OpenClaw's New "Cold Email Agents" Are INSANE

Most people think that cold email infrastructure takes a full team. You got to find the leads, you got to verify emails, you got to warm up accounts, you got to write sequences, you got to distribute campaigns, you got to check reply rates, you got to optimize the sequences. And this used to be a full-time job for multiple people. We're talking maybe 3, 4 people, and maybe it would take you a week, 2 weeks, 3 weeks to get done. But now you can do all this by yourself in an afternoon, not because you're necessarily faster, but because you have agents handling almost everything end to end. And this is how we do it with our little agents around our single brain system. Here's exactly how it works. If we look at this, so literally what I'm doing is I have a bunch of— literally at the very top over here, I'm like, hey, tell me about the Insulate campaigns this week. What's performing versus what's not? So it'll quickly, using the Insulate API, because it's hooked into my OpenClaw, it will look at the campaigns that are working, what's not working, and historically it'll show you which ones have decent reply rates versus what they don't, right? So obviously the higher the reply rate, the better. If we're sending cold emails, hopefully we can get the reply rates up to 2 to 4% or so. And you can see, like, it knows that we've paused the account. So this is hooked into our Instantly, which is our cold email, just call it our cold email infrastructure. And then I'm just asking more and more questions. Hey, what's happening over here? Oh, okay. We sent this amount of emails this week for these, you know, what are the, what's the reply rate? You can see the reply rates are pretty bad. And so what I'm doing then is I eventually, I look at some of the cold emails we sent before, successful sequences that we've had, and I basically include it. in here. And when I include it in here, it's good because here's the thing, I include it in here, it'll copy the sequences that we're getting like a 4% reply rate. It'll look at the— it's reading like a Google Drive document, and then it will reformat and rewrite sequence. And you can see if I continue to go down the list over here, and I add a whole bunch of other workflows to this thing, right? So I'm— it's like, it's having me review the sequences. And I'm like, hey, I don't— I'm literally dictating to it. I say, hey, I don't, I don't like, I don't like this, the 2 campaigns that you made over here. You've tested this, this doesn't work, etc. It'll take 2 hours. I'm like, don't take 2 hours, go faster than that. Then it'll send me this Google Doc to read, and then it'll show me— there's another workflow that I have, another skill that I've implemented into this. It will look at the different campaigns that we have, okay? And it'll score them recursively on a scale of 0 to 100, and it'll keep scoring it until it reaches 90+, and only then will it return the sequences to me. So it's still a little subjective, but at least it uses like a purported expert panel based on whatever it is that we're offering and based on the skill that we're trying to use, which in this case is persuasion, it's marketing, it's cold emailing, right? And then I asked for a bunch of other things too. So you can see, literally, I'm just telling it what to do, and it's doing that. And so then I'm asking this question, like, hey, by the way, are we using the right custom variables over here for the company names? And it's just checking, checking. Okay, checking, checking. It's checking. Okay, now this is all clean. Okay, great. So I realized that we had the wrong custom variable, and I didn't want to change all that stuff myself. So this is very detail-oriented work, and it made all these changes over here, right? So this has changed. This was changed over here. This also works. Okay, great. So, you know, by the way, like, this doesn't— like, when you're watching this video, It's very clear that you can't just hand this stuff off to AI, at least not right now. You have to make sure that you have a human in the loop that has taste, that has good judgment, then they can make the call and then they can look at these sequences over here. So I'll give you one example. So it's scoring it. Okay. Ready for my review. Okay. Yeah, we actually had a 16% positive reply rate. So obviously there's like a 16x change in what we need. And so it pulls in all the other context that I've given it with stuff that's worked in the past. Here you go. Quick question about company name over here. So this is like a campaign that we're doing for a customer right now. And literally, we're trying this on their infrastructure. Okay. So, or we're trying this on our infrastructure, I should say. A plumber in Phoenix picked up 11 hours after calls last month that would have gone to voicemail, booked $4,200 in jobs from those calls alone. I looked at company name, I think you might be leaving someone money on the table, worth a quick chat. That's not bad. We're looking for brevity when we write these emails. We hopefully, we want to, we want to have numbers in there too. And hopefully, if the numbers make sense, we can have an incentive. So for Single Grain, for example, for our products and services, We might have an incentive like a $100 Amazon gift card just to get them on a call. And that tends to work out well because who doesn't want a $100 Amazon gift card? But I've actually made changes to that now where it's not just the Amazon gift card. It can be donation to a charity of your choice, a Visa gift card, or Amazon gift card, right? You can choose what you want. So then it just, I can review these other sequences over here. And then by the way, if I think this is good to go, okay, the picked up calls, cold email sequences, they look good to go. You can go ahead and set these live for instantly for my review there. I just want to Make sure that we check the options to make sure that we have the right settings and we're good to go. So go ahead and do that. So what did I just use there? I just use Whisper Flow for that. It's great. I'm going to let it work. In the meantime, I already had it working on our single grain, our single grain cold emails, right? So I'll give you an example of what those look like right now. So basically for this one, let's see. So let's just look at step 1 over here. Okay. So this is the incentive. Okay. So we are testing instead of just the Amazon gift card, because a lot of people are using it, we're going to try something else. So your buyers are asking ChatGPT and Perplexity for vendor recommendations before they ever contact you, before they— most marketers have no idea where the brand shows up in those results. How about a 20-minute call and pick any $100 resort we want? Amazon Visa, charity donation, your call. We help brands like Salesforce and Amazon stay visible in search shifts. Like, I actually think this is pretty good. But we do need to make sure that we change— like, so this is something I caught with custom variables. If it says like Eric is your first name, but maybe the first name is not entered inside of the lead list, then it's going to fall back to there. But if it's— if you get a— if you get an email and it says there, you already know immediately it's automated, right? So you want to make sure that you fix these custom variables or you just remove the name in general. Okay. So that That's what I would say. And I don't even know if I— I don't remember if I gave it this feedback, right? So mystery shopper is dead. Okay, got it, got it, got it. Now look, I'm going back and forth like, oh, this— it's working with me, right? So I have multiple instances running at the same time. First it was like, hey, I'm going to work on the single green one. I was like, hey, in parallel, can you just start to work on this picked up calls one, which is like another sequence that we're running, right? And then like, look, I'm giving it all these rules over here. So, you know, hey, you know, these, these unit economics don't work for gift card incentives. We should draw from winning patterns. And then we also want to save this as a skill, right? Which I'm going to make sure I do afterwards. So then I have to— I don't have to repeat the same thing. So, so then if I go back to the single grain one, okay. So let's say, I think it's this one over here. Okay. So I think I gave it feedback already and let's see what it said over here. Okay. Let me prove that. Pull the context, da da da, clarify questions. Okay. So I don't, I think, okay. So let me see what it did. Okay. It loaded everything into Instantly. So I need to check to make sure that this is okay. The big loader sub-agent to create these in Instantly, redistribute leads and split accounts. So there's a lot of complexities here, right? One, you have to find the leads. 2, you have to verify the emails. 3, you have to make sure the email accounts are warmed up. Then you have to distribute the email accounts for each campaign so you're not duplicating the sends, the inbox don't get burnt out, right? But my point is all these complexities, you can now manage this. Like you as one person, you can handle all of this. You're a one-man wrecking crew. That's literally what you are, okay? Quick break. If you want the same Single Brain system running inside your company, go to singlebrain.com. We build and deploy this for businesses that want an agent fleet handling their operations the way you just saw. Links in the description. I'm pulling this up over here and let's take a look at the campaign. Let's see if this was loaded. And it looks like these were loaded. SGV4. So 1, 2, 3, 4, 5. Okay. So let's make sure we check over here that we do have leads loaded in. We do have leads loaded in, but it's only 166. So I have an issue with that. We're gonna have to fix that. The sequences. Okay. So your buyers, let's just preview this. How does this look? Your buyers asking ChatGPT to relax. So we removed the name. Okay. Okay. Not bad. Let's also try this from one of our emails. So the name's included. Let's try. This over here, like, okay, that's fine. But here's the thing. Campbell Marketing and Communications. This doesn't seem like, these don't seem like leads that we want to go after. These seem like agencies. So we're going to need to fix that. Okay. Now, one step at a time. I want to make sure that we have the right settings over here. So these are some of the emails that we have. Open tracking, disable. Send emails as text only, yes. And then stop sending emails on reply. That's good. We don't want to have a daily limit. Okay, so that's what I'm gonna— that's what I'm gonna say. And then so what I'm gonna do now is let's say I can— I'm gonna make this into a new thread. Okay, so, and I'm gonna dictate this out. All right, these look pretty decent, but we have a few problems. One, it doesn't seem like the leads were redistributed from the other campaigns that we had running. There's only 166 that went into it. And when I look at the leads for this one, they look like small companies. Doesn't really tie in for Single Grain's ICP. So, you know, Take a look at that and then figure out how we can redistribute the leads for each of these new campaigns. The other thing is I noticed that each campaign's only set to send 50 emails a day. We need to set that to zero so it's unlimited. And also I want you to double-check that the, that the email accounts for each of these is not being duplicated to other campaigns that are, that are in these V4s. Okay. So we need to make sure that we do that and feel free to ask me any clarifying questions. To make sure that we, that we get this right. So I'll drop that in there right now, but I kind of want to review some of these other ones. So like, this is new over here. We built a model that estimates how much B2B companies use when AI search engines skip over them. Okay, fine. We originally built this model for Salesforce to quantify the same blind spot in their marketing. Turns out most B2B companies over $10 million have the same exposure and almost none are measuring it. Okay, so I would, I would actually adjust this. So, well, we need to make sure this actually Paste in. So, okay. By the way, I'm looking at some of your sequences right now. So for SG V4E, the revenue leak, we want to make sure that we are talking about, you know, just leave it for most B2B companies, not over $10 million. So let's take that number out. And then the other piece is in your sequences, you tend to talk about, you tend to talk about, you know, ChatGPT and Perplexity. Let's make sure, let's just have ChatGPT, Gemini, and Claude. Those are the main ones. Let's eliminate Perplexity from all of this because Perplexity isn't really that relevant in my opinion. And going from there. So let's see what you got. And yeah, let me know when you are done. Send me a reply here. Thanks. Okay, boom. Literally, this is why Whisperflow is so good. I don't need to repeat myself over and over, but this is the whole idea. And by doing this, when we have— when you have a single brain system, this is literally what we have, a single brain system where our team has an Asian fleet and we have a world brain where people can just ask whatever questions to it. That's the single brain, right? But when you're able to do that, You are able to get a lot more things done. And this isn't it. Like, look how long this took. This video took, what, 10 minutes or so? So once I'm done with this, maybe to get this all said and done, maybe it takes an hour of my day. But okay, what about the other 7 hours of my day? What am I doing with that? There's so much more you can do in parallel by just doing it this way. And I think everything starts with having that single brain, which is exactly what we do now. There's more and more people coming to us for doing exactly what we have that you can see on my screen. So, and literally you can find out more just by looking at singlebrain.com. That's with a B. So that's the whole system. One person managing cold email infrastructure that used to take a full team. If this was useful, watch the next video on how companies like Block and Ramp are creating single brains, world brains, and why they're doing it. It's on the screen right now.