YT YouTube2Text

YouTube2Text → Transcripts → Crie Um Agente De IA Para CADA ÁREA DA SUA EMPRESA E VIDA

Crie Um Agente De IA Para CADA ÁREA DA SUA EMPRESA E VIDA — Transcript

by Leonardo Oliveira | Inteligênc[IA] · 5,972 words · 944 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00I've been seeing a lot on social media
0:02lately about people adopting bots, right
0:05?
0:06At the beginning of the year, we had
0:08OpenInterpreter launching the first
0:10agentic bot solution, then we had Hermes
0:12, which were solutions that until then
0:14required a bit of a complicated
0:16technical setup; you had to have a VPS
0:18and everything else.
0:21In August 2026, Grok ended up innovating
0:24by bringing that VPS already built into
0:27the Grokbot subscription itself.
0:30Right after, in September, Meta launched
0:33Movie, which is its own personal
0:35assistant.
0:36Then came ChatGPT with the bots about
0:38three weeks ago.
0:40And now Anthropic just announced they
0:42are going to launch the Tool Use. It’s a
0:45project that has already been released
0:46in the United States and is already on a
0:48waiting list here in Brazil.
0:50But the big question is: why are these
0:52giants betting so much on this type of
0:55solution, and why should you adopt it?
0:58That’s what we’re going to investigate
1:00in depth in this video. So stay here
1:01with me.
1:03But in practice, what do these bots have
1:05that’s different from the normal agents
1:07that were already in these tools?
1:09The first thing I want to point out here
1:11is how easy it is to configure a bot.
1:14Basically, you give it a name, start
1:16talking to it, give it some functions,
1:19and it automatically starts working
1:21proactively.
1:23This is the second point that makes
1:25these bots different.
1:27So, this new type of agent will keep
1:29tracking your applications, it will keep
1:31tracking your brain, in case you have a
1:33brain.
1:35If you don't, check out my old videos
1:37and create one, because it will make a
1:39lot of sense now for those who have all
1:42this context, for those who already have
1:44an AI brain, and this bot keeps scanning
1:46for pending tasks, ideas, or signs that
1:49you could leverage as opportunities or
1:51bottlenecks you haven't seen.
1:54So, it is very useful to have a
1:57proactive agent doing this for you. So,
2:00it’s always keeping an eye on your
2:02emails, your WhatsApp.
2:04On everything you use in your daily life
2:06, and it will seek ways to make your
2:08life easier.
2:09>> I am one of them, Leozinho, Leo's bot.
2:13Imagine a center where every task moves
2:15along and you only unlock what has
2:17stalled. In the Double Your Business
2:19Workshop. November 11th and 12th, you
2:22build yours, point the camera, and
2:24secure your spot.
2:29A recent case that happened to me, which
2:31is where I actually saw value in this
2:33type of solution.
2:35I was in a mentoring session talking to
2:37a mentee, focused on what I was doing,
2:39and automatically the GPT bot checked my
2:41email regarding an issue we were having
2:43with an invoice, since I had changed my
2:45tax address.
2:47It read the emails, understood what my
2:49secretary was resolving, and what was
2:52still needed from my side.
2:54It identified that a document was
2:56missing which I had to send to my
2:57secretary so she could resolve this
2:59pending issue. And it already grabbed
3:02this document because it was in my drive
3:04, sent it to her, and she resolved
3:06something that had been pending for two
3:08weeks in just two days. So, look how
3:11cool that is, right?
3:13Now you can be focused on your daily
3:15life, on your work, doing what you have
3:17to do, and this bot will take care of
3:18all that management for you.
3:21Another very important and different
3:23thing is that each bot will now have its
3:25own internal computer.
3:28And what does that change in practice?
3:31This is known as a VPS, which is
3:33basically a virtual machine that you
3:35rent, let's say, elsewhere so you can do
3:37the things you need to do without your
3:39computer being turned on.
3:42But if each bot has its own internal
3:45computer, you are basically renting
3:47several computers. Basically, you will
3:50have an entire company, a whole new
3:52productive capacity working at your
3:54disposal. Another thing that also
3:57impressed me is the routines.
3:59So, each bot can have more than one
4:01routine and throughout the day it does
4:03the things you programmed it to do,
4:05right?
4:08It is not the same as the routines we
4:10have, for example, in GPT and Claude,
4:12especially in Claude, for example, if
4:14more than one routine triggers at the
4:16same time, it tends to error out.
4:19So, it executes the first one and the
4:21second one it kind of skips, and then it
4:23will only run again on the next schedule
4:26it has. So that ends up getting in the
4:28way a lot, right?
4:30Because sometimes when I turn on my
4:32computer, I want a lot of things to be
4:34updated.
4:35So the bots solve this, because since
4:37they have this virtual machine, they
4:39keep working while you are sleeping,
4:41even if the computer turns off or you
4:43lose the connection to something,
4:45because they are in the cloud, they can
4:47do many more things. And now I'm going
4:50to show you in practice how easy it is
4:52to create a bot. Basically, if you are
4:54on GPT and have the pro account, right,
4:57which is about 500 reais more or less,
4:59you will be allowed to create a BOT for
5:02now, right?
5:03But they’ve announced that soon you’ll
5:05be able to do just like Grock, which
5:07released an entire team of bots.
5:09Basically, GPT will have a "create your
5:11bot" function.
5:13And then, you basically click here, and
5:16it will help you with the setup, right?
5:19I named mine Leozinho, right? "Leozinho
5:21," very much from Minas Gerais, right?
5:23I’m not from Minas, but I like the "
5:25Mineiro" way of being. So I thought it
5:27was a cool name for him.
5:29You can configure his visual appearance
5:31here.
5:32And then, basically, it’s what I told
5:34you, right?
5:35You can call him, you can trigger him on
5:38Slack, and he has his own computer.
5:40In this case, since it’s just one bot,
5:43you’ll only have one, right?
5:46But when you come to a solution like
5:48Grock Bot, you can have an entire team
5:50of bots.
5:51See? Here I have the same maestro
5:54Leozinho, but here he operates
5:55differently. The GPT Leozinho has to do
5:58everything because he’s just one.
6:00Here, since I have the option to create
6:03multiple bots, the Leozinho here is a
6:05maestro, so he’ll be responsible for
6:08creating new bots for me and also
6:10keeping them doing what they should.
6:14So he’s kind of a watchman, a guardian
6:16who will architect everything and manage
6:18these bots.
6:20That’s his only function: to keep this
6:23team of bots running. Got it? To show
6:25you how easy it is to create one, I’ll
6:27create one here in Grock for you to see
6:29: new bot.
6:30I give him a name, or I don’t even need
6:33to name him. I could create it directly
6:35through Leozinho here, but I’ll assume
6:38you’re on your first bot and you don’t
6:40have one yet, right? So, the first thing
6:43you have to do is create an orchestrator
6:45bot that will help you create all the
6:47others, right? I’ll create the "demo"
6:49here, right? Because it’s a demo bot.
6:51Basically, you choose a visual. At the
6:54beginning, you don’t worry too much
6:55about that, right? But I’ll choose this
6:57little star here and give it a little
6:59color. I’ll use my brand’s color, right
7:01? Usually it’s terracotta, but here
7:02there’s only orange. So I’ll create it
7:05here, and it will already start asking
7:07me what it should have, right? Look. Hi
7:09, I’m Demo, I just joined the team. To
7:11start being useful, tell me what you
7:13want me to do around here, right?
7:16And then, basically, it already gives
7:18you some suggestions, but the best way
7:21is for you to tell it here what it
7:22should do, right?
7:25So I can come here, I'll use Raycast,
7:27since I like to use my own dictator, but
7:29you can use its own dictation tool here
7:31, but I'll talk to it, right?
7:33Let me cancel here and do it again.
7:35Option option.
7:37Hi, Demo, how are you? I wanted you to
7:39be the orchestrator of my bots, okay? So
7:41, your first task is to help me
7:44architect a map. Right? You already have
7:47access to my drive, to my brain, you
7:49already know what it means. So, the
7:51first thing is you're going to bring me
7:53an architecture of the first 10 bots
7:55that I can create here in my team. Done
7:58. And then I'll send it here to him,
8:00you'll see he's already going to start.
8:02Uh, here he's already saying that I've
8:05used it a lot this week, look.
8:06Everything great, Léo? Let's go. I'll
8:08start by reading your drive's brain and
8:10what the bot team already has today to
8:12bring you an architecture of 10 bots.
8:15Well, while Demo works on the
8:17architecture issue, I want to show you
8:19the architecture I already have. So,
8:22when I installed this bot architecture
8:25solution, the first thing I thought
8:28about is where a 24/7 bot dedicated to
8:31that function would be useful in my
8:34company. So, first I thought about the
8:36macro sessions.
8:38Starting with the first one, which is
8:40basically what every CEO should worry
8:42about: the company's strategy.
8:45So, the first session I created here in
8:48the bots was the strategic session,
8:50where I have the brain basically that
8:52will take care of everything involving
8:54the construction of the company's brain
8:56.
8:57If you already know me and have seen
9:00videos here, I always say that your
9:02company or you need an AI brain, which
9:05is a single place with all the context
9:08that AI needs to be able to work for you
9:11, right?
9:12So here, little Brain, its concern is to
9:15document things by brain, connect things
9:17that are loose, find bottlenecks.
9:21For example, sometimes a bot creates
9:23something new, leaves it loose in the
9:24drive or creates it in the wrong place;
9:26it will be tuned in 24 hours a day to
9:28know if everything is working.
9:30OK. Got it? So that is its function. The
9:34second one I created was Vision. What
9:36does Vision do? It actually has the
9:38wrong color, because all the strategy
9:40bots here need to have the little red
9:43color, right? This is basically the
9:45color code I created here.
9:47The Vision's role, and I was even
9:49training it here, is to stay aligned
9:51with the company's strategy.
9:54So it will check if everything being
9:56done by my bot team, or by me, or by my
9:58human team, is aligned with the
10:00company's strategic vision, the 3-to-5-
10:02year plan.
10:05So it will take care of all the parts of
10:06the company's brain that contain these
10:08files.
10:10It will keep reading the strategic plan
10:12, checking if there were any changes and
10:13if it can update this document, right?
10:16So, look how cool, you will basically
10:18have an agent focused on knowing if you
10:21are following the plan, if all your
10:23actions, your priorities, your projects
10:25are leading to where you designed, where
10:28you wanted to reach. One of the ones I
10:30like the most and is already being very
10:32useful is the track.
10:33What does the track do in practice?
10:35Basically, it's an agent focused on
10:37customer success, right? So it stays
10:39focused on all the meetings, on all my
10:41students, and checks the feedback, sees
10:44if everyone is using the brain, right?
10:47So it has the success metrics.
10:49Every time it sees any of these things,
10:51it makes clips.
10:54Here's an example: I was in a meeting
10:55with a mentee, he said, "Léo, I already
10:57installed the Brain in this first week,
10:59I already had results, look how I'm
11:01using it."
11:02So it identified the moments in the
11:04meeting where it was reported how the
11:07person was using it. And it made these
11:10clips. It saved these clips in my Drive
11:12. There's an area here, a specific
11:14folder where I save the clips.
11:17And it didn't just do that, it compiled
11:20these clips and created a 60-second
11:22story for me with that mentee's success
11:25story.
11:27And it published it for me, well not it
11:28, but my Instagram agent, which we'll
11:30see in a bit, okay?
11:32But just so you can see, I'm going to
11:33play it real quick. Look how cool,
11:37look. It made this whole video for me,
11:39see, look? Look how cool, with motion,
11:41already with my brand's editing standard
11:44and everything else. So the track,
11:46that's why it's a heart. It focuses on
11:49my client's happiness and it makes me
11:51happy too. So nothing more fair than it
11:54being a heart. Another really cool
11:56strategic bot is Atlas. What does Atlas
11:59do? Atlas is basically an archetype of
12:01my client. It’s not a bot to do things,
12:04it’s a bot to receive things.
12:07So, in Greek mythology, Atlas carries
12:10the world on his back, which is
12:12basically the archetype of my type of
12:14client—that overwhelmed person,
12:16overwhelmed team, or chaotic company
12:19that needs structure and agentic
12:21solutions like this one.
12:24Basically, I trained it with my client's
12:27voice, using all their feedback, forms,
12:29meeting notes—basically, I did a full
12:31audit to create a real archetype.
12:35And then, whatever I produce, whether
12:36it’s a product or content, I can ask
12:38Atlas to evaluate it from my client’s
12:40perspective.
12:42Check out how cool this solution is,
12:43right? So, this is basically my
12:45strategic team, and it can work
12:48individually or in the way I like, the
12:50way I prefer, and what I love about
12:53Grockbot, which is their ability to talk
12:55to each other.
12:57For example, do you see here that Brain
12:59Zin sent a message to Nil?
13:02Nil is basically an agent focused on AI
13:05news.
13:06So, he’s always tuned into X, news from
13:10OpenAI, ChatGPT, and in general, what’s
13:13trending in AI, and they are able to
13:16communicate with each other.
13:19So, in this case, he sent a message to
13:22Nil.
13:23Look how cool, they can talk to each
13:25other, man, it turns into an internal
13:27chat, right?
13:29And he said: "Look, he actually received
13:31a message from Nil, and Nil asked
13:33Brainzinho to choose which part of the
13:36Brain he should save these analyses he’s
13:38doing."
13:40And then Brain went there, received it,
13:42organized it within the Brain, and sent
13:45it back, teaching Nil where his
13:48deliverables will be saved from now on.
13:52Look how surreal this is.
13:53While I was presenting this to you, the
13:56demo already finished its work. Look how
13:58cool.
13:59It creates a whole team architecture
14:01analysis for the bots, look.
14:03It’s clickable, it’s a super beautiful
14:05document.
14:07Every deliverable I’ve received from
14:10Grockbot, man, the files it outputs are
14:12extremely organized, whether it’s an
14:14audit or an HTML file.
14:17It is a bot that truly knows how to get
14:19things done.
14:20Look, it gave me 10 ideas on how I can
14:23build the architecture here, which area
14:25this bot will take care of, its status,
14:27its mission, and what it absorbs, right
14:29?
14:31Look, he even made a super nice
14:33flowchart, well-designed and very
14:36minimalist. I'm really loving what Grok
14:39is delivering, you know? See, gaps and
14:42overlaps. Look how beautiful this file
14:44is.
14:45And the sources it used from my brain.
14:48Look how cool.
14:50So, man, having an AI brain for your
14:52company is surreal, because it will help
14:55not only you and your human team, but
14:57your bot team to work in a harmonic,
15:00organized, and very well-structured way
15:03.
15:04Following along here with my bot
15:06architecture, the second area where you
15:08can place bots is the operational area,
15:10obviously, right? We use them to improve
15:12our lives, to do things for us. So, it's
15:15only fair to have an operational team.
15:18The first one that's helping me the most
15:20is Eco.
15:21What is the idea behind Eco? Eco is
15:23basically a meeting agent. So, it helps
15:26me in the pre-meeting phase. It checks
15:28my calendar and says: "Look, Léo, you
15:30have two meetings today. The last time
15:32you spoke to this person, it was about
15:34this, this, and this."
15:36Or if it's the first time, it will pull
15:38from other sources of information and
15:39prepare a mini agenda for me.
15:42Look, I'll show you one it did the other
15:44day that was really cool, right?
15:46Oh, good morning. Today there's only one
15:48meeting, and its agenda is already ready
15:50in your brain. It's a follow-up for
15:52Brainwork, the app I'm developing with
15:55BN and Adriana, who are my partners in
15:57the project.
15:59Then it gives the link to join the Meet
16:01, the context of what happened so far,
16:04and what I should pay attention to from
16:06now on for this meeting agenda; what
16:08needs to be ready by 8:45 AM, which is
16:11just before the meeting; and the
16:13material in Drive it used as a source.
16:17Look how cool this organization is,
16:18right?
16:19And it doesn't just help me pre-meeting
16:21; it also helps me post-meeting.
16:24So afterwards, it analyzes the meeting,
16:26sees the pending items or the
16:27deliverables I promised, and starts
16:28working on them ahead of time.
16:31This is very useful when I'm with a
16:33mentee, for example, when they say: "
16:35Léo, send me this HTML, send me this
16:37template, send me this skill."
16:39It understands the mentee needs this,
16:41packages everything up—obviously
16:44removing my sensitive data—and sends it
16:46straight to the mentee's WhatsApp.
16:49So this gives me a huge advantage, it
16:51makes my life so much easier, because
16:53it’s very common for me to finish one
16:55mentorship session and immediately jump
16:57into another or go straight to another
16:59meeting.
17:01And this ensures nothing is left pending
17:03, it takes care of this entire process
17:0624/7.
17:07I think you’re already understanding the
17:10real power of having a dedicated agent
17:13for every single thing within your
17:15company's brain, which is massive.
17:17A company has a gigantic flow of
17:19information.
17:21So, the more you have a proactive little
17:23guy to handle this, the better.
17:25Just imagine hiring a human for each of
17:28these functions. First of all, it’s
17:30unfeasible, it’s very expensive, and
17:33it’s also underutilizing human potential
17:35, right? Humans have the ability to deal
17:37with various sectors, but it’s very hard
17:40to have a specialist in just one thing
17:43in your company. With this type of
17:45solution, it’s not that difficult and
17:47it’s cheap if you stop to think about it
17:49.
17:50A 500-real plan, for example, here on
17:52Groq, man, you can have a whole team
17:54like this working all week with credits
17:56, right, without running out of credits
17:59.
17:59And man, it’s less than a minimum wage.
18:02Take the minimum wage in Brazil. It’s
18:041600 and something at the moment I’m
18:06recording this video. I think it’s 1621
18:08to be more precise, but that tends to
18:11increase. In a while it will be 2,000,
18:14anyway, right? So, look how cool this
18:17kind of thing is. The others I have
18:19operational are the flux. Flux only
18:21takes care of knowledge management here
18:23in the brain. So it finds things that
18:25are disconnected, information that could
18:28connect with others and turn into
18:30knowledge, turn into an SOP, turn into a
18:32skill, turn into a process, turn into an
18:35agent. So it’s always keeping an eye on
18:37this loose information.
18:39The follow-up, which is the FOP,
18:40basically its function will be to give
18:42follow-ups, right? Whether with clients
18:44or with the team. So it will have this
18:46dual function. It will be an internal
18:48team agent and also an external agent
18:50helping me with clients. And the DU,
18:52basically will take care of the tasks,
18:55right?
18:56And you can see here that I haven’t
18:58triggered all of them yet, because it’s
19:01really hard to structure everything in
19:03just a few days.
19:05What I’m doing is having Claude Code
19:07interview me on this bot architecture of
19:10mine, and through this interview, I’m
19:12giving clear functions to each agent
19:14until I manage to finish all the
19:16operational files for these agents.
19:20Something very important I want you to
19:22understand about this type of bot is
19:23that they will document things
19:25internally within the tool.
19:27So, Grokbot will have a Grok documents
19:30area and will place information there.
19:33And GPT will do the same, but again, it
19:36goes against the principles of the brain
19:38I create, right?
19:40The brain, which I always talk about,
19:42the idea of the brain is to never leave
19:44important things inside the tools, but
19:47to transfer them to a universal brain
19:49that feeds them all back at the same
19:51time and gives you control over your own
19:54information.
19:55In my case, I chose Google Drive to
19:57store this, but some people want a
20:00dedicated database, their own server,
20:02OneDrive, anyway, it really depends on
20:04each person, on each company, but in
20:06practice, the operation is the same.
20:10When I'm talking about these bots, what
20:12I recommend you do within your brain
20:15solution is create a folder called "
20:17agents" and inside it, you will put a
20:19subfolder for each of these bots you
20:21create.
20:24And inside them, it will only have two
20:26files, which is an expertise markdown
20:28file.
20:30So, in this Atlas case, it has the Atlas
20:32expertise, that is, how Atlas will be
20:34learning over time.
20:36It's a kind of learning memory that it
20:39refines by itself.
20:41And the second file is the soul, that is
20:44, the soul of your agent. What does it
20:46have to do in practice? What are the
20:48roles? What is the tone of voice it uses
20:50, right?
20:52What is its personality, and how can you
20:54call your agent at any moment?
20:57You can actually call it, and it can
20:59talk to you; within Grokbot, for example
21:02, you can choose a different voice for
21:04each of them. Inside the others, it only
21:07has one voice there, and you can't
21:09change it for now.
21:11But look how interesting, if I create a
21:13personality for each one, when I call
21:15this agent, the conversation will be
21:16very pleasant, right?
21:19Basically, I'm talking to an entity, a
21:21human being. So much so that when I look
21:23here at the agents, where do I store
21:25them within my brain structure? It stays
21:28here inside "areas," see? So, inside "
21:30areas," there will be the "agents"
21:32folder.
21:32What are areas within the Alma Brain
21:35structure?
21:37Remember I've already had a video about
21:38this talking about my PARA structure,
21:40right?
21:42How I organize over 70,000 files in here
21:44using only six folders: projects, memory
21:47, areas, resources, inbox, and archived
21:49.
21:50Speaking more specifically about areas,
21:52areas are the company's areas of
21:54responsibility, they are sectors, it's
21:56that which is continuous, which never
21:58deactivates. Projects have a beginning,
22:00a middle, and an end. An area is harder
22:02, right? Obviously, I could have an area
22:05, I don't know, for a podcast studio,
22:07and then I deactivate it over time. I'm
22:10not going to make podcasts anymore.
22:12That can happen, but in this case here,
22:14the agents will always be a part of it
22:16because it's like a real team.
22:18It is in this structure that I'm always
22:21talking about the human-agentic company
22:23, which is a company where humans and
22:25agents work together in the universal
22:27structure that will serve any company,
22:30which is strategy and operation.
22:33Now you have to start looking at sector
22:35sessions, right?
22:36How do you create these sessions in
22:38practice?
22:39When you create a bot here inside
22:41Grockbot, you right-click and you can
22:43move it to a session.
22:45At first, there won't be any sessions
22:47created, so you come here to create
22:49session, and then you can give it a name
22:51, in this case here, demo session, and
22:53it stays inside this channel, let's say
22:56, okay? This makes life much easier.
22:59And if you want to create a shared chat
23:02between your bots, basically you have on
23:04Mac, it's Command Shift N. And then I
23:06select who will be part of this
23:08conversation.
23:10I will make the bots talk to each other
23:11. And if it's on Windows, it's Control
23:14Shift N, okay? Just change command for
23:17control. It will work.
23:19In the sectors part, as my company is
23:21educational, there is the education
23:23sector.
23:25For now, I only have Eduin, who will
23:27take care of the entire pedagogical plan
23:29, both the general pedagogical plan and
23:31the specific pedagogical plan of the
23:33tracks on our platform.
23:35So now I'm creating a cloud code track,
23:37I'm going to create a bot track like
23:40this one.
23:41So if you want to be a part of it, we're
23:43having a Black Friday sale.
23:45I truly recommend you take part in the
23:47Double Your Company workshop.
23:49There will be a lot of cool stuff there
23:51and there will be unique opportunities
23:53for you to be part of our community and
23:54get access to these tracks.
23:56So there's not much of a secret here in
23:58the educational area. Basically, it will
24:00take care of the entire lesson plan,
24:02educational plan, and everything else.
24:04Within finance, specifically, I created
24:07two bots, Good Money and Big Money. What
24:10is the difference between the two? The
24:12Good Money focuses on day-to-day
24:14operational cash. So, cash flow, who
24:17needs to be paid, in short, paying taxes
24:19.
24:21He just takes care of the company's
24:22financial operations, updating the
24:23financial spreadsheet, and all that
24:25rigmarole we're already familiar with.
24:28Big Money does something that's not very
24:30common in a company, right?
24:33If you don't have the capacity to hire a
24:35CFO, that is, a Chief Financial Officer
24:37, someone who actually takes care of
24:40this financial vision and scaling the
24:42company, you can create a bot like Big
24:44Money, right?
24:45Basically, he'll look at the scale,
24:48he'll start finding financial
24:49bottlenecks or financial opportunities,
24:52whether in products, adjusting service
24:54pricing, and so on.
24:56So he'll take care of that, of the scale
24:58, of truly scaling the money and the
25:00business model in a general way.
25:03Now, moving on to marketing, this is
25:05more common; you'll see the bot
25:06community, right? They'll always teach
25:08marketing. So this is all a bit cliché.
25:11I'll just go over it quickly so you
25:12understand. The one I'm enjoying the
25:14most is Gram. It's the one I've been
25:16activating the most lately. What does
25:18Gram actually do in practice? Basically
25:20, it takes care of the Instagram channel
25:22, right?
25:23And this is where a great function of
25:25having these dedicated computers comes
25:27in.
25:28Since each of these bots has its own
25:30computer, Gram will have the computer,
25:33the browser already logged into my
25:35Instagram. Why is this cool? Because
25:39here in his routines, one of the
25:41routines he has is to stay in the
25:43Instagram explore section.
25:45So, at three times a day, at 10 AM, 2 PM
25:48, and 5 PM, he will be looking at the
25:50posts that are trending most in my niche
25:53, okay?
25:56You can't do this with API, you can do
25:58this with scrapers, like Apify or some
26:01other scraping service, but it doesn't
26:04stay specifically on the Instagram
26:06explore tab.
26:08So, what do I want Gram to study in
26:10practice? What makes a post appear in
26:13the explore section?
26:15Because if you appear in the explore
26:17section, you reach new audiences and
26:19achieve his goal, which is to grow my
26:21base there on Instagram and attract more
26:23customers, right?
26:25So this is very cool. I have Gram
26:27specifically for Instagram, and he also
26:29takes care of the scheduling, see? I've
26:31even asked him to schedule the new posts
26:34.
26:35He already scheduled one for Sunday and
26:36two for Monday, one at 9 AM and another
26:38at 6:18 PM.
26:40And see how well it uses my drive, right
26:42, my brain, because inside the brain
26:44there's an organized folder
26:48to search there.
26:50So when I say, "Look, schedule the new
26:52content," it already knows where to go,
26:54checks it, sees if there's anything new
26:56, looks at the publication calendar, and
26:58posts it for me, right?
26:59So, in the same vein as Gran, I have IUT
27:01, which takes care of YouTube.
27:04So, same thing, IUT's computer stays
27:06connected to YouTube and it even does
27:08things I can't do, like an MCP, like
27:11posting to the YouTube community.
27:14We have services like Zeno, which does
27:16automatic publishing, but it doesn't do
27:18it for the community.
27:20It handles automatic video publishing
27:22and automatic shorts publishing, but not
27:25community posts. So, IUT is already
27:27taking care of that for me. And it's
27:30really cool.
27:31This is really cool in practice, because
27:34they can work in parallel, right?
27:37It's very common for me to create a
27:40video like this on YouTube and then ask
27:42the other bots, like Gren, to make
27:45carousel versions of this same content,
27:47but optimized for Instagram, the link
27:50for LinkedIn, the adzin for ads, and so
27:52on.
27:54So, it's a team that can work in
27:56parallel, and that's how I like to work
27:58on my content.
28:00I create a single idea for the week and
28:02then I distribute it across various
28:03social media, obviously adapting that
28:05content to the language of each channel
28:07.
28:09This will obviously help me to have a
28:11powerful 24/7 scale.
28:14And the last two categories I'm going to
28:16introduce to you are the learning
28:18categories.
28:19The first of them are the masters. The
28:22masters are the bots that will basically
28:24take care of an important skill here
28:27within my brain.
28:29And they will be like teachers for the
28:31other bots.
28:33They will always be focused on teaching
28:35skills, on refining the skills of the
28:38sectorized, operational, and strategic
28:40bots. What do I mean? Pen will be a bot
28:44specialized in storytelling.
28:47And then it will train all the bots that
28:50use storytelling in practice, like the
28:52copywriter, the marketing ones, and even
28:55the educational one will need
28:56storytelling to some extent to explain
28:59certain things.
29:01For example, the history of Alma
29:03Educação, our mission, the history of
29:05our products.
29:06All of this involves knowing how to tell
29:08good stories, having good narratives,
29:10brand positioning, and so on.
29:12So Pen will be there to train this whole
29:15group.
29:16There’s nothing in any of them yet, you
29:18know, in Pix, Human, or Money, because
29:20I’m going to do that next week,
29:21otherwise I'll burn through my tokens,
29:23right?
29:24Imagine training 30, 40 bots at the same
29:27time, it consumes a lot of tokens.
29:29So, in this case, I decided to wait, to
29:31focus more on the operational ones,
29:33which will already help me, and the
29:34strategic ones.
29:36And then, over time, I’ll increase the
29:38scale of their training.
29:40But I already know what they’ll have,
29:42because inside my brain, like I told you
29:43, there are over 60,000 files in there.
29:46And I have an entire playbook, I have
29:48notes from storytelling books I’ve read
29:50, summaries, and so on.
29:52So, I’ll use all of that to train Pen so
29:55he can be the expert in storytelling.
29:57Obviously, he’ll be able to do research
30:00every day if he wants, or once a week,
30:03to improve his skills.
30:05Pixi will be the same thing, only he
30:08will be focused on sales. I named him
30:10that because it reminds me of Pix, right
30:12? Pix in the account. The 'C' is for
30:14cinema, 'Human' for human psychology in
30:16general, and 'Money' just for money. Got
30:19it?
30:20And following this same vibe, but now in
30:22a different way, the "geniuses," right?
30:25What are the geniuses?
30:27It’s that tree of references from great
30:29minds that I base myself on to do
30:31something, right?
30:34So I always follow few people, but these
30:36few people are excellent at what they do
30:39.
30:40So human beings always need good
30:42references, right? Sometimes we have a
30:45teacher or a mentor who helps us a lot.
30:48Just the fact of following them, just
30:49the fact of consuming their content
30:51already helps us a lot.
30:53And I hope that perhaps I am one of them
30:55for you, right, a genius, let’s put it
30:57that way, right?
30:58And so, what’s the deal? Here I’ve put
31:00people I always follow.
31:02So, one of them has passed away, which
31:04is Leo da Vinci, but I have his book, it
31:05stays here on my desk.
31:07He helps me be creative, to think
31:10outside the box, to observe more, right
31:13?
31:14To draw, to sketch, to actually believe
31:17in ideas, and to connect things.
31:20So he will be a clone of Leonardo da
31:22Vinci’s mind to help me with this type
31:24of challenge.
31:25Russellzin is Russell Brunson, right? If
31:28you already know him, he’s from the
31:30digital world, you’ve probably already
31:32met him, because he’s a legend. So it
31:34will help me with sales funnels, offer
31:35creation, and generally speaking,
31:37everything related to that. Steve Jobs,
31:39right?
31:41He’s the guy for innovation, brand
31:43strategy, branding—everything we already
31:46know Steve Jobs is fantastic at.
31:49Hormozi, right, he’ll be focused on
31:52offers, pricing, and sales.
31:54He pairs really well with Russell. I can
31:56consult one of them individually using
31:58the specific framework I prefer from
32:00either of them.
32:02So, I’ll prefer Russell for funnels and
32:04Hormozi for offers, value equations, and
32:06, well, every model he teaches.
32:09Or I can make them talk to each other,
32:11right? I can come here and create a chat
32:14with them, right?
32:15"Russelzinho" and "Ormozinho."
32:18Okay? Then I select both and ask them
32:21about some challenge I might have, right
32:24? Like, "Hey, I'm going to position a
32:26new offer, what do you think?" Then each
32:28one will bring their own perspective and
32:30help me with that.
32:32And Matt Gray is for systems in general
32:34, social media scale—honestly, the guy
32:36is insane at that.
32:39What is the big advantage of this? You
32:41remove the cognitive bias of your "
32:43brain" only having your own opinions.
32:46The idea of a brain, of an AI brain,
32:48isn't just for you to have your own
32:50opinions or for it to agree with you on
32:52everything.
32:53That’s not what it’s about.
32:55It’s about expanding your intelligence,
32:57or rather, "Inteligar."
33:00Why do I always say at the end of the
33:02video: "I'm Leonardo and this is
33:03Inteligar."
33:05Because "Inteligar" is different from
33:06just human intelligence. Human
33:08intelligence is focused only on our
33:10repertoire, only on what we have already
33:13lived, tested, and validated. So it's
33:15very finite compared to something
33:17infinite, which is collective
33:18intelligence.
33:20So here you can bring in this new layer
33:23of intelligence, operation, strategy,
33:26and scale in a way, okay?
33:30So I hope this video made sense to you,
33:32and I truly invite you to be part of the
33:35"Empresa em Dobro" workshop.
33:38Because there, we’ll discuss how to
33:41structure a human company. For those of
33:43us who use this type of solution, it’s
33:46not just about creating bots—that's easy
33:48—but what about the architecture of that
33:50team? What needs to be in those files of
33:53soul and expertise?
33:55How to actually build a corporate brain
33:57for these bots to manage.
33:59So there’s a lot of back-end stuff going
34:02on behind the scenes that you haven't
34:04mastered yet. And there's nothing better
34:06than learning from someone who already
34:08knows, who does it in practice, and who
34:10has been teaching companies for 4 years
34:12. I go to companies personally, and not
34:15just any companies, but major ones. So I
34:18truly invite you to be a part of this
34:21workshop, which will be fantastic. I am
34:23certain I will open your mind and
34:25prepare you for this new era we are
34:28entering, which is the human-agentic era
34:30. I am Leonardo Oliveira and this is
34:33Inteligar.

About this transcript

This page contains the full transcript of Crie Um Agente De IA Para CADA ÁREA DA SUA EMPRESA E VIDA by Leonardo Oliveira | Inteligênc[IA], generated from the public captions YouTube serves with the video. The transcript has 5,972 words across 944 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.