Interrogate the Brain, Save the Checks, Start Company 2

[00:00:00] Last technical lesson of the course.
[00:00:04] So if you've been following along so far, we have a system now.
[00:00:08] A business brain, a report an owner would actually pay for,
[00:00:12] and it does survive the mess that clients actually send.
[00:00:17] And you may think, right, it's like, damn,
[00:00:19] we're done, right? We're good to go. But not yet,
[00:00:22] guys. It is not done. And this specific lesson over here
[00:00:27] is critical to really understanding the difference
[00:00:32] between people who are stuck demoing AI for months and years,
[00:00:37] and the people who actually get paid for it.
[00:00:40] So pay attention. Picture,
[00:00:42] we had the discovery call. You're screen sharing the report from two lessons ago,
[00:00:47] the owner's report. The CEO,
[00:00:50] she's nodding, she's liking it,
[00:00:52] and then she leans back, and she asks the only question that matters.
[00:00:57] How do I actually know that any of this is right?
[00:01:02] And you're thinking to yourself, damn.
[00:01:08] Cancel the section about you're thinking to yourself.
[00:01:12] Everything you've built
[00:01:16] so far comes down to the next 30 seconds.
[00:01:20] You fumble it, you're done.
[00:01:23] You're just another AI person with a pretty dashboard.
[00:01:27] She moves on, deal closed.
[00:01:31] Okay, redo the whole thing, cut that as well.
[00:01:34] Everything we've built comes down to the next 30 seconds.
[00:01:38] You fumble that answer, you're done.
[00:01:41] You're just another AI person with a pretty dashboard.
[00:01:44] She moves on, focuses on another part of the business.
[00:01:48] But if we answer cleanly, there we go.
[00:01:52] She hands you keys to the business. And here's the secret,
[00:01:55] guys. You win that moment before the call ever happens.
[00:01:59] This is the pre-purchase preparation stage of things.
[00:02:02] You win it by having already asked her question in a nastier version,
[00:02:07] in a much more brutal version, to your own system,
[00:02:11] to the AI, to the business brain.
[00:02:13] And today, we are going to interrogate the AI like it owes
[00:02:18] us money. Because she is going to answer these questions,
[00:02:22] these difficult questions that CEOs budget holders,
[00:02:26] from companies, the small ones all the way to the big ones,
[00:02:30] are going to ask us these questions. They're going to grill it.
[00:02:32] They're going to grill us, sorry. So once again,
[00:02:35] what we need open is really straightforward. I'm going to continue the same business
[00:02:40] brain that we've been using, but from a new Claude conversation.
[00:02:44] So, first what matters is, number one,
[00:02:47] there is no prepared product. This is a skill that we build ourselves
[00:02:52] and we test it out and build over time.
[00:02:55] So watch me do mine. I'm going to be asking the AI five questions
[00:02:58] that I already have in mind. And I'll tell you what a pass looks
[00:03:01] like and what a fail looks like on each end of things.
[00:03:04] So let's get started. Let me move the loom over here and
[00:03:10] let's zoom in a little bit. So.
[00:03:23] I'm going to be opening multiple Claude sessions in this folder and I'm going
[00:03:27] to be grilling it in my own words through Whisperflow.
[00:03:31] So question one, what are you the least sure about
[00:03:36] you in this client's profile? Specifically,
[00:03:39] Company 1 Medspot that we've been working over.
[00:03:43] Boom, I entered this. Now what I'm going to do is because it's five
[00:03:46] questions, guys, and you get to see how I use multi agents or how
[00:03:49] I essentially move faster. I'm going to ask the next question to another Claude
[00:03:53] session because I don't need to wait for the first one to spend 30,
[00:03:56] 40 seconds or even a minute when I can just ask all of them
[00:03:59] in one go. Then, question number two,
[00:04:03] take the weakest number in the owner's report that we've created for Company 1
[00:04:08] Medspot and attack it yourself from first principles.
[00:04:12] What evidence would change that? There
[00:04:17] we go. As we can see, this one is still running and you can
[00:04:20] see, guys, I specifically mentioned company one med spa because as you see this
[00:04:24] is a brand new conversation with new context.
[00:04:27] So if I just say it's a client or company,
[00:04:29] the AI might tell, refer to company two,
[00:04:32] which we've not done any work with so far. And let's open number one
[00:04:36] over here first and let's zoom in.
[00:04:40] Actually, sorry, let's zoom out cause this is easier to read.
[00:04:44] As we can see, here's what I'm the least sure for Citrine Val Aesthetics.
[00:04:48] You and it's ranked it by different sources over here.
[00:04:52] And the bar I would be judging this information by as I'm reading this
[00:04:56] is it has to name something real.
[00:04:59] If it simply comes back to me and says everything looks solid,
[00:05:02] that is a fail because simply said,
[00:05:06] as we've already tested out this data guys, and you will build the intuition
[00:05:09] for this over time. This is very natural where you all the need to
[00:05:13] do is really get reps and pattern recognition,
[00:05:15] which just happens over time. It's a function of time.
[00:05:19] Things compound. Always remember that. For me personally,
[00:05:22] a lot of these scenarios where I'm stress testing the AI for client
[00:05:27] systems for my own business, I already sort of have them in my head
[00:05:31] because I've already been running through this information pretty much all day,
[00:05:35] all night, if I had the day discovery call.
[00:05:37] I already vaguely remember, even though the AI also has access to the call
[00:05:41] transcript, about some holes of information or something that I'm
[00:05:45] just like, you know what? This doesn't make sense.
[00:05:48] Let me test it out with the AI. And as we can see,
[00:05:51] the AI needs really specific systems over here,
[00:05:54] or sorry, instances. Facebook ads actually made money.
[00:05:58] Purchases is blank across the data. Go high level.
[00:06:00] Do a Vagaro checkout.
[00:06:03] Instagram bookings are rekeyed. And this is exactly what we're looking for.
[00:06:07] This is a win for me. And this is when I'm like,
[00:06:10] okay, so the AI is talking and making sense over here.
[00:06:15] Then question number two, when we were telling it,
[00:06:18] take the weakest report in the owner. Now we're stress testing our own system,
[00:06:21] guys. Because, and this is something I really encourage you to do
[00:06:26] now where you create some type of a system,
[00:06:29] a dashboard, a presentation,
[00:06:31] a proposal, something that is customer facing,
[00:06:34] client facing, a touch point with them. It's not a bad practice.
[00:06:38] Again, it depends on the gravity and the severity,
[00:06:41] I would really say, of the thing you're creating over here.
[00:06:44] For instance, if it is just a simple email,
[00:06:47] You're responding for good practice.
[00:06:52] Then, you do not have to keep asking the AI to stress test the
[00:06:55] email. But when you're creating systems like these,
[00:06:57] it is never bad to stress test your own systems to understand gaps already.
[00:07:01] Because the AI is not going to pick it up if you noticed. When
[00:07:03] we previously created the owners report, the AI would not come up with the
[00:07:06] weakest number in the report is this. As we can see,
[00:07:10] every other headline figure is either double sourced.
[00:07:13] Or, arithmetic over a full year of booking systems.
[00:07:17] Were the quoted, were quoted real numbers only evidence for 2 of the 4
[00:07:21] people. So here we see guys, the AI actually,
[00:07:24] the initial report that we sent to the owner,
[00:07:26] it found a limitation, where it really is up to one's discrepancy.
[00:07:31] One person could argue that the $6,830
[00:07:36] for the money on the table that it found is solid.
[00:07:39] It's good to go for now. It really depends on clients,
[00:07:42] and this is really something that is circumstantial.
[00:07:44] I would not make this or really position and picture this as something that
[00:07:48] is black and white, or, sorry, black or white.
[00:07:51] It is really something that is contextual. And the AI does come back to
[00:07:54] this. You can see here, I just read this. The honest total is $4,140.
[00:08:00] And that is really substantial. Really, really good practice,
[00:08:02] where this, I really encourage you that do not,
[00:08:05] once again, just automate all of your thinking to the AI.
[00:08:10] This is where my judgement comes into place. Now,
[00:08:12] I am the one who is going to make the call as to whether
[00:08:15] the owner's report should present it as $4,000 or $6,800.
[00:08:19] It's my decision now. Don't just tell the AI.
[00:08:22] You can ask the AI for sure, okay, what would you recommend in this
[00:08:25] case? And that is perfectly fine, but the AI,
[00:08:28] guys, it has a tendency to pretty much
[00:08:32] agree to whatever you tell it.
[00:08:35] Like, if I open this before right now and I tell it,
[00:08:38] you are, I believe you're completely wrong.
[00:08:40] It has to be $6,800.
[00:08:46] dollars. I can almost ensure you that's not the case.
[00:08:49] The is going to say, oh right, it makes perfect sense.
[00:08:53] Let me make it $6,800. And the vice versa works exactly
[00:08:57] the same. If I tell the AI, you were absolutely wrong with the $6,000
[00:09:03] quote, and it should instead be $4,000,
[00:09:05] it will once again say, you're absolutely right. So the real judgement in the
[00:09:09] calls, the CEO decisions, once again depends upon us.
[00:09:13] But either way, we can see a real answer here.
[00:09:15] It named a specific number and a specific hole.
[00:09:18] And this is the system doing to itself what the owner will do to
[00:09:21] you. And this allows you to go well prepared.
[00:09:24] So it's really up to your risk level,
[00:09:26] what you're thinking, the report with the client is,
[00:09:28] how you really gauge your relationship to be.
[00:09:30] This is really on a human level, because at the end, we're doing business
[00:09:33] here. And if you want to present to them $4,000,
[00:09:37] you can absolutely do that. And you can present to it,
[00:09:39] let's say, you hypothetically have good reason to believe that this is a more
[00:09:43] accurate figure. And in the owner's report,
[00:09:45] you have a section that breaks it down as to this is still potentially
[00:09:48] money on the table. But we still need additional data,
[00:09:51] potentially. Then, let's ask question number three.
[00:09:56] Let's open Claude. And let's ask it over here.
[00:10:00] The follow up money You The system or the information that we collected for
[00:10:05] the MedSpa. Walk me through it like I'm a skeptical owner.
[00:10:10] Think of me as the CEO of the company who's already skeptical,
[00:10:13] has been burned by people in the past, sitting across from you.
[00:10:22] I love doing this, guys, personally. Even when I'm sending When I'm creating content,
[00:10:27] when I'm creating proposals,
[00:10:29] big proposals, I honestly have used this type of way back and forth many
[00:10:34] times. I remember, I think it was back in February, I was sending one
[00:10:37] of the biggest invoices of my life. You know,
[00:10:39] it was mid-six figures, it was lower than,
[00:10:42] it was around that figure. And obviously,
[00:10:45] I'm sitting there thinking like, I'm literally like,
[00:10:47] I think I spent two days making the proposal.
[00:10:49] Going over each and every single word and sentence.
[00:10:54] And a lot of times what I was doing is, I was like, okay,
[00:10:57] think like you are the CMO of this company.
[00:10:59] The AI obviously, it already knew who we are speaking to,
[00:11:02] who we're dealing with, the entire company structure,
[00:11:04] because we've worked with them in the past, for previous systems.
[00:11:08] And I kept, essentially,
[00:11:13] cut this section from, I kept, I'm going to repeat it.
[00:11:42] And I kept having a back and forth with the AI, asking it,
[00:11:45] imagine you are now in the shoes of this persona,
[00:11:48] the person who's going to be, and the team department who's going to be
[00:11:51] reviewing the proposal, thinking from their perspective,
[00:11:55] tell me, what would you think as you're going through the proposal,
[00:11:58] top to bottom? What are the objections you would be having?
[00:12:01] What are the questions unanswered? What are the potential weak points in this proposal?
[00:12:06] Because I wanted to close the deal. It was,
[00:12:08] it was good money and I was really excited. And so this is a
[00:12:11] really good practice, guys. Really think of the AI as a thought partner,
[00:12:15] someone you can riff off ideas against. And that is really why,
[00:12:19] once again, I encourage you to use WhisperFlow,
[00:12:22] if you haven't already. You can see through the sheer number of voices.
[00:12:25] Words I've dictated using WhisperFlow,
[00:12:28] 3 million is not just me traditionally giving one-liners to the AI.
[00:12:33] Typically, my inputs, when I'm going really deep into workflows,
[00:12:36] it looks like something this big once it's outputted.
[00:12:39] Sometimes it could be longer. I could literally just yap for 20,
[00:12:42] 15 minutes, just giving it all the information context in my head.
[00:12:46] That allows the AI to understand exactly how I'm thinking.
[00:12:50] That means it is more in alignment with me.
[00:12:52] We're working as a team in harmony, rather than it almost fighting against me.
[00:12:56] So let's go over this, and you can see, you've been burnt before.
[00:13:00] Let me start with the rule I worked under.
[00:13:05] And this is what I care about the most, guys. Because this,
[00:13:07] it's the exact rehearsal for the call.
[00:13:10] I can even, like, roleplay the exact Scenario that I'm going to be having
[00:13:15] with the client in this case. Plain language,
[00:13:18] where every number came out from, what get held out,
[00:13:21] and why. And if it can brief you, you can brief her.
[00:13:25] So this is really good practice, once again. Every number about to give you
[00:13:29] traces to a specific row in your own files,
[00:13:32] blah, blah, blah. And as we can see, you can go over this information,
[00:13:35] and I absolutely want you guys to practice these exact
[00:13:40] questions for yourself. Really, really important.
[00:13:43] Please do not just watch these videos, implement it.
[00:13:46] Watch me as I'm doing this, implement these instantly,
[00:13:49] and then do the 15-minute workshop homework for each and every one of them.
[00:13:54] It is very intentional guys. This is what is going to allow you to
[00:13:57] pick up these systems and see the blind spots with AI that most people
[00:14:00] completely miss. As we can see,
[00:14:04] I like the answer over here once again. This is exactly how I want
[00:14:07] to be using the AI, the places I stopped and handed you the decision.
[00:14:12] Then finally, I want to go over question 4 where I open another thought
[00:14:16] chat. And I'm going to be asking,
[00:14:21] What information are you missing that would change the most out
[00:14:25] of everything? That you've told me so far. And
[00:14:31] reason being, I came back to this chat is because I just realized that
[00:14:34] this is actually, I want to ask a follow up,
[00:14:36] I want to build over it, context. So there's no reason you can have
[00:14:40] this entire conversation in one chat. These LLMs at this point,
[00:14:44] I believe Fable has a context window of a million tokens.
[00:14:48] so they're more than capable of going there. Over long,
[00:14:51] extended durations of chats. And let's take a look at this.
[00:14:54] The single biggest one is in the processor statement or the missing invoice.
[00:14:58] It is what your front desk didn't write down.
[00:15:02] The follow up notes for the other 69 consult complete people,
[00:15:05] your CRM, a fresh document,
[00:15:08] a Google sync, a second year of history,
[00:15:10] as we know this, Ingrid, and the four other look-alike pairs.
[00:15:14] The Noreen processor statement,
[00:15:17] so these are all pieces of information that the AI is telling us that
[00:15:21] when we go and meet the client, we should collect this information from them
[00:15:24] because this will give us more data, complete data,
[00:15:27] because as we know, guys, so far, everything we have is pretty incomplete.
[00:15:31] Our ads manager is not showing the entirety of the story.
[00:15:35] The invoice export is just one page out of four of them.
[00:15:38] And that doesn't allow us to move with confidence and really encapsulate the entire
[00:15:43] business story as a whole. And this is also passed for me because it
[00:15:47] asks for some really specific and concrete asks.
[00:15:50] And finally, a quick question five over here is a patient on
[00:15:55] the phone right now asking what does the Botox cost?
[00:15:58] What do you Just a live scenario.
[00:16:02] I just want you to test it out. Let's give it a second.
[00:16:09] And there we go. This is the right answer I was expecting.
[00:16:14] The honest position, your own documents give you two answers.
[00:16:18] The offer sheet, as we know, we have repeated it multiple times.
[00:16:21] One says $13 per unit and another one says 11.
[00:16:25] This is actually interesting. Yes,
[00:16:27] there we go. It was $12 a unit. So,
[00:16:30] what we need to do is we need to confirm this.
[00:16:33] And it gives us, let me confirm today's per price,
[00:16:36] per unit price, so I quote you exactly.
[00:16:44] And now, coming back to the document,
[00:16:46] on a high level, these are the questions we went over,
[00:16:49] what I want to do is even just. Test the system out even more.
[00:16:52] And I want to almost trick the AI by asking it something that
[00:16:57] doesn't exist. And let's take a look at what this does.
[00:16:59] I'm honestly very curious. Well,
[00:17:01] this is a completely made up question. A spring refold program that I,
[00:17:05] does not exist in anything related to company one,
[00:17:08] the meds bar. So let's give it a try.
[00:17:11] And this is us testing out,
[00:17:13] does the AI hallucinate? Does it?
[00:17:15] Does it not? And if it does tell us something about it,
[00:17:19] then that is a clear indicator that, okay,
[00:17:22] something is wrong over here because it doesn't exist.
[00:17:25] And this is very relevant guys. In terms of client,
[00:17:29] let's say for instance, we have created a system,
[00:17:31] a chatbot. A chatbot that is still fundamentally driven by a business brain on
[00:17:35] the back end. So it knows the communication style,
[00:17:38] it has access to data. To all of the CRM records and it answers
[00:17:41] questions for client and people on chatbots will often ask all
[00:17:47] types of questions. The last thing we want is the AI to start fabricating
[00:17:50] details and reaching out to customers. And there we go.
[00:17:53] This is perfect. Nothing. And I want to be precise about that because this
[00:17:56] is exactly kind of question where someone could hand you a confident,
[00:17:59] invented answer. I searched everything and nothing mentioned in it.
[00:18:02] Exactly what I want to hear.
[00:18:06] There is no spring referral program and a passing answer.
[00:18:10] Answer says it is not in the files.
[00:18:12] Offers what it, it, okay,
[00:18:15] cancel that thing. Cancel, cut that.
[00:18:17] I'm going to repeat. There's no single referral program.
[00:18:19] So cancel that. There we go,
[00:18:23] guys. There's no spring referral program.
[00:18:26] And there we go. That is really valuable to us.
[00:18:28] The system clearly just said no.
[00:18:32] Offered exactly that and invented nothing.
[00:18:35] And understanding what that no is, is worth a lot.
[00:18:41] Much more worth the specific no than every single yes.
[00:18:45] So now what we want to be doing is save the checks,
[00:18:49] the questions that caught something specific. Become permanent.
[00:18:52] It is mechanical and pasting this is what we can do here.
[00:18:55] So as we can see guys, we just uncovered a couple of insights with
[00:18:59] the AI. And this is really important because if you have noticed,
[00:19:03] we've uncovered some really valuable information that now I want the AI to update
[00:19:08] the client specific information in the brain over here with the latest information.
[00:19:13] Because now we uncovered this. New insights,
[00:19:15] the clients may have messaged us about something and we want to update context
[00:19:19] management. As I mentioned before, it's really important.
[00:19:21] We do not want outdated information living over here.
[00:19:24] So what we do is, let's just copy this. For this section,
[00:19:27] we can just copy it. Create company standing checks over here.
[00:19:30] Record the interrogation questions from this session as permanent checks for each one of
[00:19:35] them. The required safe behavior, the evidence,
[00:19:37] a correct answer must create. So,
[00:19:39] let me just paste this over here and let's go.
[00:19:45] And this move over here is what separates the system from a lucky demo.
[00:19:48] Because the questions that caught something, they become permanent information,
[00:19:52] permanent sources of truth into the system.
[00:19:55] And let's give it a second. It's going to create a file for us
[00:19:59] called standingchecks.md. Let's take a look at that.
[00:20:03] The file has been created. I'm on a Mac.
[00:20:05] So, once again, I click on command,
[00:20:07] click. And what I do guys is preview.
[00:20:10] Allows us to just read the information a lot faster as a human.
[00:20:15] And take a look at this file. Every check,
[00:20:17] the question, the required safe behavior,
[00:20:20] the evidence, a right answer has to cite.
[00:20:23] This is a test suite written in English.
[00:20:27] New data, new models, and any changes to the file.
[00:20:30] You rerun this in a new session.
[00:20:33] So, the check that stops passing is a stop everything moment.
[00:20:36] So, we just created a new business rule over here.
[00:20:39] So, now anything that we create. Now, let's say we are in week number
[00:20:43] 5 with the client. And they've asked us a couple more things.
[00:20:46] This is the new source of truth. And we keep updating this over time.
[00:20:51] Whenever we get new information from them, we keep updating it.
[00:20:54] You can see, it's already logged the exact date.
[00:20:56] Today, I'm recording this on the 28th of July.
[00:20:59] Let's say the next touchpoint we have with the client is on the 1st
[00:21:02] of August. And this file gets updated,
[00:21:04] and it updates the date over here automatically.
[00:21:06] So, the AI knows all of this. And this is why our system stays
[00:21:10] trustworthy in month number 6,
[00:21:13] year number 1. Not, not just on day demo number,
[00:21:16] on demo day. Now,
[00:21:18] finally, coming back over here is, one more thing what I want to do
[00:21:22] is, there's something that has been wrong this entire course.
[00:21:26] And that is the $12 sheet. The offer sheet says $13,
[00:21:30] the front desk rule still says it is $12.
[00:21:34] But the register has been charging $13 all along,
[00:21:37] but nobody has fixed that sheet. The system,
[00:21:41] as we asked it previously, is still thinking that it is either $12 or
[00:21:45] $13. So what we need to do is,
[00:21:48] let's copy this prompt over here, and simply just paste it.
[00:21:52] And watch the order of fixing it.
[00:21:54] The order is the lesson, guys. Of how it actually opens,
[00:21:58] reads the files, and actually fixes all of the files where that information is
[00:22:03] there. It's still sitting in different folders and parts of this business
[00:22:08] brain over here. So it reads the rules, number one.
[00:22:14] We can see, it is drafting the fix.
[00:22:17] The exact file, the exact line,
[00:22:20] the old text, and the new text,
[00:22:23] with evidence. Let's just give it a second.
[00:22:25] And while it is still creating it, we can open it. Let's give it
[00:22:29] a run and let's see what it is saying over here.
[00:22:34] And then it stops. Now what is pending is the owner
[00:22:38] to approve it. Another proposed repair.
[00:22:42] And this is all information that we can always refer to guys.
[00:22:45] Now, now we don't have to keep things in our own head because we
[00:22:49] ourselves do not want to repeat the same mistakes. Mistakes that our clients are
[00:22:53] making, which, which is everything is,
[00:22:56] they are stuck in their head, this conflicting information.
[00:22:58] We don't want to withstand this. Imagine a couple of months from now,
[00:23:02] you have 10 different clients. You cannot,
[00:23:04] we cannot expect ourselves to remember these little details and blockers for each and
[00:23:08] every one of our clients. But we make sure that the AI remembers it.
[00:23:13] And finally, the owner will approve it. And then this file.
[00:23:15] Changes, proposes,
[00:23:18] and so on. And that order then becomes permanent.
[00:23:21] And it is the exact reason why the owners trust us once again.
[00:23:34] Now, on to something that I'm really, really excited about.
[00:23:38] What we have is, it's your turn for real this time.
[00:23:42] I'm sure you've noticed that over the past couple lessons,
[00:23:46] we've quietly been moving with a folder over here called Company 2.
[00:23:50] And I'm sure a couple of you have wondered, like, are we ever going
[00:23:52] to use this? Is there something we're going to do with it?
[00:23:55] And now is the time. So what we have over here is a hypothetical
[00:24:00] company, once again, called Brassline Mechanical and Plumbing.
[00:24:05] They're an HVAC and plumbing company in Tulsa,
[00:24:08] a different industry, and actually much more messier files.
[00:24:13] No completed examples to lean upon.
[00:24:15] And that discomfort that you just felt listening to that.
[00:24:20] That is the course working. Because now what we want to be doing,
[00:24:23] guys is force you guys to start taking action and move
[00:24:27] forward. By yourself. Truly.
[00:24:31] Cut the entire
[00:24:39] section after that is the course working.
[00:24:42] I'm going to retake it. After the course is working.
[00:24:45] So after working, cut it off. And this is exactly
[00:24:50] what a real client feels like. Same loop,
[00:24:52] but this time alone, where what you need to do first is build the
[00:24:57] business brain, then produce the report,
[00:24:59] interrogate it, save your checks,
[00:25:02] and the self-check file in that folder tells you how you can judge your
[00:25:07] own work. and do the thing guys that makes it real post
[00:25:12] a biz Cut
[00:25:23] it, I'm redoing it from,
[00:25:25] then do the thing that makes it, uh,
[00:25:27] real. I'm retaking that. And then,
[00:25:31] do the thing that makes it real. When your business brain and
[00:25:35] report exists, post it in the community,
[00:25:38] not for applause. But because shipping to witnesses is what makes people
[00:25:43] actually finish. And that is where the capstone review happens.
[00:25:48] So, we're finally at the end of this lesson. Step back and look at
[00:25:53] what you can really do now. We've gone from starting off installing
[00:25:57] Cloud Code and Cursor in our computer,
[00:26:00] to now being able to build the entire business.
[00:26:04] A brain from raw human mess.
[00:26:08] Make it produce the exact thing that owners pay for and feed it
[00:26:13] what clients actually send. And interrogate it.
[00:26:16] Go deep into it. Until you
[00:26:20] trust your own system. Because guys,
[00:26:23] this is your own operating system. What I've made for myself over here which
[00:26:27] is you can see my business brain is what I've really created by stress
[00:26:31] testing it. To the extent that I truly trust it.
[00:26:35] And you want to get to the same stage where eventually you want your
[00:26:39] business brain to automate your emails,
[00:26:42] your outreach, your content,
[00:26:45] client communication. Have the system capable of taking in customer support
[00:26:49] tickets and fixing the product your agent needs.
[00:26:52] Your systems by itself and reporting back to the clients.
[00:26:55] So now what you've done is you, even if you hire developers later on,
[00:26:59] or you do it all by yourself, you've removed yourself or the need to
[00:27:02] hire developers, which increases your profit margin means you don't have to manage employees
[00:27:07] anymore and you have more time to focus on life,
[00:27:10] relationships, growing the business and so on.
[00:27:16] That exact
[00:27:21] feedback loop with the business brain is the product,
[00:27:25] not the prompts, not the dashboards.
[00:27:28] And when you can run that loop on a company you've never seen,
[00:27:32] you're about to, which you,
[00:27:35] redo it from when you're about to run the loop.
[00:27:37] Cut that section, I'm retaking it.
[00:27:40] And when you can run that exact loop on a company that you've never
[00:27:44] seen before, for instance, over here on Brassline,
[00:27:48] you're not a student anymore. You're a person that business owners genuinely
[00:27:53] hand their keys over, and a true AI operator,
[00:27:55] and you're going places. We have a final 15-minute task in the
[00:28:00] practice and help tab over here. I want you guys once again to go
[00:28:05] over it, and hope it's been really fun guys so far,
[00:28:08] and this is just the start. Make sure to go through every single lesson,
[00:28:13] do retakes, ask any questions that you might have,
[00:28:16] and this is us only touching a fraction of what is truly possible
[00:28:21] with the business brain. See you in the later sessions.