The AI Sales Lies Founders Still Believe (Claude In Sales Series)

Speaker A: Hey, sprinters. Today we talk about lies. No more lies. You say something to the agent, to your cloth agents, and they say, yeah, yeah, I'm gonna do it. And then they don't do it. No more lies. How can we get rid of this, ah, uh, uh, feeling when working with AI? And how do we get to a feeling of, oh, yeah, thank you for doing the work that I dread doing? Because some of you, they are getting really heavy ROI on using AI and, um, some, um, are not. So if you are in the camp, um, this video is for you. Because there are a few things that you can change in your behavior, in how you use your agents. And m, I'm going to show you exactly what it is. And I'm telling this from somebody who's not technical, but has achieved an incredible ROI. It's April, and in April alone, 2026, we have made an additional $81,000 in sales. And that's just three things that AI helped us see in the pipeline that I would have missed. One was an email that I didn't see and that I responded to. Another one was an opportunity that I didn't understand that was there because it was said just on the side. En passant mentioned something that the, that Claude, uh, understood. And that agent brought us to their advisor, and the advisor brought it to me. And I could close the deal by following up on that. And I'm not even talking about. And the other one was a licensing deal that I couldn't have done without it, and I will show you why. And I'm not even mentioning all the subscriptions that we could cancel. We canceled subscriptions to our CRMs to make IO, to Vidiq, to a ton of software that we don't use anymore. Asana, we don't use any of that anymore because we have JARVIS now. And JARVIS is basically cloud code, but built in a way that you will see soon. We are three humans. We are 33 agents. And those 33 agents, they report to five advisors. And that makes a difference. But let's go into practice, right? Let's look at it. You might like. I did also have completed the entropic course on cloud code, right? You started with Cloud 101. You did, uh, cowork, you did cloud code in action. You did AI fluency building with the cloud API. You learned about the mcps, you learned about, um, even Bedrock and Verdicts, and you learned about agents and workflows and that you should use workflows for things that are deterministic. You know, the outcome and that you should use agents for things where you don't know the outcome yet. Okay, that's all fine and dandy, except it doesn't really work. So we started applying all of that. This is our GitHub repo. And some would work, but then after two, three days they would forget it again. And so we had more and more rework to do and more and more. And by the way, we have documented the journey here if you want to, if you want to follow, uh, along. But it didn't work. And so last Monday we started building with our community the thing that actually works. So in the Jetpack series, we started building one agent, one advisor, and then chaining that loop together that makes the difference. So of course level one is. Claude knows who you are, right? The Claude md, the connectors, Granulo Obsidian. Level two is. Then Claude starts working for you. Now you have agents and advisors and goals. Most people don't have the advisors and most people don't have the goals. So without the advisors and without the goals, if you only have the agents, you will be frustrated. You will be very frustrated. They cannot do the work for you. So it's really important that you start with your monthly goals and then you have your current work, which is probably some form of a Sprint dashboard of a Kanban board. So your tasks. And now you have to point the agents to this gap here and ask them to track this gap and help you close this gap. How do you do that? Let me show you. Because level three then is cloud runs your company when you don't. And here is an example. You can see our Sprint dashboard and there is the humans and there is the agents. And everybody has their task and they're tracking it. So it works. We created quickly a ghost buyer that, uh, went out there, checked all of offers online, bought them, and gave us feedback on what to improve. And the feedback was really, really good. And then we have of course the tools, the 274 tools in the Sprint Club community that people are using to actually move the needle forward, move their business forward. And this is what you need to build. You need to build the Agent Advisor loop. What is it? In this example, we have 36 of these loops going on right now, and they're reporting to discord. And when one of the agents forgets something, now the advisor can pick it up, but only if you create that complete loop. Let me show you what I mean. You cannot really expect from one agent to do that. And that's also what's, what's missing a Little. So first thing is you do the task. So you're the human, you do the task. And of course, Claude is on your computer. So Claude sees what you're doing. And the next time you say, okay, the one thing that I just did, let's say it's creating content, right? You created a YouTube video and you say, all right, you saw me just record a video, then write an article, then post the social media about the same topic, right? Uh, now next time that I record a video, you turn it into an article and into three social media posts by doing exactly what I did. Do it once. Show me. So you have now built an agent that does, uh, your content. You create a video agent, does all the rest, and moves it forward. This is what people are missing. The quality will never be great. It will be around 40%. Good. Now you need an advisor that takes this from, let's say, 40% quality back and says, you know what? Do it again and improve it this way. Now it might get to 60%, and then the agent gets it back, reworks it accordingly. And now you might be at 85%. And that 85% now is something that you will feel good about. Below that, you're just frustrated. That's the RA feeling do not expect. And this is actually something that you can learn in the entropic courses. Uh, they explain very clearly why you cannot expect more than 40%, 50% from the first shot of anything, AI anything. It's not built to do that. It's built to work in iterations. And that's why, yeah, you will learn that in the course. Let me keep this video tight. Uh, you need this loop. You need Agent Advisor, Agent rework Advisor, and now Infinite improvement loop. But this might sound theoretical. So let's, uh, show you an example from this week's, um, Jetpack workshop. So we said, guys, let's build an agent. And then we had Kyle, uh, sending in this question. I want to build a seamless operations process to track partnership introductions per subscriber, and automatically send each subscriber a weekly pipeline report showing status, progress, stage of development, retain clients longer without adding admin, um, hours. So the five things that you have to ask, five things that you have to. The five questions that you have to answer when building any. Agents are always the same. And so we applied this life on Monday in our mastermind to Kyle's question. But you can apply this right now to your workflow. So think about your. Your current business, and where is the bottleneck? Is it in, um, generating attention into Nurturing that attention until they know you, like you, trust you. Is it into converting their attention into committed work, which is closing the deal, or is it activating them after you closed now onboarding and activating them? Or is it in retaining them and expanding the work via referrals or more work? Which of those is the bottleneck right now that if you solve that, it moves really the boat forward for you now that you have that bottleneck. In this case, it's retainment, uh, referrals if you have your bottleneck in front of you. Now pick that workflow and let's go through the five questions. So you answer those five questions for yourself. Question one, what data does it need to do that? In this case, it needs the subscriber list. If it was a closing question, it needs the CRM data. If it's an attention question, it needs the buyer's journey data or the website visits data, or the free trial, um, form completion data or the cart abandonment data. Question two is what do you want it to do in this case? You want it to scan for updates, to calculate progress per subscriber, and to format a weekly report in a specific way that you want. You have to describe that very clearly. Also something, um, that you learn in the entropic course, how to describe clearly. Question three, what's the output? And question four, how often do you want it? Now question five is my favorite. Where does it get smarter? This is where you build in loops that will learn even when you don't work. Your company will learn. And that's important because this is the compounding effect. This is the thing that really, really compounds and that makes the difference. If you are experience Siri style, uh, or Jarvis style, wow. The first time you do the task, that's the human. Here you are doing the task, but the computer is with you. Obviously you're working on your computer. If you're watching this and then the next time the agent will do the work, saves the output, the advisor reads the output, scores it, writes feedback to memory, agent reads memory, applies the feedback, does, uh, better version of that first work, the advisor reads the better output, writes a sharper feedback, and this thing goes on ad infinitum, um, until it is at 100% and it can be shipped. Now, these loops go insanely fast. We had loops take three hours, loop takes 20 minutes and some loops take 10 minutes. So you go from idea, uh, to an actual shipped campaign, shipped, lead magnet shipped, follow up email in a few minutes, which is incredible. It makes your head explode when you when you see it for the first time and it makes you feel really, really in God mode. That's why we call it Jetpack, because, yeah, it feels like, you know, six coffees in your veins. And that's just because you add the self improving loop. This is really the big difference, guys. It's, it's nothing magical. We're only using cloud code. We're paying 200 bucks a month. But you can do this also with the hundred bucks a month version for a while. And um, this is the secret. This is the difference between having a Siri feeling or a Jarvis feeling. Uh, um, when we wake up in the morning, we don't have to go to Open Calendar, open Gmail. Who needs what from me? What's the day we get briefed? We are three humans. We have five advisors and three agents. Each agent reports to one advisor. This is important. Why? Because of that loop, right? Let's say the YouTube agent is improving the tags of a YouTube episode on your YouTube channel. So it improved the tagging and the headlines of the last three videos you published for a B testing. Now that agent has one direct advisor that he reports to. In our case, it's the distribution advisor, it's Greg. And it says, greg, I improved the tagging, I improved the headlines of these three videos. Here is the report, gets feedback back, does it again until it sets. And when it sits, it goes to board and board tells in discord. We completed this. So that's one loop and it can take two minutes, it can take 20 minutes depending on how complex it is. This is not very complex, but you have to measure back the difference that the tags made. So it takes some time now because in the end it's always about outcomes, right? It's not about the inputs, is about what should it accomplish. It should accomplish that the video performs better. So you have to wait for the performance data of that video to close that loop. But you don't have to think about all these things. You should think about what do your people need and how can you serve them? How can you create value Right now, everything else should really be taken from you here. What I really like is having a system that does a B test all the time. All the time, all the time. And I don't want to do it like you probably also, right, There are three, four, five tasks that you really enjoy doing. Uh, for example, my five tasks are I like recording videos, I like closing deals, I like meeting new strategic partners. Uh, like Jay Abraham that we wrote just a book together or today I'm meeting Bob Berg. Yesterday I met David Allen. I like having these conversations. I like meeting my clients in the mastermind and helping them solve big business problems that they face right now. Those are the tasks that are for me really the task that I love doing. Everything else I want to be stripped away from me and that's what the Jetpack series is doing. In April we have built the foundations, we have built one agent, one advisor, one self improving loop for each participant in there. In May we will build your board of advisors. Three to five advisors that work for you. The sales pipeline will be built in June. In July we build your content machine system. You just record everything else will be done then how to onboard clients, make them successful, activate them, optimize the revenues, delegate the most boring 60% of your business, improve the A, B testing and advanced loops and by December you have stripped away everything else that's boring or that's not really human because reconciling two spreadsheets, I've done a ton of them but it's not the most. The highest expression of my humanity is just reconciling two spreadsheets. Um, the machines can do that better. This is the Jetpack series guys. And to sum it up, if you have experienced this bah feeling with AI oh my God, this is dumb. This doesn't work. This doesn't remember anything. Try the Agent Advisor Agent Advisor loop and see if there is a difference. And if you have any questions, shoot us an email or join the Sprint club and let's build together@stragysprints.com keep rolling everyone.