AEO, modern paid advertising strategy, and a dedicated go-to-market engineering team. Everything CGM does gets tracked, measured, and improved over time. That means more pipeline for you. This works because they were started by a former VP of marketing. Like I said, John, he really gets this space. I think that matters. By the way, for B2B tech companies, CGM has a special offer for people who listen to this podcast. You can get two free sessions focused on media planning to learn what channels are driving the best ROI, how much you should be investing in demand, capture, versus creation, and how to invest up funnel effectively. Hey, this is a really smart offer. Go check it out. Compoundgrowthmarketing.com/exit5. Go check this out. Get some free help. That's compoundgrowthmarketing.com/exit5. Bye-bye. Okay, so that's the first round. I want to back something up. I want to show you, I'm a man of action here. Last week or two weeks ago, we did this little shout-out thing where I said, "Hey, man, there's a lot of podcast fans, a lot of people who listen to our show here on Exit 5 Live. Can you go and post on LinkedIn? Take a screenshot about it. Dasha, I'm supposed to shout someone out here, but I absolutely cannot see. It's in tiny micro-and-size font, so you got me. All right. Tyler Howard, thank you so much for being a listener of the show. Anna Bradshaw, you're amazing. Thank you for posting about it with your LinkedIn followers. So, Jana Baluraman, thank you so much for listening to the show. I really appreciate your post about it. Meredith, Meredith, shout out to you, Meredith Paps. I like that last name. That's a fun last name. And last but certainly not least, Bethany West McCarter, thank you for listening to the show. Thank you for the lovely screenshot. We appreciate you. We see you. Yeah, appreciate that. And just real people shouting out real people and carrying pierce in the chat is, "Are we really going to have time to go through the presentations?" Well, yeah, girl, we got 30 minutes left. We got a whole plan here. This is a real production. You'll see, if you don't like it, you can leave me a review after. But shout out to all you, but we got a surprise guest. This just in. Oh, now we're throwing a merissa says pole, pole, pole. Are we doing the pole, pole, pole? Are we doing surprise guests? We're going to do both. We're going to do the pole. We got a pole here. So of the second round, of those three that you just saw, who are you going to vote to the next round? Who are you going to vote? You know, who's going to be joining Maddie out there? Maybe Alan Finn is going to be the wild card from the first round. Maybe one of these. Wow. Overwhelmingly, Mr. Lasers, I think it was the Lasers, honestly, that did it for us. How do I can't say his name? Can you say? Michiel Michiel. Michiel. Yeah, there you go. Michiel. Michiel. I'm a Jew. I have that. I have that. I have that. I got that. So Michiel not required for this name, not required. It's certainly not required. No problem. We welcome all here. Love this. Okay. Overwhelming amount of votes. And then do we have the. Hey, okay, producers of my year were rolling. Surprise guests. Bring them up. Send them up. Get the surprise guests on stage. What do you know? Ladies and gentlemen, exit five's very own growth guy with a bonus agent. Mr. Ben Reed, welcome to the show. Yeah, I'd be here. I don't know if you want to just go right into it. We do, man. Let's go. You're up. You filled up these bookshelves last night for, you know, to be here. So let's do it. Yeah. So I'm a big believer in learning. I try and follow as many newsletters as I can. The problem is now I've got weighting and notifications, weighting and newsletters. So I like to consume knowledge, but there's no time. That's why I created an agent that was going to help me just digest that. So it's going in the phases of how we've built it. phase one was just, hey, agent, I would read my subs to that every day. I like like, hey, here's what you missed. There's all this noise. Take a look at these five. So that was really cool. It scanned everything. It goes over the weekends, does all this stuff. But then I wanted to go a little bit deeper with that. So I updated the NDE file, made it bigger to where I was like, all right, there's a lot of areas of growth marketing. So I set up this way of like, all right, Monday, it's going to give me a rundown about growth marketing Tuesday and run down about paid ads, Wednesday, strategy, SEO, wild card, best of the week. So every day, it's scanning my sub stack, finding the lessons that I need to do. It's not just saying like, hey, here's a popular post, but it's saying like, based on what I know you're working on from our other conversations and where you want to be, this is the learning to have. So just give me that digest of not like, check out this newsletter. But like, here's the end takeaway. And I won't process like, I'm going to make it where you can't do this with the Xs5 newsletter because you do need to be reading that Tuesday, Thursday, and possibly more. So it went from there to now. Oh, and then with it, I use the tool Resend, it does free emails. So it's like, this isn't just in my cloud. I'm getting an email every morning of here's your takeaways. But now I've made it even better thanks to another tool called earcast that takes any article newsletter goes to your inbox and it creates a podcast. So now I actually have a podcast just for myself. So I don't care how it sounds. I'm not going to go on Spotify, find growth marketing, run down, and every day listen to my, here's my five lessons for the day. So just as I'm on my walk, it's probably a little thing. So that's a fun thing. Lastly, follow me on LinkedIn so I can win this contest. You should add a QR code. All right, go follow growth guy Ben, handmade those shelves in his in his background too. So all right, we're back. So we got three people, we can't give them all to you. But look, I do see your chats. We are a content business. We take all of the feedback and signals from these and we're going to find ways to give you more of these plays. Have people break them down, whether it's in the newsletter podcast webinar, stick with us, we got your back. Let's bring up Maddie. So we got double the amount of time. We got four minutes here. And now we're going to take a little bit of a deeper dive into each one of these agents. And let's kick it off with back to you, Maddie. All right, sounds great. Thank you for voting for me. Exciting to share my agent a little more in depth. So this is the messy middle part. And as many of you know, content guidelines aren't tidy. There are a lot of different pieces that you have to chain together style guide your brain positioning SEO guidelines, legal guidelines, how you talk about the competition. And they're not always in agreement with each other. So that's part of the problem. And you know, ultimately, it's really not a technology problem. It's us humans and the fact that we have Google docs and confluence and notion, you know, everything's living in all these different places. So how do we kind of bring it together in a way that makes sense where they're not talking over each other. They're just handling whatever the most important job is. So the skill chain is really just me getting that all out of my head, out of these documents that, you know, I probably know better than like the writers I asked to use them. And you also notice that the number of skills change since I pitched this agent because I add a new one at least every week. And that's, that's just me coming across this problem and trying to solve it. So I want to zoom in on a couple of the skills just to give you a little bit of a deeper dive into what they actually are. This first one is a reader advocate checker. And it basically reads the draft kind of the same way and uninformed human wood at digital ocean. We're dealing with a lot of more technical concepts. And so I just want to make sure even though our audience is developers that the terminology makes sense to them, especially if it is more of like a new AI type term. And so it looks at it kind of as an editor on behalf of the reader. If flags would a reader would actually stumble on maybe like a concept that's mentioned but never explained or an acronym that doesn't spelled out or even like a claim the piece raises and then just drops without closing the loop. So next we have what I call the fidelity checkers brief fidelity outline fidelity guideline fidelity. We have all these documents we've already created and we've already infused you know with the important details but basically like did everything that we agreed this piece actually needs to cover make it into the final draft that we're planning to publish. So this used to be one of the more tedious parts of my job and involved having a lot of browser tabs open. I still have to obviously look because AI is not deterministic and it makes mistakes but a lot more of this process is automated and flagged so I'm kind of informed us to at least the baseline of this going into my human head around. So this is one of maybe the most important details that I alluded to at the beginning here which is that these documents they live in different places and the other thing is that they're not static files. Quad you can use projects and you can upload documents but with skills and I could be wrong maybe there's a way to do this with projects too but with skills you can run the live document and you can check it for changes and I'm going to tell you a little bit more about how I do that in just a moment here but just like understand that it can grab the live version of a file that you might have changed you might have new guidelines you might have updated a brief etc and so basically the skill checks whether the source has changed since the last run and repulsed if it has and yeah it kind of avoids that knowledge being frozen in place. So the editing chain is really just a half of my workflow. I also have this draft content creation workflow it helps me generate on brand CTA's internal link finding finding relevant media to weave and even taking a stab at the intro and then it works with the draft agent and then or sorry the editor agent. I start the brief as a human I use the skill for that too and then there's a final human edit so I want to make sure I get this in cashier prompts which is what I said before use markdown to save on tokens and use a project manager skill so that the skills work together effectively. I love that. Hey if you are a member of the exit five team and you're listening to that do what she did because we're blasting through every 12 hours someone's like can you add more money to the clawed. So I had to tell Matt I'm like stop checking the weather with clawed please. All right. Meadle. Are you ready for your four-minute timer. Yeah, me and my answer's dog. The animations are sick. This is awesome. I have more than just lasers. I also have a really cool X-ray mode. Look at this one. I'm going to say I have substance. No. I love animation so yeah I just wanted to show you like where all the data comes from and I'll jump right into the agent and show you how I actually built it but yeah you can see the data comes from LinkedIn. It's actually doing research with built with because like we as a company we really care about the products that we have. So like when we want to sell you a CMS or an experimentation platform we need to figure out not just who you are as a company but also what you care about and how you talk. So we like I mentioned earlier we look at job open job applications everything like that and then we basically create a pitch out of this. I think the core thing that I would love to give to people is that the goal here is not to essentially throw everything that you have on a page. It's to like a hub page is often a term that I hear. It's actually take what people care about and try to figure out how to contextualize everything that you have and how to leave stuff out. So you can see here it's actually creating a net new pitch out of the real concept that we have pre-approved phrases. It's all complete and then on top of that we even have an edit mode so that you can see here internal people can actually give suggestions like cloud code style and the agents basically working in the background to do the edits including salespeople they can literally on a call go in and basically create their own landing pages directly and then you can see here this is the actual agent running in the background so on daily schedule we run it on Tuesdays and Thursdays because that's when I found most updates happen in CRM. This is our agent platform by the way so I built out this little workflow and it says on Tuesdays and Thursdays I want you to go grab a list from Salesforce and then for every company you find on that list do all the research and spin up a landing page and I can show you how that works here so it's essentially a really long agent that we built that's taking on the role of an ABM manager and then the good part about this is we actually tell it how to do a good job so that's the difference from maybe a normal agent where you write out a prompt and tell it how to do the work we actually say why it's important and what we expect to be in there and that's kind of how we get it to run at such huge scale and make it repeatable and then every week it also updates the hundred oldest pages so that it kind of refreshes each of them and keeps them fresh say they move through the deal cycle or we win an opportunity it knows to like move those out and update them but yeah you can see here all the tools that we use like Salesforce it browsers the web the searches takes screenshots updates are CMS gets their brand so yes I saw that question on the chat as well the lasers come in multiple colors it actually looks at the brand of the company and uses their accent colors throughout the page which is kind of fun and then even uses our digital asset manager to like store all the files yeah and you see here here's so my job now essentially is just to manage this at scale right it's running for thousands of pages I don't have time to review thousands of pages so actually send out a ping on teams with instructions for the sales team on how to activate it and then track how people engage with the pages and based on that I kind of have this platform that I can go through and and look it's actually like what the quality is I can even give the agent coaching and say if you're going off the rails and not doing it in the way that I like I want you to stop and I mentioned it earlier but you can see the number here it's ran over 14,000 times and it's almost too hard for me to I have to zoom in to actually focus on it but yeah there's only 120 cases where it wasn't able to do its job so which is in nine out of 10 times just someone didn't update stuff in Salesforce or we need to clean up some of our data on our side. So that's kind of what we built out. And then my favorite part of this is actually seeing people at events too. So we actually bring it to events now. We literally throw in a lead list at events to like show them their pages, which is kind of fun. So. Well. Wow. With one second is there. All right. We got Mr. Wildcard. Huge. You had to have known that it was coming. I mean, you're sitting there calculating the votes. You know, we produce a merissa called you from behind the scenes. So Alan's back. It's been a minute. He was here, you know, when it was like 12, 10 p.m. So let's go. All right. So this basically breaks down into five steps. Each one of those steps has steps, but find filter in rich score and share. Find these combination of a tool called trigify and fire crawl use mcp for both of these things. And I have a massive list of cool places. So probably 30 or so Twitter accounts, CPG brand aggregators, paid newsletters, podcasts, YouTube channels that I ripped the transcripts of it searches all of that stuff. It makes a list of brands that are mentioned. I get a gigantic list dumped into Clawed on Tuesday morning. Then we filter that list through a combination of this neat little bot, my Revops team built in Slack that allows us to check two sales force instances because we have two sales forces to make sure someone is not a customer. They're not active pipeline. If they are neither of those things use fire crawl mcp again to go and find a URL for the brand. Once we have all of the URLs, we kick them into clay where a ton of stuff happens, but we start with thermographic enrichment. So looking for headcount and where they're based and what kind of category they're in, what year they were founded. And if at any point in filling in this information, they're de-cued, they just don't move forward to try to save a little money on tokens. So if they're founded 30 years ago, I don't care about them for this purpose. This will build like part one of what will be scored eventually after this technographic. So we use store leads. We use built with, as I mentioned before, store leads is generally not super valuable from my perspective when we're looking for nascent brands, but some of them have something. Maybe we'll see they're using clavio already or they're using post script for SMS. And those are generally good indicators that these are people who know what they're doing. So they're worth us talking to. It's also good to know if they're on Shopify, Shopify Plus or neither. If they're on neither, they're auto de-cued. Site traffic is pretty useless here. If they have site traffic, it's a great sign, but no site traffic is not a helpful indicator. Did they raise money? How much did they raise? Obviously, we'll have this information if the way we found them was through a VC or through an aggregator that said, hey, CPG Brand X raised some cash. But generally speaking, this is somewhat helpful. Having money is great. Not having money is also totally fine, particularly if you're good. Then we look for the team. For new brands, LinkedIn is kind of useless in the CPG e-commerce space. So this will usually mean some more fire crawl MCP to go and find the people associated with Brand who were mentioned in articles and then find where they used to work and give me a list of who they are. Their LinkedIn profile, their last three jobs, their role in the current organization. Then we determine, is it a fit? Pretty obvious. Is it a new brand? Is it a consumable? It's yes. If they're established or they sell one-time purchase type product or they're not on Shopify, we have to pass even if it's dope. Life at the grains. This is how the scoring all breaks down. We're biased towards team pedigree subscription fit and freshness. We care less about revenue. Fundraising and traffic growth for the reasons I explained. I'm going way over time. You were doing great on time. I also really like your slides. Were they designed for the AI? They were all a lot of design. You can see the shitty little eyebrow in the top corners of the dead giveaway. See, that's my issue. I think we give Claude too much credit. Claude was trained on all of our stuff and now whenever we do something, we're like, yeah, Claude did that. I'm like, no, I did that, mother. You know what I mean? You did that, Alan. You made those slides. All right. Dash, I've been popping off. Do you know what we're supposed to do? Yeah, it's got to pick a winner. It's the time to nominate the actual, actual, actual winner of this competition. The actual winner. I don't want who had the best presentation I want. What should these is most useful for you? Which one do you think you might actually take and go put into play in your company? Is it going to be Maddie? Maddie popular among the attendees in the chat. Love to see that. Michele 2600 personalized landing pages with laser beams and Alan finding your next big customer for your competitors. No, who is it? Oh, look at this. This is the closest it's ever been. Wow. Whoever was in the chat said they don't like the voting format. I got some going to say to you, dog. This is awesome. All right. 38. You see the live voting? Like you guys seeing the numbers? So close right now. All right. We're going to call it. Should we call it? 10, 8, 7, 6, 5, 4, 3, 2, 1. Call it. Oh, no. Oh my god. The percentage. Wait, hold on. This is a tie. The percentage is a tie. But Maddie has 97 votes and Michele has 95 votes. How is that possible? Yes. We have two winners. We'll feature you both in upcoming newsletters. We'll plug you all. Congratulations to you both. Dan, you're going to have to find two trophies to send out. Excellent work. Let's just give it up. The chat was awesome today. Electric chat. I see you. I see so many familiar faces. I'm going to see next week. I saw Brendan. I saw Charlie Riley. I saw Tara. So many people out there. I'm excited for Drive next week. Thank you all for being a part of this session. Look noted. I am. I do like to be silly guy. But we do see all the stuff in the chat as comments feedback. We are working hard. Shout out to Marissa and Maria, our team behind the scenes. Ben, amazing work that goes into this. We're taking all those notes. We're going to figure out how to get more of this. You can always pop in the community. Our newsletter is a great resource for you. And then from here, we're going to see you back in person at Drive Dash. Has anybody missed anything before we hang up here today? Let's talk about experts network real quick. If you want to be on a webinar live, exit five live, like this, not a webinar, you should join the experts network. And we're going to be having a session again in two weeks on September 25th, which is creative tasks with AI is the theme. So if you enjoyed this one, you should be in the next room as well. Let's put the word out for that really quick. We're looking for speakers for that. So if you want to be on, if you are someone that is in the world of creative creative brand and you're embracing AI to help you embark on your job as a creator, creative, we'd love to see examples. Hit us up. Find Maria at exit five.com or LinkedIn or come see us in the in the community. All right. I think that's a wrap. We're out of here with 12 minutes back. So put that in your ratings and reviews and we gave you 12 minutes back here. And I think we're supposed to rate and review this session for us. Is everybody still here? Should we roll that poll? Can we get a quick gut check on this? Let us know. Was this session valuable for you? We love all this feedback. How would you rate this session on a scale of one to five? I usually say things here that skew the votes in some way. I'm doing my best to not say anything right now. So you find, you know, tell everyone how excited you are for Drive. Yes. I'm so excited. I appreciate y'all coming in. I really don't care what you rate this webinar. We're going to keep doing them anyway. But thank you. It's good for Dan and everybody else to know. We're out of here. Five all day. We could have used the extra time to provide more of the info we wanted. Possibly true. We're out. Dash. I'll see you in person next week. Yeah. Everybody. Thank you all. Great session. Have a good long weekend. All right. Have one. Hey, thanks for listening to this podcast. If you liked this episode, you know what? I'm not even going to ask you to subscribe and leave a review because I don't really care about that. I have something better for you. So we've built the number one private community for B2B marketers at exit five. And you can go and check that out. Instead of leaving a rating or review, go check it out right now on our website exit five dot com. Our mission at exit five is to help you grow your career in B2B marketing. And there's no better place to do that than with us at exit five. There's nearly 5,000 members now in our community. People are in their posting every day asking questions about things like marketing planning ideas inspiration asking questions and getting feedback from your peers building your own network of marketers who are doing the same thing you are so you can have a peer group or maybe just venting about your boss when you need to get in there and get something off your chest. It's 100% free to join for seven days so you can go and check it out risk free. And then there's a small annual fee to pay if you want to become a member for the year go check it out learn more exit five dot com and I will see you over there in the community.