Only 2% of Founders Understand This About AI

Here's why you need to put your Open Claw inside of Slack. So people talk about using it in Telegram, people talk about using it in Discord. I used it in Telegram previously and it was great. I mean, I would have a bunch of different groups, I'd be talking with it, but I realized one thing, that I was becoming the bottleneck. So what I'm going to show you is I'm going to show you how it interacts with my sales team, my SEO team, my organic social side of things, recruiting, also even leadership too. When you get these concepts in mind, you're going to be able to unlock way more superpowers because it's not just you doing it now, it's an entire squad that's doing it. And if you stay till the end, I'm going to give you this revolutionary new concept that's going to help make your agents even more powerful than the 100x that I'm about to give you right now. Let's get into it. Okay, so this is my sales channel. And then we had Ashley this morning post something about email personalization, was sharing this post over here. I actually didn't even read the post because I just assumed that Ashley's judgment is really good. You know, I'm like, okay, well, this is— she's probably giving tips to the sales team on how to make things better. I'm like, you know what? I'm going to tell my sales agent over here, Arrow. Arrow is an AI outbound sales agent. So I put Arrow into my Slack, right? What happened was I was like, hey Arrow, can you read this? Take this into account when building cold outbound, because Arrow actually helps me build cold outbound sequences and comes up with prospects to go after and finds the emails as well. And then it can actually send those emails out. You can all manage it within here. In here, I just like, oh, Arrow's like, got it. Great post from Kyle Lacey, CMO at Jellyfish. Here's what I'm pulling. From it for our outbound. Hyper-personalized subject lines work. Don't jump from personalization to pitch. That's the biggest mistake. Speak to how they get measured, not your product features. Lead with proof and outcomes. Social proof is the multiplier. The bar is high right now because AI makes personalization so easy. By giving this nugget from Ashley, passing this nugget over to Arrow, now this is added into whatever outbound sequences that Arrow will make in the future. This is now baked into the outbound framework. So Ashley sees it and she's like, whoa, that's amazing, right? In the morning, this is what I'm testing right now. It's only 6 daily prospects right now because we're just testing this. So you want to make sure that you nail something before you scale it. So 6 prospects ready for review across one source. Deal of the day. Okay, this person from Klaviyo over here, Alana, partner marketing manager, score 63. Richardson Sales Performance, Laura Morrison. And then we have Adecco. So it gives you all of these people that you should be reaching out to. So Arrow, what it does is it'll look at all my LinkedIn connections. Those are more warm connections to maybe try to figure out an angle, and then Arrow will figure that out. But also we'll look at deals that are stalled or deals to revive. So maybe deals that we lost over 90 days ago, over the past 2 years, maybe we should just message all those people. We should just email all those people. Not only that, we should just text those people too. So not only is this, this claw of Aero an autonomous agent, but it's an autonomous agent that now collaborates with your team. Do I give it a lot of permissions? Like, for example, my personal email or credit cards or passwords and things like that? No, I don't give it that. In fact, the only access it really has, my OpenCLAW instance has, is a vault for passwords that's only dedicated to that, right? And so I'm trying to minimize the attack vector for myself. And so there's always going to be vulnerabilities. But from a security standpoint, if you're like, for example, my CLAW is on a local device, right? And so I would just say that doing this alone, you know, again, I want to just reinforce this. You get one nugget from Ashley, okay? And then I pass the nugget over to Arrow, and then Arrow reinforces that and then builds it for the future processes. So it compounds and then the rest of the team sees how this works. And then the rest of the team can work with these daily prospects and they can also give arrow feedback, right? So for example, so over here you can see, I'm asking it questions within the daily process. Hey, what prospects or what copy did you send? What was the copy that you sent? Give me examples of the first 2. So this one over here, subject Klaviyo Inc., following up on our conversation. Hey, Alana, it's been 107 days since we last connected. I'd love to schedule a quick 15-minute call, view word things. And so it's very generic messaging. So by the way, now that I have this Kyle Lacy thing, I might just say, hey, use this Kyle Lacy framework to redo the copy, right? And so it comes up with, I'm like, hey, go find like 15 more fresh prospects with personalized copy. And then it finds all these people over here, which I don't know who these people are. Then what happens is, where's the sheet? And so it gives me all this information over here. So by the way, I'll say this, watch. Hey, I gave you the Kyle Lacy format for outbound. So give me an example with one of these prospects on how you would use it because I want to see it in action. So, okay, we just got the answer over here. So I just asked it for the Kyle Lacy format. We've written using the Kyle Lacy framework. I'll break it down each piece so you can see the structure. So gone are the days of being a point solution. So I don't like that. I think we should just usually include the company name. But Raul, that ad exchanger quote stuck with me. You told Impact View succeeded by making contrarian bets, building on Amazon's DSP when everyone else is chasing DTC. So this is cool. So I think that's actually pretty good. So it's hyper-personal, okay, from a public interview. So, but the challenge with that is like, how recent was that public interview? Opening acknowledge, opening acknowledges the specific thing, references the exact ad exchange or piece and just contrarian bets and then bridge their metrics. Okay, that's good. Proof with numbers. That's also very good as well. And then CTA tied to business problem. You can set this to be fully autonomous if you want it to be. If you stay to the end, I'm going to show you the concept on how to tie all this together, because what you don't want to do is you don't want to be managing back and forth. You don't, you do want this to be a lot more autonomous and you can get there, right? You can have the agent sitting in here. This is the future of doing work. Okay, so now we're on to the next one, which is SEO. So my SEO bot is called Oracle. You know, I actually asked over here, I was like, hey, so the team knows what abilities do you have and what tools are you connected to? So I am Oracle, Single Grain's autonomous SEO and AEO execution agent. I run 24/7 on a dedicated Mac Mini, not a chatbot you have to prompt. I analyze, recommend, execute with approval, verify, and learn. So there's connections to GSC, Google Analytics, WordPress API, your passwords, and ClickFlow, which is our content optimization product that we actually use to create content, right? And so you can see these are the daily cron jobs that it has running. I ran like a small test, like, hey, optimize 117 meta descriptions. And this is cool. You saw the click-through rate for these go from 0.16% to 0.22%. And then you can see we noindexed, you know, a couple pages and there's these tier 1 pages, right? And I'll tell you kind of what challenge I'm having with this right now. So there's a lot of cool insights with this. The problem is I am not— this is not my full-time job. Like I'm making these agents for my team to use. And then eventually we want our clients to— we want our team to be using these for our clients, right? What's interesting here is it came up with some ideas on outside the box ideas. So it's like, Number 6, answer engine bait. So, you know, in terms of SEO strategies that we should be trying. So traditional SEOs are optimized for Google's ranking algorithm. You know, I can reformat existing high pages into high authority pages into LLM optimized structure and create new definitive guide pages for terms we want to own. Think what is LLMo. Amy actually asked a question here. So how would you grade our existing site content against your own optimization levers? Please separate based on subpage groupings and URL structures. This is cool, right? So it's giving ideas here and she just jumps into the conversation. She's like, okay, well, how's this working right now? Then Oracle just starts to work and it's looking through all these pages over here, right? All these PSEO pages that we have, we have 776. It grades us across schema markup, entity clarity, C-, C+, D+. Services, you know, how are we doing on this? You know, C, D, D. So not that good, right? Blog over here, 2,000+. The content engine, Mostly Bs and Cs, all this stuff over here. So I'm like, oh, okay, TL;DR, look, C, C+, C-, D+, D+. So Amy, from this nugget over here, starts working with it and then tags 2 other people in, tags 3 other people in, okay? And this is cool. The whole idea here is that the agents that you have inside of Slack, as your team responds to them more, they're gonna continue to get smarter over time. So they have memory so they can Continually improve over time, right? That's the whole idea with these. So you as one person, you're only going to be able to do so much. But when your team— the whole idea is hopefully you have smart people on your team. You have smart people on your team that are able to feed their good judgment into this. And so the machine gets better faster. And not only that, it's also being used to do work faster. You're not building these things ideally just for you to use them, not just a personal assistant. You are looking for an army fleet that will help you grow whatever it is that you're doing to help you drive more revenue. That's what you're trying to do. All right, so next up, let's talk about recruiting. So I have an AI recruiting channel, and our AI recruiting bot here is called Cyborg. So it's our recruiting agent. You can see it's actually Cyborg from Mortal Kombat. So I'll just give you an example here. So we have— there's a role that we're looking for right now. So we're looking to hire a GM/COO type that can, you know, make sure that the trains are running on time and also driving AI fluency and adoption throughout the organization, right? And so I had Cyborg working and we have one of our recruiters also in this chat group as well and she's working with it and she's basically said that Cyborg in its current form right now, it's still very early, it's already a 6 out of 10 versus the product that we pay for that just raised $36 million, they're a 7 out of 10. So Cyborg's already getting a lot better, a lot quicker and the whole idea here is like you're not going to get it perfect the first time. The whole idea here is that you're collaborating with these things, you're feeding it more data and it's getting better with you over time. I want to repeat that concept over time because a lot of people are, oh, it's just not there yet, it's not there yet, right? Um, those are people that just don't want to try and they, they don't want to help themselves. Don't be one of those people, okay? So the cool thing is with Cyborg, I can like— every day it's coming up with, um, you know, new candidates to reach out to. And what we've actually built with Cyborg today, uh, we're tying it in with the, these, uh, these email inbox that we're warming up right now. It's going to send, uh, emails in a sequence. It can even text message people. Automatically as well after they've responded to us. You overall, you want to hand off a lot of these tasks and you want your team to be managing these things. That's why I say this is such an unlock right now because once your team sees this in action versus you just talking about AI all the time, then they're really going to be drinking the Kool-Aid, right? So that's what I would say about that. All right, so next up, let's talk about the social media side of things. So social media, my bot is called Flash and this is our content agent. So actually Flash right now is actually helping me create a video clipper that is finding the right moments. And then also it's not only finding the right moments, it's editing as well to make sure that we have the right hook in the beginning and also making sure that there's captions in there and everything. So there's a lot of tools for this already, but I found that a lot of these tools just aren't that great with finding the best moments. So what I have right now, it's already pretty close. I would say it's like 8 out of 10. I'd like to get it to 9 out of 10 before we start to kind of repurpose content and then push it out there. And so, you know, when I introduced Flash into this channel, Flash is like, hey, I'm Flash. I'm the content intelligence bot. We're good for trend scouting, content repurposing, content intelligence. I work with Oracle and Alfred to help Eric's content machine running fast. When Oracle spots keyword gap, I find content to fill it, right? Speed is the whole point. Good content shouldn't die after one platform. Let's go. I actually asked Flash to scan this channel because I'd been pointing out a lot of issues from the last 30 days or so, and it actually did a good job. So I like, for one of our team members here, it's like, oh, And we have podcast ad placements here. Spotify podcast feed was down. Apple Podcasts host migration still blocked. YouTube videos dying out the gate. Packaging problems, right? Agency mentions over here. So these are all problems to fix. These are all problems that we want to, want to attack. But we actually have 9 things here, right? So the pattern that I keep calling out is he shouldn't have to be the one catching these errors. The feedback from weeks ago isn't being consistently applied. He wants to be able to trust the process without looking over his shoulders. John, if you can address these with next steps, that would go a long way. And so here, you know, these are all very pointed questions for John on our team. And, you know, we're fans of public feedback and John can take public feedback. So we have this over here, right? I actually asked it to create some content on my Instagram right now. When I use Flash, it averages about 85,000 views per X post that I put out there. Not only that, I put out an image that it created a couple of days ago and I put it on my Instagram and that post has 30,000 views on it. So not bad. It's like a data post that's tied to AI and marketing and business. So the TAM, the total addressable market for that is not that high. So really sky's the limit. A lot of the things that you used to think that you, you, you, one, you hate doing or you don't, didn't have the time to do, you can do all those things now and then hand it off, right? Once your team sees this, that's when the compounding truly starts to happen. Quick break. If you want to run personalized LinkedIn ads and have personalized landing pages to convert your customers at a much higher rate on LinkedIn, check out Karrot. That's K-A-R-R-O-T.ai. Karrot allows you to do things that LinkedIn ads does not allow you to do. Right now, check it out. There are publicly traded companies such as Semrush and Sitecore that are using this right now, and you can use this to get ahead because if you don't use this, it will take you weeks, if not even a little longer than that, to make these ads because again, you cannot do this on LinkedIn. So again, go to www.karrot.ai, and we'll see you on the other side. All right. So last but not least, by the way, you want to stay to the end because I'm going to show you this other concept. But last but not least. I am showing you how Alfred, which is my chief of staff bot. So Alfred is the chief of staff right here. You can see a picture of him. So chief of staff agent, basically ClickFlow, which we just launched very recently. Right now we have it on self-service and we're kind of seeing how people are. It's our SEO AEO product that helps create high-quality content. And you can see, so right now it's like, okay, 13 paying customers in 2 weeks. And this is, there's, we're all kind of testing it right now. We haven't put a big marketing push behind it. Right. So it's continuing to grow right now. And then Nat from the team comes in and says, hey, does our butler have access to Mixpanel? I'd like to see where we're losing prospects and what we can do about it. So Alfred's like, yeah, I've got Mixpanel API keys for both ClickFlow and Carrot. Happy to funnel data and figure out where prospects are dropping off. What specifically would be more useful? Sign up to activation flow, trial to paid conversion, or the full top of funnel visit to sign up? So visit to sign up and where are the warning signs, signals, and are we losing prospects in the current flow? So I have a single API secret, blah, blah, blah. Okay, boom. The key thing, I don't want to bore you with all the details. It's spotting the trends over here and it's like, hey, like this seems like there's a big drop-off over here. There's a big drop over here. Cancellations are trending up. And then Alfred's like, want me to dig into these? Yes. And Nat's like, yes, please. And then we see these other, these other things, right? So we can see the onboarding drop-off waterfall over here. It makes these tables and all that. So the key thing is this is being exposed front and center in front of my leadership team. And then what happens is, you know, there's, look, there's like 29, there's 29 replies basically in this chat over here and you just keep going down and then you can keep asking all these data, this data to Mixpanel. This actually spurred action and it spurred a discussion and that's what you're looking for ultimately. So you are looking for, I just look at this as just you're passing ideas along and people are deciding to take action on ideas and the faster you take action on these things, the faster you get things done as a company, the The faster you grow as a company. So that's how I see it, and that's how it's working right now in real time in our leadership chat. All right, so the final thing is I want to tie you in with a concept here known as calibration agents, because the stuff I'm showing you right now, you're seeing a lot of back and forth with my agents. The question is, how do you make them fully autonomous? And arguably, I can make a full video on this for the next video, but I'm just going to cover the high level around calibration loops, right? For your, your CLAWS and whatever agents that you have. So the whole idea here is that, look, if your claws need your permission all the time, you are the bottleneck, which is what was happening with me before when I was just using Telegram. But even if we use Slack, there's still kind of a human bottleneck there. But you also can't just let it do its thing in the beginning. You have to kind of, you know, train it and let it spread its wings over time. So the concept that I was thinking of here was that you want to constantly calibrate these things and there's loops of calibration and there's times to calibrate more, there's times to calibrate less. But also, nothing is ever set in stone because your business is always changing. And so what if you had the ability to set up a calibration meter for all of your agents that, you know, that they're allowed to monitor, right? And so the whole idea here is that, you know, this calibration score will directly control how much autonomy that your AI gets. So 0 to 25% supervised mode, 25 to 50% guided mode, 50 to 75% assistant mode, and then 75 to 100% autonomy. Autonomous mode, right? And then you can kind of calibrate per channel, and then you have this feedback loop, right? So by the way, when your product changes, okay, you probably need to recalibrate how you're going about, you know, you're doing your outreach, or when your ICP changes, you should probably recalibrate it. And so the whole idea here is that you have an operating system here where, by the way, this is just a Google Sheet over here, and you can see, oh, for recruiting, we've kind of calibrated it quite a bit over here. So we've approved quite a bit of things, subject lines, X content, sales stuff, cold email. The more we approve, the more it feels like it's calibrated and it's the more it's learning and then it can more or less kind of operate on its own, right? But if you don't train the agent, it is not going to be able to be the most helpful to you. So that's the whole idea. So look, I packed a lot into this one video. I think that's it for this one. Let me know if you have any questions. Let me know what else you'd like to hear about in the comments below in terms of the clause or AI in general, Claude Code, whatever it is exactly. Drop a comment below and drop a like as well, and we'll see you in the next video.