YT YouTube2Text

YouTube2Text → Transcripts → Dame 25 Minutos y Te Doy 1000 Horas De Conocimiento Sobre Claude

Dame 25 Minutos y Te Doy 1000 Horas De Conocimiento Sobre Claude — Transcript

by Ramiro Cubria · 6,141 words · 881 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00I use the cloud absolutely every single
0:01day of my life. I use it, my team uses
0:03it, and I'm not exaggerating when I
0:05tell you that between us all, we've
0:07logged thousands of hours using this
0:09tool. With the cloud, I create 95%of my
0:12content, manage my ads, analyze my
0:14sales, and I even built a team of AI
0:16employees that works while I sleep. And
0:19that system runs a business that bills
0:20hundreds of thousands of dollars per
0:22month. Now, here's the thing. Most
0:24people use the cloud at level one, and
0:26there are five levels. At level one,
0:28you're a tourist, you use the cloud as
0:30if it were Google, and at the CCO level
0:31, you're the owner. You have systems
0:33that work by themselves, correct
0:35themselves, and report to you while
0:37you're even sleeping. And the
0:38difference between level one and the
0:40CCO level isn't one of productivity,
0:41it's one of money. It's literally the
0:43difference between asking AI for
0:45recipes and having a business that
0:47mostly runs itself. I was saying. In
0:49this video, I'm going to explain the
0:50five levels one by one. How to know
0:51where you are and move to the next one.
0:53All step-by-step with real examples
0:55from my business. So grab something to
0:57write down because this is going to be
0:58one of the most valuable videos you'll
1:00see on this channel. Let's get started.
1:01First level, the tourist. What is a
1:03tourist? A tourist is someone who uses
1:05the Cloud as if it were Google on
1:07steroids. They come in, walk around,
1:09ask questions, take a photo, and leave.
1:10They open a chat, have a question, copy
1:12the answer, and close it. They say, "
1:14Write me this email, give me content
1:16ideas, write me this script, summarize
1:18this text, recommend gifts for my
1:20mother." That's level one. And why is
1:22it a problem? Because tourists always
1:24get robbed, always. In all the big
1:26cities in the world, tourists pay three
1:28times more for a taxi, eat at the most
1:30popular restaurants, which are usually
1:32not that good, and the locals never go.
1:34They buy crafts, trout, and fall for
1:36very common scams because they don't
1:38know anything. Well, it's the same with
1:39artificial intelligence. Tourists pay
1:42Cloud with crappy questions and prompts
1:44, and Cloud returns generic and crappy
1:46responses. And then those same people
1:49are the ones who say, "No, AI gives me
1:50generic responses, it answers me with
1:51any number of things, it makes bad
1:53scripts, it makes a lot of mistakes, it
1:54says information that isn't true." And
1:56let's see, the reality is that they're
1:58right, but the diagnosis is wrong. AI
2:00isn't bad. The problem is that Cloud
2:01knows absolutely nothing about you.
2:03Every time you open a new chat, you
2:04start from scratch. It's like having an
2:06employee with amnesia who forgets
2:07everything. You always have to explain
2:09who you are, what you do, how you like
2:11them to work, how you like them to
2:12write, and give them context again, and
2:14that's extremely exhausting. But anyway
2:15, how do you fix it? With three things.
2:17First, customization. You go to
2:19settings and write the basic
2:21instructions. Who are you? What do you
2:23do? What are you going to normally ask?
2:25And how do you want me to respond? In
2:26my case, it would be something like,
2:28I'm Ramiro, a content creator, I have a
2:30social media business, I help someone
2:32sell more through social media using
2:34artificial intelligence and organic
2:36content on Instagram. I'm going to ask
2:39you many times to analyze my
2:40competitors 'profiles, give me ideas
2:42for script content, analysis, copy,
2:43metrics, respond to me directly with
2:45short sentences, without beating around
2:47the bush, and also in Argentinean
2:48Spanish. And that's it, with each new
2:51chat you have, at least start knowing
2:52who you're talking to and how they
2:54should respond. It's something very
2:56basic, but it makes a world of
2:57difference. Obviously, in the basic
2:59instructions, what you can also do is
3:01dictate to the cloud all the context
3:02you can with the greatest amount of
3:04detail and that same transcription,
3:06which is literally the post talking to
3:08the computer, you just paste it into
3:10the basic instructions. Again, this is
3:12already a huge difference, it's a huge
3:15context. Second, the memory, activate
3:17it in capabilities and Cloud starts to
3:19remember what you talk about with him
3:20between chats. So, first you already
3:22gave him context, but now the memory
3:24will be cross-referenced between all
3:25the chats. That means that he will
3:27start to remember all the conversations
3:28you have and will add them to the
3:30context. And an extra trick, you can
3:31tell him, "Ask me questions to get to
3:33know me better." There Cloud will start
3:35to ask you questions and you can even
3:36answer him via audio if you feel like
3:38writing in a separate chat, because
3:39remember that now the memories between
3:41chats work and add them to the context
3:42and will continue to build that memory
3:44base about you. Third, pinned chats. If
3:47you have a task that you always repeat,
3:48for example, answering emails, always
3:50use the same chat and leave it pinned
3:51at the top so that you don't lose it at
3:53the bottom. The idea is that you use
3:54that chat as much as possible and never
3:56use a new one. You can correct it in
3:58that chat. I like this shorter, this
3:59more formal, this more informal. When
4:01you respond to this person, respond
4:03this way. When I send an email to an
4:04employee, send it this way. Great
4:06detail with the feedback you're giving
4:08them. That chat will be trained and
4:10pinned right up there so that you can
4:11go in and not have to start a new chat.
4:14Although Cloud has a memory between
4:15chats, it's better to use the pinned
4:17one that already has all the
4:18information that you liked how it
4:19delivered it to you. It's more
4:21practical, it doesn't have to keep
4:22searching, it won't get confused with
4:23other information that it's looking for
4:24in all your context, in other words, I
4:26prefer that you use that chat. Now, how
4:27do you know if you're at level one?
4:29Look at your recent chats. If they're
4:31all short chats of five or 10 messages
4:33where you asked something and left,
4:35you're a tourist. Period. Now, all this
4:37is great, but memory has its limits.
4:39You can't fit your entire business into
4:41a memory. And that's where level two
4:43comes in. Level two is the coach. What
4:46is it? Well, the coach stops visiting
4:49Cloud and starts training it. They stop
4:51using it as a chat and turn it into
4:53their workspace. And the key tool here
4:55is projects. What is a project? Well,
4:58it's a folder with a brain. You load
5:00all the information about a topic into
5:02it, the documents, the transcripts, the
5:04data, the instructions, and each chat
5:06that you open within that project
5:08responds to you with prior context on
5:09top. That means that it doesn't start
5:11from scratch, it starts knowing
5:12everything. You might ask me how a
5:14project is different from the memory
5:15that you have between chats. The memory
5:18saves summaries. The project saves all
5:20the textual material. When you talk in
5:22chats, Cloud doesn't save the entire
5:23conversation. Imagine if they had to
5:26save the entire conversation of all the
5:27users, all the conversations wouldn't
5:29have enough data centers in the world
5:30to store so much information. For
5:33example, from a conversation you had
5:34with Cloud, he saves something like
5:36Rama is a content creator, he creates
5:37scripts, he likes direct and
5:38straightforward answers. The other day
5:40we did a summary about something, but
5:42he saves summaries. And on top of that,
5:43Cloud decides what to save and what not
5:45. And sometimes he can leave out an
5:46important detail. If you explained the
5:48entire structure of a script in a chat,
5:50he may not have saved it in great
5:51detail in his memory, as you wanted.
5:53There is no summary in the project. All
5:55the textual information you give him,
5:57he saves all the textual information
5:58and reviews it before giving you an
6:00answer. Cloud has all your information
6:02saved word for word in the project. For
6:04example, what I uploaded to him were
6:06all the transcripts of all my YouTube
6:08videos. So when I want to put together
6:09a new script, he creates a perfect base
6:11for me adapted to my voice and my
6:13identity. That's the difference, before
6:15answering me, he reviews all the
6:16transcripts of all my past videos and
6:18then gives me the structure. Why is
6:20this a game changer? Because this is
6:21where the guide stops writing generic
6:23goo. Let me give you my real-life
6:24example. I have a cloud project for
6:26each content format that I make. I have
6:28my carousel project, my reel script
6:30project, my YouTube video project. And
6:33what does each project have loaded? The
6:34transcripts of all my previous content,
6:36my script structures, my communication
6:38rules, and all the data on what content
6:40worked best for me to sell. So when I
6:42ask it to write me a script, it doesn't
6:43just write me some generic crappy
6:45script that looks like AI. It writes
6:46exactly how I write, because it has
6:48thousands of authentic scripts loaded
6:50with my voice. Even back in the days
6:52before AI existed, I've sent it all the
6:54transcripts of every video I've ever
6:56made. And mind you, this isn't just for
6:57content. If you have an online store,
6:59you upload your products, your sales,
7:01your customers' frequently asked
7:02questions, and you ask it, for example,
7:04"Create product combos for me based on
7:05everything that has been purchased
7:07together historically." It will respond
7:09with your data, not by making things up
7:10. If you have a service business, you
7:12upload all your transcribed sales calls
7:14and it will tell you exactly why people
7:16are buying from you, what the common
7:17objections are, and what pain points
7:19you should address in the content to
7:21attract more qualified customers. The
7:23logic is always the same. All the
7:25information that Cloud has saved is
7:26information that you never have to
7:28explain again, and the more you upload,
7:29the better. Now, how do you do it? Easy
7:31, you create the project, you give it
7:33instructions, what this project is,
7:35what you want to achieve, how it should
7:37respond, and then in the context
7:38section you upload as many documents as
7:40possible, PDFs, texts, spreadsheets,
7:42books, customer databases, whatever. If
7:44you dedicate 15 minutes to properly
7:45putting together the data for a project
7:47, you will save yourself hours for the
7:48rest of your life and you will have
7:49much better answers. Now, how do you
7:51know if you are at level two? Simple,
7:53are you already using projects or not
7:55yet? If you don't have a single project
7:56set up, you are still a tourist and
7:58tourists get robbed, remember that. Now
8:00, pay attention because up to this
8:01point, cloud only thinks better, writes
8:03better, responds better, but there is
8:04something it still does not do and that
8:06is work. For example, you ask it to
8:07write an email, it gives you the text,
8:09you copy it, paste it in your inbox and
8:11send it yourself and that is the point.
8:12Cloud does not do it, you are doing it.
8:14You take the information, copy it,
8:16paste it, but you are not automating
8:17tasks yet. Let's say, you trained a
8:19virtual employee, but you are the one
8:21who ends up executing. It only tells
8:23you what to do. And that's what changes
8:25at level three, which for me is the
8:27level where those who only play with AI
8:29and those who make money with AI are
8:31separated. And now, level three, the
8:34boss. What does a boss do? A boss
8:35doesn't do the tasks, he delegates them
8:37. And this level is exactly that, Cloud
8:39executing tasks for you, not telling
8:40you how to do them. There are two tools
8:43here, connectors and skills. Let's
8:45start with connectors. What are they?
8:47They're connections between Cloud and
8:49any other platform, for example, Gmail,
8:51Google Drive, Google Calendar, Notion,
8:52whatever tool you want. Go to settings,
8:55connectors, and plug them in. Why does
8:57this matter? Well, because the moment
8:59Cloud is connected to your tools, we
9:01can ask it things like, "Review all the
9:02emails that have come to me in the last
9:0424 hours and respond to all of them the
9:06way I want you to respond, for example.
9:07" And you could tell it something like,
9:10"If someone asked me to schedule a
9:11meeting, reserve a slot on my calendar
9:13and send them the meeting invite." And
9:15then Cloud goes and does it, it doesn't
9:17explain to you how to do it, it goes
9:19and executes it. That's the difference
9:20between a daytime assistant who gives
9:22their opinion and someone who is a
9:23virtual employee. I'll give you a real
9:25example of how connectors can be used.
9:27There's a connector that's for Facebook
9:28Ads. You literally look for it in the
9:30connectors section and it says Facebook
9:31Ads. There you can log in with your
9:33meta account and you can tell it, for
9:34example, adjust the budget for the
9:35campaigns. Turn off the campaigns that
9:37have a cost per result greater than
9:39that. Identify audiences that are
9:41burned out and new audiences that I can
9:43test and install within my ad sets.
9:45Well, I also told you about the example
9:46of emails. You can ask it, "Read me all
9:48the emails and reply to all the ones
9:49from the last week." Or, for example, I
9:51have a connector for my Instagram
9:52account. You connect Cloud to your
9:54Instagram account and tell it, for
9:56example, "Create a detailed report of
9:57my content from the last month with all
9:59the retention metrics. Also tell me
10:01which topics generate the most interest
10:03, which content generates the most
10:05comments, and which are all my winning
10:06hooks and it simply runs and brings the
10:08report. Any of these things used to
10:11require hiring a person to do, but
10:13today Cloud does it for you based on
10:15the connectors, period. And then there
10:17are the skills. What is a skill? It is
10:19a mini instruction manual for Cloud to
10:21do a task always the same way, always
10:23the same, and always well. You document
10:25once how something is done, how you put
10:26together your scripts, your documents,
10:28your presentations, with what structure
10:30, with what rules and you upload it to
10:31Cloud as a skill. From there, Cloud
10:33repeats it perfectly every time. For
10:36example, I have a script skill where I
10:38have my entire scriptwriting
10:39methodology saved in the skill and I
10:40literally write a mini command in the
10:42chat before asking it to do a Script,
10:43with that mini command you are already
10:45giving it a whole series of previous
10:46instructions that you loaded again once
10:48. It's as if I told you to condense a
10:51mega prompt into a tiny mini line that
10:53you type before your previous prompt
10:55instructions. It's impressive. You have
10:58to use skills again. And since you have
10:59your entire methodology loaded, for
11:01example, in that script skill, it
11:02doesn't just pull out any script, it
11:04pulls out my script. And since it has
11:05the entire explanation of how to script
11:07and also my context, because let's say
11:08I can use the skill within a project,
11:10it's impossible for the script to sound
11:11like someone else's, impossible for it
11:13to be generic, it's going to sound the
11:14same as me. Suppose I use the script
11:17skill within my project that has the
11:19entire historical transcription of my
11:20scripts. There you go, I won. Now, how
11:23do you know if you're at level three?
11:24Ask yourself this. Does Cloud tell you
11:26what to do or does Cloud do things for
11:28you? If the answer is the first, you're
11:30not the boss yet. Now, up to here,
11:32Cloud works with the tools. that you
11:34connected to it, that already exist.
11:36But what happens when the tool you need
11:38doesn't exist? Well, you build it and
11:40here we enter level four, which is
11:42called the manufacturer. What is the
11:44manufacturer? It's the person that
11:46Cloud uses to manufacture its own tools
11:48. For example, custom computer programs
11:50. Before you tell me," But I don't know
11:52how to program. "The answer is, neither
11:54do I. But it's not necessary because
11:55that's why Cloud Code exists. What is
11:57Cloud Code? It's cloud, but focused on
11:59building. You tell it in Spanish," What
12:01do you want? For example, I want a web
12:03page with the face of a bear and a
12:05button that says, I want my free bear.
12:06I want a tool that shows me this data.
12:08I want a content dashboard that shows
12:10me my Instagram metrics in real time. "
12:12It builds it. Obviously, it throws you
12:14different versions and you tell it,"
12:15Change this, change the color of this
12:17title, this button takes me to such
12:18place. "But you build a custom program
12:21for yourself and literally in a moment
12:23you have something working that used to
12:25cost you thousands of dollars because
12:27you had to hire a programmer. Why? This
12:30is crazy. Because generic tools are
12:31made for the average person and you
12:33don't need what the average person has,
12:35you need your own. Let me show you, for
12:37example, what we build without
12:38programmers, 100%with AI. We created
12:40this platform to create videos with
12:42artificial intelligence where I upload
12:44a script or my spoken audio and the
12:46platform generates an AI clone, a video
12:48of me, but it's not real, it's me. A
12:51clone with artificial intelligence.
12:52It's crazy. It's a media program,
12:54literally the cloning platform. You
12:55train the platform with a video of you
12:57speaking for 10 minutes and you already
12:59have your clone trained to record
13:00content without having to record
13:02content. Or for example, you built a
13:03content dashboard to see the metrics of
13:05my content in real time, even see how
13:07many sales each reel I uploaded to
13:08Instagram brought me. And we also made
13:10an extension of Google Chrome exports
13:12all the information from Instagram
13:14messages so we can find out why people
13:16buy from me and what they ask me the
13:17most. None of that existed before. And
13:19if you wanted it to exist, you had to
13:21pay $ 3,000 per month full time or $
13:233,000 to $ 5,000 per month, full time
13:25to a programmer to build all that for
13:27you. Today, with Cloud Code, you have
13:29it all figured out, and all of that was
13:31custom-made because that's exactly what
13:32was needed. And how would you start?
13:34Well, don't start out wanting to build
13:36a gigantic program, you'll get
13:38frustrated, but start right away, a
13:42calculator for your business, a page to
13:44put your resources, a dashboard to see
13:46your numbers. You see something
13:47specific well-explained in Cloud Code
13:49and little by little it increases in
13:50complexity. The skill you're going to
13:52learn to develop is not knowing how to
13:53program, but knowing how to ask
13:55correctly. Now, how do you know if
13:56you're at level four? Look at your
13:57business now and ask yourself, is there
13:59a tool that didn't exist on the market
14:00before that would help you? are we
14:02using that we have manufactured with AI
14:04? If the answer is no, you are still at
14:06level three and you are not a
14:07manufacturer because the manufacturer
14:09does not rent tools, they create them.
14:11And now, if level four seemed tough to
14:13you, hold on tight because level five
14:15is where it gets really absurd. Level
14:17five is the final boss. I call him the
14:20owner. What is the owner? Well, he is
14:21the one who no longer uses AI. The
14:23owner designs AI systems that work by
14:24themselves. He manufactures employees,
14:26he does not manufacture tools. The boss
14:28delegates tasks, but the owner owns the
14:30entire factory. And this is not theory
14:32or a nice concept for this video. This
14:34runs in my business today, every day
14:35and it is a business that currently
14:37invoices half a million dollars a month
14:38. And I will show you piece by piece.
14:40The factory is called OpenCl. What is
14:42OpenCl? It is an artificial
14:44intelligence brain that lives inside
14:46your WhatsApp. It can live in many
14:48places. It can live directly from your
14:51computer terminal, but you can
14:52communicate with it by Telegram, by
14:53WhatsApp, by Discord, by the
14:55communication platform you want. I
14:57choose WhatsApp because we all use
14:58WhatsApp all day long and I use it like
15:00any other contact on my phone, nothing
15:01more. I can send it photos, audio, text
15:03, videos and it understands everything.
15:05I write to it as if I were writing to
15:07any person. The difference is that
15:08there is no person on the other side.
15:09There is an agent with access to tools
15:11that can execute tasks, can create
15:13other agents, can run automations and
15:15can return anything you want right
15:16there in your WhatsApp chat. I mean,
15:18it's insane. And why, for example, on
15:20WhatsApp and not on a futuristic NASA
15:22dashboard, on a Jarvis? Well, because
15:23WhatsApp is your life, that's where you
15:24manage everything, your business, your
15:26personal life. So, the experience of
15:27having AI employees, you want to have
15:29them where you talk to your normal
15:31employees. If I'm on the street, for
15:32example, I send an audio to Mora, who
15:34is an artificial intelligence employee
15:36who is in charge of my content and when
15:37I get home I have the variants
15:39developed. If I want to know how things
15:40are going metrics, for example, I ask
15:42Bruno, who is the data nerd, and he
15:43answers me and yes, I give them names
15:45and you will understand why. And so
15:48that you see that this is not a lie,
15:49here I am in the OpenClock chat and I
15:51am going to tell him, give me all the
15:52metrics of my Lead Magnet ads and give
15:54me a table where we order from 1 to 10
15:56the best historical ad sets of that
15:57campaign. And give me all the metrics
16:00of those ad sets. There I sent it to my
16:02OpenCl which will redirect it to Bruno,
16:03who is the data reader and now he will
16:05give me all the data. Notice that I
16:06sent him an audio. Let's wait a moment
16:08and then he will answer me. Now, how
16:10does this that I just showed you on the
16:11back work? Three pieces. And pay
16:13attention because here is all the magic
16:15. First, the subagents. You have the
16:17master Open Cloud and that master can
16:19create other agents that are the
16:20employees. Each employee is created
16:23separately with their own instructions,
16:25their own process and their own memory.
16:27It's like each one has their own office
16:28, they don't cross paths, they don't
16:29step on each other, and the one who
16:30manages them all is the master. Well,
16:32look, here he answered me, not only did
16:34he answer the audio, but he also sent
16:35me an Excel database. He tells me,"
16:37Historical Magnet Lead, top 10. "Look
16:39at what he tells me, it's crazy what he
16:40tells me. He tells me," How much was
16:42spent historically, how many records
16:43did I have? How many records did I have
16:45that earn more than $ 2,000 per month?
16:46How many agendas? How many sales? How
16:48much cash collected? What was my ROAS?
16:50"And well, here inside is the database
16:52with winning ads. This is incredible.
16:53This was given to me by Bruno, the data
16:54ner. I mean, it's impressive. And the
16:56best thing is that they can be
16:57concatenated. You can tell him, first
16:59this employee does this thing, then
17:00another one comes along, takes what
17:01that one did and does something else.
17:03And why does this matter? Because if
17:04you ask an agent for everything, at
17:06some point it gets stuck and your
17:07process breaks. But with linked agents,
17:09you can make an orchestra of employees
17:11working in unison without anything
17:12breaking. Two, APIs. An API is the
17:17official gateway to a service. It's
17:21exactly the same as when a new employee
17:22joins your company and you give them
17:24access: their corporate email, their
17:26database access, their company email
17:28and password. Well, here instead of
17:30giving them an email and password, you
17:32give the employees the API key. For
17:35example, I give Bruno, who is a miner,
17:37the API key so he can access my
17:39Instagram, my customer database, my
17:41email, and a bunch of services and
17:43things that I use. I also give Fede,
17:46who is the one who uploads the content,
17:48access via API so he can upload my
17:49content to social networks. I also have
17:51a digital trafficker that has API
17:52access to Meta's Business Manager to
17:54modify campaigns or increase budgets,
17:56rotate creatives, and do whatever else
17:58I need to do. And each employee has the
18:00access rights for their position, no
18:01more, no less. And finally, the third
18:03point, the crons. A chrome is a
18:05scheduled task that runs by itself
18:06every x amount of time, every 30
18:08minutes, every day, once a week, once a
18:10year, once a month, whatever. It's the
18:12employee's work schedule, but the good
18:14thing is that you can tell it to run a
18:15task every 5 minutes forever. It'll do
18:17it forever, without resting. And for
18:19example, I don't have to tell Bruno,"
18:20Hey, send me the metrics. "I tell Bruno
18:22," Hey, every day at 10 in the morning,
18:24send me the metrics. "and he
18:25religiously sends me the metrics at 10
18:27in the morning every day without fail,
18:28without you chasing him, even if you
18:30never answer him, even if you're
18:31sleeping, he'll send them to you. And
18:33now let me introduce you to the
18:34permanent staff because the whole thing
18:36is crazy. I have Bruno, who is the data
18:37nerf, he pulls the daily metrics from
18:39all the platforms and gives me a report
18:41and on top of that he downloads all the
18:43transcripts of all my competitors '
18:44videos and tells me which content
18:46angles are working the most for them.
18:48It's crazy. I have Nico, who is the
18:49strategist who takes that data. He has
18:52all the transcripts of my content and,
18:53based on the data of what's working for
18:55my competitors and what's working for
18:56me, who is also connected to my sales
18:58CRM, tells me what content ideas I
18:59should cover next week or, more than
19:01ideas, what topics or angles I should
19:02cover in the next week of content. Then
19:05I have Mora, who is the ideator, who
19:06takes Nico's data and produces all the
19:08ideas. Then there's Thano, the
19:10scriptwriter who takes Mora's ideas and
19:12scripts all the content, who also has
19:13my entire video history loaded as
19:15information so he can script exactly as
19:16I speak. I also have Fede, who is the
19:18community manager who takes the
19:20recorded and edited content and uploads
19:21it. And finally, I have Gina, who is
19:23the boss who oversees that all the
19:24other digital employees have delivered
19:25everything on time and in the right way
19:27. And if something goes wrong, she lets
19:29me know, but like the example I just
19:30showed you, where I asked Bruno to send
19:32me the metrics in real time, I can talk
19:33to any of them and they will respond.
19:35And also, Shina, who is the content
19:36manager, oversees that everything has
19:37been delivered on time and in the right
19:39way. And if something goes wrong, she
19:40lets me know. Now you realize what was
19:41created there, a closed loop. The data
19:43decides the calendar, the calendar asks
19:45for ideas, the ideas ask for scripts,
19:46the scripts are published, the
19:47publication generates new data and a
19:48boss makes sure that no one falls
19:50asleep. It's an entire business
19:51ecosystem running on WhatsApp alone.
19:53And do you know how long it takes me to
19:54create one of these employees? One
19:56minute. What's more, right now I'm
19:57going to show you an example of how I
19:59asked him to create an employee with
20:00artificial intelligence live the other
20:01day, and you'll see it. Let's play the
20:03clip. Now we're going to create an
20:06employee whose role is to manage our
20:08advertising campaigns. I need you to
20:11create a sub-agent to review the
20:13remarketing advertising campaigns,
20:15which are the campaigns called Mofu.
20:18Also, review it carefully and tell me
20:21the exact name of the campaign. I need
20:23him to review every single one of the
20:24ad sets within that campaign on a daily
20:26basis. I need him to turn them off if
20:29the frequency of those ads is greater
20:31than 10. And I also need him to send me
20:34a complete report every day at 10 in
20:36the morning and at 10 at night with all
20:38the metrics of the advertising
20:40campaigns, reach, CTR, cost per
20:42campaign, frequency, and all the ad
20:44metrics you can think of. If you can,
20:48send me a complete report every day
20:49from 10 a.m. to 10 p.m., so that the
20:51employee can create it and we can
20:53confirm it here when it's ready. Okay,
20:55and then we'll send them an audio and
20:57then they'll start creating an employee
20:59who will perform that task every day
21:00and will do exactly what I asked. We
21:03can see the bubble indicating that
21:04they're responding to me. Let's just
21:06wait a little while and then they'll
21:08send me back the completed employee.
21:10Okay, and then we can see that they
21:12responded to me, it says," Done, Mofu
21:14employee created and working. "This is
21:16the exact campaign that's currently in
21:17effect. It says ID, it says active
21:19status, it tells me," I found that the
21:21old employee was looking at this
21:23campaign, but uh, that campaign is
21:24already paused. I fixed it so that you
21:26only see the active campaign. It looks
21:28like this. It checks the ad sets, not
21:30the individual ads. It checks the
21:32lifetime frequency every hour. If an
21:34active set exceeds frequency 10, which
21:36is an example we gave, it automatically
21:38pauses it. It sends a full report every
21:40day at 10 a.m. and 10 p.m. It includes
21:42spend, reach, impressions, frequency,
21:44CPM, clicks, CTR, reach, CPC, agendas,
21:46uh, cost per agenda, conversations,
21:48cost per conversation, responses and a
21:50breakdown of all the ads. If the
21:52employee fails, it also notifies you.
21:53Look, it's also Covani, it chases. I
21:55tested it against vivo. It checked 23
21:57sets. There are 21 active and none
21:59exceeds 10. The current maximum per set
22:00is around 2.41. The first automatic
22:02report arrives today at 10 p.m. And so
22:04you can see. that works, we're going to
22:07tell him, send me the first metrics
22:09report right now. And there I'll ask
22:12him and I'll get the metrics report in
22:14a minute. Done. Here it sends me the
22:16complete report. It says, "Complete
22:19report today, spent in total campaign
22:21time." It tells me, uh, by ad set. It
22:24gives me all the metrics. What I can do
22:26is ask it to give me these metrics in
22:28the form of a dashboard, that is, to
22:29give me a nice, complete dashboard.
22:31What I can even do is connect it to
22:32Cloud Code and have it create a super
22:34aesthetic dashboard, like the ones that
22:35the cloud creates. But this is
22:36basically so you can see how it works.
22:38I mean, I literally sent it an audio
22:39via WhatsApp, it created the employee
22:41and the employee reviews the ad sets
22:43every hour and does what I told it to.
22:44Now I'm going to tell it to turn it off
22:46because it's clearly useless for me to
22:47have, I did it as an example. Turning
22:49it off was just an example to show how
22:50I make AI employees and we're going to
22:52see how it pays. I mean, if you want to
22:53fire one of these employees that you
22:55want, you just have to tell OpenClock
22:56to turn them off via WhatsApp and it
22:58turns them off for you. And here it
22:59told me to turn it off. I deactivated
23:01the employee who checked frequency and
23:02could pause adsets, the automatic
23:04reports of the 10:00 a.m. and 10:00
23:05p.m. Here I'm going to ask him if he
23:07turned off another one that I do use
23:08and if I need it to stay on. Then I'm
23:10going to tell him, "Hey, just in case,
23:11the Tofu one, which is another campaign
23:12I'm running, you didn't touch it, right
23:14? That one has to keep working as usual
23:18. Just in case I kill them both with
23:19one shot." And I didn't want that.
23:21Let's remember that the guide, just
23:22like a person, can make mistakes. In
23:24any case, they tend to make fewer
23:25mistakes, but they can clearly make
23:27mistakes and precisely to prevent them
23:29from making mistakes, there are times
23:30when we have to make sure and ask them
23:32so that they don't make them, basically
23:33. Now, an important detail is that Open
23:35Clow is the factory, but you have to
23:36plug a brain into the factory, which is
23:38the artificial intelligence model that
23:39runs behind it. And here the rule is
23:41simple, give it the best brain that
23:42exists at the time you are watching
23:44this, because brains are always being
23:45updated. Now, the real difference with
23:47all of the above is the scale. A prompt
23:48solves a task for you once when you ask
23:50it. An employee solves that task every
23:52day without you asking them to do it,
23:54without you following up, and forever.
23:56And they also report to you. And when
23:57you have 6, 10, 20, 30 employees
23:59working for you, connected to each
24:00other, you no longer have loose tasks.
24:02You have an ecosystem. and a business
24:04that starts to walk without you
24:05necessarily having to step on it all
24:07day or review it all day long. Every
24:09time I detect a repetitive function, I
24:11don't hire anyone. I go to WhatsApp,
24:12send an audio, create the employee,
24:14leave it running and maybe add it to
24:15the supervision flow in China so that
24:17no one misses the deliveries of that
24:18employee. How do you know if you are at
24:20level five? With a single question. Did
24:22the AI do something for you while you
24:24slept? If the answer is no, you are not
24:26the owner yet. If the answer is yes,
24:27well, congratulations, you are part of
24:29a very small club. And an important
24:30clarification so that you do not get
24:32overwhelmed. You do not need to reach
24:33level five to earn money with
24:34artificial intelligence. Money is made
24:36from level three. My real
24:38recommendation is this, master levels 1
24:39, 2 and 3, which with a little
24:40perseverance during the week you will
24:42have covered and start building from
24:43there. Whoever becomes a boss has an
24:46advantage over people who are still in
24:48Tourist asking for recipes or stupid
24:50things. Now, let's recap the five
24:52levels. Level one, the tourist. You use
24:54the cloud like Google. Level two, the
24:56coach. Projects with all your
24:57information. Level three, the boss.
24:58Connectors and skills. Cloud executes
25:00for you. Level four, the maker, you
25:02create your own tools with Cloud Code.
25:04Level five, you own it. You create day
25:06employees who work for you. And while
25:07you sleep, which is the reality now,
25:08most people are going to watch this
25:10video, they're going to say, "Um,
25:11that's cool.""And tomorrow he's going
25:12to keep using Cloud in tourist mode
25:13asking stupid questions. Don't be that
25:15person. I chose a level, the one that
25:16follows yours, and I'll level up this
25:18week. It's literally a matter of
25:19sitting down one afternoon. Now, to
25:21help you with all this and save you
25:22years of experience, you have a link
25:24down here to contact me and have all
25:26the information to work with me and ask
25:27me the questions you need. And one more
25:29thing, before you go, everything you
25:31saw, the systems, the products that are
25:32built with this, is exactly where
25:34everything I've been putting together
25:36is going. In the coming weeks I'm going
25:38to be sharing on this very large
25:39channel, so subscribe and activate the
25:40bell because I know you don't want to
25:41miss it. You have a link down here in
25:43the description to contact me and scale
25:45your business or start your business
25:46with artificial intelligence.

About this transcript

This page contains the full transcript of Dame 25 Minutos y Te Doy 1000 Horas De Conocimiento Sobre Claude by Ramiro Cubria, generated from the public captions YouTube serves with the video. The transcript has 6,141 words across 881 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.