# Is Grok Bot the Future of Agentic Work?

Channel: ICOR with Tom & Paco
Video: https://www.youtube.com/watch?v=zOdIV8frGMI
Duration: 1 hr 2 min
Language: English
Words: 10101
Transcript page: https://viewrankai.com/tools/youtube-transcript/zOdIV8frGMI

---

[0:02] Hello everyone. Sorry about the delay. Well, technical issues, we are not over them even with AI in the era of AI. And well, who knows? Maybe they caused all the issues after all. So, restart your computer, switch it off and on at least once a year. At least at least once a year. Yeah. Yeah. Correct. But here we are. Well, and uh now let me

[0:36] open quickly the chat. Okay. All right. Let's let's dive into this. Paco, what the heck have we done again, man? Well, keep enjoying our lives as always. That's the thing. That's the thing. Oh, let's let's dive into this. That is a very interesting episode about AI orchestration, agent orchestration, agentic systems and things like that. And also we will talk about the big picture that we see people keep

[1:21] losing or not headed in the first place. They dive into something, they think this is the only thing they need to know and then something changes in their life and everything crumbles. And that's where we need to talk about the basics and we need to talk about our insights about a few things and so on. So before I keep startling, let's dive into this because you saw already the thumbnail and the title. It is about Grockbot and that Paku and I we both started using Grockbot to see what is capable of and this is where people go nuts because we build our own tools now we talk about Grockbot and you don't

[2:04] even know that this is not the same thing it's different things and that's what we want to talk about and uh yeah I don't know how we want to start Paco well I think I just published a video on YouTube where I told all my story. So if you want to dive deeper into this, there's a 30inut video where I tell you the whole story from April until today. So I give

[2:25] this time to Paco. You know I will. You know I will. Yeah. No, the thing is that um little by little we're moving towards what we envision since all this AI story started that it was that AI work space. Okay. of course cloud-based uh as we always said okay but uh also um the thing I will share my my thoughts okay as I see it because they are a little bit abstract but I will try to explain myself as as well as possible I've been perceiving since AI started okay and the way to interacting with the front end or the software behind uh was a a chat okay that uh since that moment

[3:19] we won't navigate the front end of fatul the exact same way I explain myself I cannot see ourselves in the near future and even uh even today okay and that's why all these changes and all these pivots from one software solution to to other I cannot see ourselves navigating through the classical menus and going in and out and interacting with the front end. Okay, I think that it's things are going and that's what we are trying to emulate again. We are at a transition period and all these are little step towards that destination in my opinion

[4:00] that we interact with the front end. Okay, for those who don't know what the front end is, it's just the screen we are interacting with. Okay, how we send and retrieve information from a software. Okay, that uh stores that information and also those features and uh I think that we start interacting little by little more and more in this way that is I give an input and the output is delivered on the go. Okay, completely on the fly. Okay. And that completely changes the way we were used to interact. Okay. It's the exact same way thinking about how we store information. Okay. We started with the

[4:44] folders and then it cames the tags and then it came AI and everything changed. Okay. Because the way we look for information is not the same. Okay. Initially we had to go folder by folder. Okay. Trying to find the things. No matter that we had a a search very basic search feature. And now in AI everything seems that it's interconnected because that's the reality information was always interconnected. Okay, the point is that the front end was enabled and the infraction to behind to deliver that type of organization. Okay, that's why the moment that software started to create these connections, okay, and all those back links and all that story,

[5:26] everything started to crumble. Okay, and people get lost because they were used to follow a path to subfolders and find the information and also when they store the information when they think that you don't care about what the information is stored, okay, you just say store this information and the system takes care of it. And I think that talking about interacting with the front end, we are in the exact same stage, okay? Where we are not going through menus. It doesn't

[5:55] matter if they're on top or in the side. And you structure how you interact with the content and the feature of a tool that way. And that's what all those AI assistants start to understand. Okay, I think that this is the first stage of that AI work space that is not based on typical terminal sessions. Okay, for the

[6:22] people is like chatting to someone. Okay, but in reality there are a lot of different sessions and things in the back end. Right now they are hiding the the the layers that we had at the beginning of this story. Because if we go backwards and I don't want to replicate what Tom has just explained in the video, but if we go a little bit backwards when we were able to interact with AI and being able to store the things, okay, we started that moment storing things locally. Okay. And the only way to interact with something local was using a terminal local session that allows you to interact with that information little

[7:07] by little. And as we predicted okay information is going to the cloud obviously because we are not going to list all the advantages of that okay and now little by little those approaches of the concept of a session itself of course accessing something locally okay all that is starting to disappear and also at a very fast speed because right now we are in an era that is the fastest talking about software development and

[7:38] things are changing almost week by week. Okay. And that's why it's so important to have a flexible protetic system that you understand based on a solid foundational core to be able to pivot the moment that you consider that you can get advantage of these news features. Yeah. And we see here the confirmation just on YouTube. I'm I'm confused every week new feature where to begin. And the thing is to be clear, it's not us putting you onto the ride. We are in the same boat still and we getting hit by each wave. And it's not even weekly. It is daily when you wake up and there's a

[8:19] new AI release or new technology release and so on. And this was already confusing back in the time with the productivity tools that people just got used to notion or something like that. and then more and more came out that had a bit different uh bells and whistles and and so on. But I want to share the screen here because something that we have brought back to better explain all this inside the membership and now there is no screen showing. What the heck? Oh no, that's the Paco screen now mine. See

[8:53] user error. Um, one second.

[9:06] Now I'm centered too. Good. Mhm. So to better explain to you this, we brought back the tool workflow creator inside the membership. So you guys can use this too if you want to lay out your tool workflows what I'm doing right now here to better explain and get a better understanding what is going on in your technology. We have the ICO framework where you lay out your tool stack where you understand what tools do you use for what. You can even have a ven diagram here and there you see already on all the tools that I'm using here is still my custommade tool for everybody who

[9:41] built your own tools. It doesn't mean we change something now and everything is now crockpot. And this is something the foundation behind everything is the icon methodology that never changes. It still applies and you need to always compare it against these concepts and workflows in your life. How does this new tool change affect your life? And do I need to change anything at all? Just because we are being on this ride, it doesn't mean that you need to follow along. that that's the point. Uh you can look at it, we test the things for you and it doesn't mean that you need to keep switching the moment we do something

[10:20] new. Okay. But um here you see this is my self-built application anic and this replaced a lot of tools in the past which was Hepabase and uh uh essentially was just Hepabase in the end and then read read wise and then it was Obsidian after Hepabase when I had the local folder approach and all this got reflected inside the IO framework but we see here there's Google calendar there's to-d doist there's ClickUp there's Gmail this never changed and once you get this understanding that there's a just a part um a section or part of your system changing you will look at it differently and don't feel that everything changes

[11:02] now and that's why it's important uh to realize this and then we also talked in the past about back end and front end which means that we have the data that lives here also in ClickUp or in to-d doist there's a database in the back end that holds all your data and then on top of it there is this interface that shows you and presents the data in a certain way. So if I'm going into notion it looks different than it looks in clickup but it is very similar when it comes to the data structure in the behind the scenes they just represent it differently it lets

[11:39] you interact differently with the data and so on. So using tools like Grockbot or uh even claw code or any other AI tools that you use chat GPT that lets you connect to other tools. This is what's happening. They are connecting to the back end of these tools. It's just reading reading the databases behind the scenes. It doesn't care about how ClickUp looks like, how to-doist looks like. It cares about the data behind the scenes to gather the things together for you to work with or what you ask for or to up create a calendar event inside your Google calendar or things like that. That's why I wanted to bring back

[12:25] this workflow creator because now it makes easier to explain this in more detail. So we have now Grockbot and what did we have else before? It was cla clo and it is still there. That's the whole point. It's still there. But in the beginning, we only had Claude code. Uh Claude code, man. Claude code. And then we had folders created because there was the pivotal moment where it was suddenly possible to connect your AI to a local folder. And then we gave instructions in this local folder how this AI should behave and rules and skills and all this. And now Grockbot replaced this combination when it comes

[13:15] down to how my AI works. It doesn't replace my knowledge base, my PKM, all this data. It is just how the AI works itself and how it interacts with my data. So this means for you if you've been using folders and you used it I don't know I call for life scaffold uh and also we can bring in here obsidian as the typical example when you want to look into an interface because that's exactly it obsidian is just the interface to your back end which is the folder here okay so you can treat folders as your back And obsidian in this case would be the

[14:02] front end. And now we used claw code. And people say well clawed code it work or claude it works perfectly with obsidian. I can interact with it. I can do whatever I want. No you don't. you interact with the folders and you just you can hook into Obsidian to give it the right rules through Obsidian how the folder which it's called a vault is set up and uh the specifics for this interface to show you this what is inside a folder properly but in the end all you're manipulating is just files in a folder so now if you keep this in mind and I think now about

[14:44] my own tool I have the same system. Okay. I have my own tool. There is the front and back end. And I can bring this in twice. So we have a front end back end. Okay. That's what I built to visualize my information that I have in here. So now what happens? I always were able to hook into the back end and manipulate my anic uh information but I could now sit just in this terminal and that's what Paco just mentioned and keep talking to AI and say hey I have this idea about something or I want to create a new journal entry and so on and

[15:31] all AI does it manipulates or updates or edits the database of any tool that you want. This is not happening at all in clawed code. And even if you have memory switched on, you know, this is not your personal knowledge management. You have to have some sort of tools around this AI ecosystem. So many people start with Grockboard or they start with even earlier right with ChatGpt and they start sharing their whole life in chat GPT and they have several chats open and then they consider that this is a PKM system personal knowledge management system but it's not because even that you have a lot of things shared in there there's no

[16:15] structure to it it's just all over the place and if you ask um your AI it will find it but it looks through all these things and that's where it becomes ine inefficient and eventually you want to access this data in a proper form too maybe and that's not possible if you keep everything inside these tools that's why again you end up with a front end so one last example and then I let Paco add to this let's use notion because it's a common tool and here we have the same principle again whoever used notion they know it is based on databases and you have cross-connected databases but it is also

[16:54] a note takingaking tool you can very easily interact with it and this is exactly what happens there again we have the notion which is the front end that you open up to type and uh add information data into the database and we have the back end of notion which are the tables that you don't see you see it through this interface okay it's always served towards this so now again I can hook cloud code into notion many did

[17:23] this but you hook get into the back end. You will never be in the front end. Even we could argue now, okay, computer use and you use your Chrome ext. That's not what I'm talking about. That's simulating you interacting with the interface. And as you know, it takes forever to do because it's the inefficient way. The efficient way to interact with this tool is go into the back end to talk to it. Okay. Now we've been able to build all this in the past ourselves and say, "Hey Claude, can you actually connect to notion and it will build the connection for you?" earlier times it was much more complex. You had

[18:01] to build the setting files yourself and so on. Now I just ask claude, can you connect to notion or in cloud desktop I even have a connector. I click a button I connect directly to notion and that's what's happening and this happens through the MCP or API but most of the cases MCP so now why Grockbot is shocked me over the weekend to be honest and I used it I started using it last week but then the more I used it was really first of all the ease of connection to other things yes it's easy with claw code but here I was just sitting in the car I said

[18:37] connect to the things and it just connects directly via the MCPS and gives me a page where I need to confirm and that's all it. So now I hope you better understand that there's a layer of this AI and this is now the thing that removes the friction of interacting with my tools. I need to open my tool. I need to go to the place where to find the information. I need to search it. And uh here I can now have this as my capture device and say hey keep this in mind update my uh AI tool create a task in to-d doist uh because I need to work on this um and

[19:18] or or or ask about what are my tasks and all this and this is all the same principles. So now I have a central hub and one single point of contact tag that I can go to to talk to. And in my self-built version or if I'm using AI inside notion or if I build something that works inside Obsidian, those are isolated AIS again that work in this ecosystem that I built. That's why I forced an to be the all-in-one solution, but it will never be. And even if you're

[19:50] building it yourself, it will never be. And it ends usually when you realize, oh, I have a team actually I want to work with. How can I now get my team into my self-built tool? And that's again where we see people misunderstand how it works because I can actually connect to team tools and then I can interact with it this way. Again, I have here back end of the team tool whatever it may be. And once this clicks then you understand what we are talking about and why it's so easy for us to say okay let's use crockbot or now let's see let's use chat GBT

[20:25] connected to all the tools and see if it is more natural or cloud code beware the moment it becomes available in Europe we might you know we at least will test clot duet to see if this works but grockbot well I have to say it did We have here, well, we have here a whole new course where you can build your own agent harness and you can go in here and uh it looks like this. You can build your own agent harness that works like uh Larry and his team because Grockbot is literally just what we built in the beginning of the year just now that there's a huge company behind it

[21:11] and they are much better in making all this work. So why would we now invest the time to build these things for ourselves if 90% of it is exactly there and then rather think about can we actually made other tools around it that support my work in a better way and this guys wasn't there a week ago just in the last week there were so many updates come coming to Grockbot that it drew my attention to test it as well Paco mentioned it already since a month or so that beware they really moving fast and so And yeah, we've been aware of this and now we both dove in and see saw how

[21:48] far we can go and here we are. Yeah, I think that the the problem when AI started was there were two main problems. Okay, the back end where do we store all the information all the insights all the things that we have exchanged with the AI the LLM model and once that was covered okay because first we were able to create files that could store that information and pretty soon that's why SQL light came into town as the main database manager because it's a simple database manager that you can store locally and I mean it it's a file the database itself it's a it's a file

[22:34] in the local folder okay and with that we always had the problems of the front end okay because once that you were able to store everything in that backend okay in that database or in those files how do you see the content of it okay and that's why we started with what we call the cockpit okay that it was an HTML that show the content of that database

[23:04] in ways that were interesting for you. Imagine that you want to see the goals, the projects and the task. Well, you need to retrieve that. Okay? And the point is that if you don't create a front end, it's impossible, okay, to see the the content the way you want. Okay? That's why all the different tools because we have to understand that we are in a transition period. Okay, we have started in the cave going back to the terminal that was a tool that non end user was using at all. It was a tool for developers. Okay, until this moment. Okay, but we are talking that this has happened in just

[23:46] one year. Okay, because it seems that we were using the the terminal, you know, and it was just I would say two or three weeks ago, okay, more or less and I would say even less, okay, that now for instance the cloud desktop app can compete directly with the CLI, okay, the software that you use on terminal to interact with cloud. We are surprised how the club desktop app has improved because now we have everything that we were missing when comparing to

[24:21] use the terminal version of the CLA. Okay. And this happened just they have released okay and they are really stating the people from anthropic that now the cloud desktop app and we can tell you that it goes that in fact I'm super happy okay because I have left if I stop using the the terminal okay and I hope to say forever okay I hope not to go back to the terminal in 2026 and the 21st century okay because it doesn't make any sense so imagine how primitive this technology was uh when we started uh last year. Okay. Right now we can use the cloud desktop app to have an amazing

[25:02] UX where we can interact with an LLM a little bit more naturally than the terminal version. Okay. But the thing is that even that version has get outdated thanks to this AI assistant like for instance groach and um in the case of anthropic do it okay because now everything is changing talking about how you create the infrastructure tour around the the LLM. Okay, because initially you you need you needed an orchestrator and you need to put some guardrails and you need to take care of the context that you were adding to the session. Okay, because if there was a lot everything collapsed, okay, and you were

[25:49] consuming constantly tokens and tokens and tokens due to sending the whole chat in every interaction. Now there has been a lot of improvement talking about how automemory works. Okay. So you are talking and that's something that we experienced and we were shocked how the heck we were able to create an amazing improvement or our branding in our myor portal. Okay just with one session okay without need to resume to cut things to split things and use different sessions with different purposes. The LLM was doing exactly what we expect from software, okay? Doing that tough job, okay, and being able to deliver the outcome we expected in no time without

[26:34] confusion, without hallucination. And now we are even seeing people from Anthropic saying that please remove the clot MD file that it's the the basic and initial setup that any session not so long ago needed to have an overall understanding of your context. Okay. Well, now they are saying that that creates a limitation for the current LLM models. Okay. And it's a something that for instance I am experiences also with growbook. Okay. You start talking in just one chat and he decides decides what he wants to preserve, what he consider that is interesting, what is your contextual information and all this is happening in real time and letting

[27:20] the users totally confused because we are constantly creating new rules that are immediately destroyed just a couple of weeks after because way that this is improving. It's insane. Okay, we have never experienced such a fast speed to deliver new features and to improve the outcome that we expect out of an income that we have put into software. Okay. And I think that that's

[27:47] where we are and we need to be aware. Okay. Because things are constantly changing and what we need to have is a solid foundational core that we can trust on so that we can pivot at any time here. not Tom or myself. We have members that have pivoted perfectly between the cockpit then they move to the uh obsidian approach then they move to building their own app and they will keep moving without any problem because they have understood the core principles and workflows that you need to understand that that's why we created the echo methodology because people can say hey guys you are constantly changing of tools by the way something that they

[28:30] say before AI okay because we were constantly changing of tools 2017 opportunities. Okay. And this is the thing that right now we are constantly seeing opportunities because one we always said that this local approach was something uh ephemeral I would say okay and just something trans that was a transition because it's impossible okay nowadays to trust we know all the tradeoff of local in my case I have several I work in several places my home and my office like many people okay where's the mobile that was something then I need to I needed to create a server on my iMac by the way my iMac all days on waiting for

[29:16] me hoping that okay the electricity doesn't fail because if not I don't have access on mobile a lot of things okay well I won't tell about mobile devices like iPhone or iPad where you cannot see the outcome properly okay because you need to redesign and then we learned okay and show our people how they can create their own iOS app all those things are part of a transition period that by the way at this speed I don't know when this will stop because I think that people should get used to pivot the moment that some specific features are for instance better developed or delivered by these tools for instance

[29:58] Tom said well we keep using cloud code yes it's my case and his case because what we have seen that we are using robbot as the front end as a different way of showing and retrieve the information and interact with our productivity system but the delivery okay the development of our own productivity system in this case at least is my experience in this weekend I'm consuming tokens and I'm killing a 200 okay tier that it's the highest okay then after that I need to start paying token by token okay and that's something that I'm not going to do so I'm delegating all those development because still it's an early stage when I ask for

[30:44] instance I want to deliver my weekly planning okay in my local tool I have no problem I have a a screen that has been perfectly designed and customized by the way the the most beautiful okay planner I have had in my life okay because I have designed it by myself but it's custommade the problem is when I ask

[31:06] that to robot he has nothing to show me. I need to create a specific okay outcome that represents in real time that weekly planning. The beautiful thing is as I have the the screen okay the design in my local tool I can perfectly share and tell well when I ask you for the weekly planning please deliver this page because it can deliver pages pages that you can interact like we could do using clot okay with the artifacts okay they can deliver outcomes in real time yeah but also inside the tools that you have been using that's the point you know that you can say create me a notion

[31:47] page with this or create in my own tool create me the page directly there and this is also something people think I need to now create a separate file that ends up again in a folder and then I need to organize it no if you have already organized the system there's no need for that yes and those are the advantages of betting on these powerful tools because for instance the moment that you bet on robbook or duet or any any other tool dots from chat gbt okay The thing is that we cannot it it's not easy to be in just one tool. So imagine checking absolutely everything but well it's the

[32:23] same approach because what these guys are delivering is the possibility of deliver the front end coming out of AI and all the back end that you have created. Okay, by the way, that we were saying I'm using torso as a solution uh from a database in the cloud and we were thinking about trying to move our databases to software that is much easier. We could say to our users for instance air table, well I make a quick report, okay, and it's impossible for me. Okay, I have 428 tables, 125 triggers, okay, and 300,45 uh 345,000 rows. So that's something that air table cannot handle. Okay, that's why it's

[33:12] important that you get the point based on all the different approaches. Okay, that's why first off, you don't need to feel overwhelmed. I need to change immediately. No the people there are people who already stay with the copic and they are working perfectly and we are talking about something that happened almost I wouldn't say one year ago but eight years ago okay and they work and there's also what did we calculate.5% of people building their own thing 2% of the people and maybe it changed but a bit but 2% of the US Americans are you paying for a subscription in AI so there are people who are still working and

[33:52] productive without touching AI at all. And I know my wife, she will never use anything like Rockbot and she will perfectly organize everything that she needs and it will actually add anxiety and friction to her life. If I, for example, would force her now to use something just because it works for me. By the way, again, what you would experience with your own team. So if something works for you, it doesn't mean it works also for your team. If they're already working, great. No need to change anything. And this again applies to a system as well. And what you just mentioned there with the tokens, I just

[34:26] want to say because I just just right now on the other video, there was a comment popping up exactly mentioning this. I burned through the ultra plan in no time. I'm at 70% and still 5 days left. I have no idea what we will do. But we both both came to the same conclusion after talking this morning to each other that we need to hand off the complex work to other tools. Uh Elon Musk stated that there is OPUS 5.5 because there's it's running on their servers and there's also Midjourney and Zoom and all this but there's no transparency whatsoever what models are used and I used Grock just to give it a

[35:03] go to code tools for me to make the whole video editing process and the thumbnail process and all these things to see how far can I go and yes it works and it seems it uses the right models and the output was great But man, this was burning money where claw code is so much more efficient. And that's where we came for now to the conclusion that Grockbot is great for operations. I would use this on a daily basis to quickly check what support tickets are there, what do I need to do and uh creating journal entries, all these things. But when it comes to coding, I say maybe we plan it

[35:41] out and then I hand it over to Claude code again. And by the way, if you saw anybody telling you that you can just log into your cloud subscription uh because these Grockbot have now their own terminals and this and use your don't do that. You will get banned by AI. You will have a permanent ban from Entropic that you're no longer allowed to use Claude if you do this with your subscription. And that's the stupid thing that annoys me so much because we have now something like Grockbot, Claude, uh, J Chat GBT and they all hook perfectly in all my tools that I can interact with. But why the heck can we

[36:21] not connect these AI tools with each other? I would use Grobot maybe for communication and then it hands over the work to my cloud subscription and then maybe I use Chat GPT for generating images. I don't know. But why the heck is this now the problem? And that's the limitation and that's what we again go back to a manual workflow where I have now decide okay this is coding work this is a specific workflow I will hand this over manually to claw code because if I do it automatically I risk to get banned nonsense and not transparent at all you know it's really a gray zone everybody

[36:59] is working in here right now but this being said be careful about the tokens if you know $200 for ultra I have $200 on the max plan. That's not what the typical us are looking for. You might have a $20 pro plan and that's already a lot of investment for you per month. Who knows? And this is where I understand that not everybody can use all the tools. But having something that works more generalistic uh growth works perfectly. And for most people just want to check their emails and want to do some certain things. this would last a lot longer for for of course that's why

[37:36] we pushed it so hard to figure out where the yeah but the thing is that it's important that no matter how simple your life is or you consider that it is because normally it isn't okay and you start using just grow bot on the go without any mental framework mental model or methodology behind and it will be chaotic okay one of the things of working properly is the moment with grubbook or any of these AI assistant is having a structure that you understand and also that lives in tools that are prepared for that. Okay. Tools prepared to manage information. Tools prepared for manage task management. Tools

[38:14] prepared for a specific use cases like for instance email management or whatever because if not it will become chaotic the exact same way that people uh went mad when they were start using uh PKM tools in the past or task management tool or whatever and they were merging information and and action. I think that

[38:35] No, go ahead. No, you finished. All right. Yeah. Yeah. No, the the only thing I want to add here for the end of the session um two two comments that are in the membership right now and again I shared the wrong one here. So one is I have been sticking to Hermas because it's all local and all my business info IP context stays local but you're suggesting that I can use crockbot to reach my local files so that it works with local files but works blah blah blah then you know it's about I can move on to the next tool and then I not losing my business or context and that's

[39:16] the point here uh two things that's what we do all the time and that's it was never easier to move on. Um, I pointed Grogbot to my AI team folder and I said, "Get the agents out there, create yourself." Because under the hood, there is no magic. Grockbot still works on skill files and it still uses these instructions. And this brings me to the next ch comment here. Wouldn't it be the same making Holista which is the custom AI that we build inside our membership fully accessible with communication channel like telegram for example no other player okay that was exactly the feeling I had and that's why

[39:59] we published this new course where you can build this completely customized running on your cloud subscription plan no Grock plot in involved you can build it for yourself if you're up for it but not everybody has the even the patience because I'm I'm getting you know annoyed too Paco as well. We don't want to maintain all the time our tools and update the things and the more features you add to it the more complex it becomes. And that was for Grockot the eye opener when I was driving the car talking via voice chat it perfectly integrated with all the tools. It pulled the data in and in my head it was going

[40:37] in. Yeah. But I could build this and I could be and how did they do this? Yeah because I know. But the thing is I came to the conclusion but I don't want and can I actually live now with this latest iteration of AI solution that we have here which is just the beginning and I can so far live with Grockbot without the need you know and then we have this where we hand over to claude code and so on because if you say okay but I want to have telegram and all this build it for yourself you can do it if you have a crystal clear plan do it But

[41:13] I built a voice chat for a whole year and I made it responsive. I made it that you know there's a lot that goes there. People think okay it's just a voice that goes on top. No, no, no. It's completely different APIs. You need to connect with different platforms and just talking about it. I know I lose 80% of the viewership here just talking about the different API connections and what platforms you use. And we also realized talking about tours and superb base those people who followed us they know what we are talking about but most people and that's the point it will not even be necessary anymore to understand

[41:48] all this in the coming future because these simplified things like Grockbot will take it away and now I want to point out here something like um uh that involved the privacy okay Grockbot I thought it's a black box Initially I thought there's nothing I can get out of it to understand what is going on and so on. No, they have their own computer and all the files are available inside this computer. All instructions that you give to it. Everything all the things that it created and you can just tell it to download it and uh and I can download everything whenever I want. I can also directly

[42:30] access my data, the instructions, the skills and all these things on Grockot. And let's point it out. The Grockbot is the only thing that we can use in Europe right now. There's a reason why dots and muse and all the things are not allowed in in Europe. And I I don't know about the privacy thing and how much I can access, but Grockbot is just a virtual machine in the web in the cloud that I have this local folder at for my AI team. And now if you think about it, it's the same thing what I did here with my Mac Mini where I had made now on my

[43:06] Mac mini the local folder and there was my AI team running on and I made it accessible from the outside. It is the same principles but now Elon Musk and his team is managing the cloud service for me and I don't need to take care about this and not everybody wants to build a second machine just to run a server locally and all these things that we've been fighting against in the past where it became nerdy and they have their n system to at home to back up everything and all these things or own servers. Yeah, you have control over it, but you also have to have understanding

[43:42] how secure is it? How do you ensure the security is there? All these things. So here that's where for me it was the final thing and I pointed this out in the video too. It the transparency is there. I can really see what is going on and what is saved and where it is saved and so on. And that's uh I think an important point to yeah to point out

[44:02] here. And the thing is that we people need to be aware that we need to test absolutely everything as much as we can so that we can deliver as much information as possible to our members and followers. And that's why we take things always to the stream to see if it resist or not. an inner strategy. Okay, Tom went all in into the world of developing my own app and seeing if I'm able to absolutely kill all my tool stack and deliver whereas I was in another path. Okay, that it was I will try to keep my existing tools as much as possible and just create a front end as simple as

[44:41] possible just to retrieve the information out of those tools. Okay. And by doing that simplistic approach and also because well I was paying attention from robbot. But before that I was paying a lot of attention about Grock when it was just you know a voice you can talk to okay a chat okay but I was amazed by how natural was the voice experience okay and the moment that they launched robog for me resonated a lot because it was a super easy because what Tom has said h implementing on your own voice natural that you can interrupt well many things Okay. Many things in I made the implementation just 3 weeks ago

[45:27] and I think that I used I think it was five or six different softwares. Okay. Because we need to understand all the different steps from text to speech, speech to text, how it manages all the information, the text while you are talking. Well, a lot of things, the different voice tones, the different well specialized LMS that are voice trained and can deliver different things, the cost of those products, okay, because they are not for free, okay? and uh well all that compared to how easy this was. Okay. And the point is that we need to try out those type of things because that's how we can deliver

[46:08] absolutely all the different options for our different members because we have all kinds of profile people who love getting into the nerdiness of technology. Okay. and creating something that is fully customized and for them it's something amazing because they allow you to grow. But we also need to pay attention to all those people who based on their criteria, their time, whatever their profile, they want to do

[46:36] things and become much more productive. And with these approaches, okay, you can be productive and much more productive I would say than in the past. Okay. And we need to deliver absolutely everything. And then it's the person who based on the things and the options that we deliver because we will never say because we have never say that use this tool and then this other tool and this other tool. No, we never say that we are always tool agnostic. We are based on concepts and workflows and that's why we defend the icon methodology because it's showing how we can perfectly implement it no matter the tool. I don't care if

[47:12] today's is bro uh grog box and tomorrow will be another one. I don't care. it will work because we haven't changed the methodology and we haven't changed things like the back end and where we store the things all those type of things are the ones that keep assisting and we are just playing with the last layer on top that is just the implementation and in this case the front end okay the front end because since I started with the local folder I never trusted on files okay and I bet on a local uh database because I want and that in last November. Uh since then I

[47:51] haven't changed torso. Okay. I I started with a local database and then as I was aware that I needed something in the cloud, okay, to remove friction and have access no matter the device to my data. I've been using torso all time long and I will keep using a standard database and I don't care the front end that I use to interact with it. It could be initially just it was just a a web page a static one then I brought dynamism then I create my own app and then right now I can bet on rock or any other tool that emerges out of the blue because I will point that tool

[48:28] to that database and I will retrieve my my data and that's how we can give to you as many ways as possible so that no matter the level of knowledge that you have the time that you want to invest into this you become more productive because you dedicate to the productivity system the time that you think that it deserves. Okay? Because if there's a moment that you're dedicating and that's something that we detected in our coaching sessions, the private ones that we run every single Tuesday. There were people who were saying, "Man, I'm dedicating much more time to my

[49:03] productivity system than my own work." That it's the first signal that this is not working properly for that person. Okay. the exact same way that we saw in the past. people dedicating time and time to set up notion setups or Tana I don't know how many people dedicated hours and hours to set up amazing Tana setups but at the end they were concluding that there were implementing things that they could be done much better with existing tools and that's what we need to deliver and show to the people those tools so that they dedicate time to use those tool to get things done okay or to better manage the

[49:43] information but not to dedicate time to the setup or maintaining that uh because now people have become the moment that we have taught the people how they can deliver and create their own app they are being aware and I think that they will face software in a complete different way how difficult it is and now I think that uh m of those people will start not complaining when they see that I don't know to do is has a failure or a book or a glitch they will say well man I cannot imagine what these guys you know companies with hundred of people you know having to deliver the amazing

[50:24] solutions that they are delivering. Yeah and just to close the loop here um I want to point out that nothing obviously is wasted or changed. We just reached what we wanted to get to to deliver absolutely everything to you inside my I core and we accomplished this point with the launch of the course today that you can build your own agent harness the way that you wanted to and I want to make here uh this comment here from Michael please don't change prior

[50:57] workflow and instructions for the app. So what he's referring to I think is the implementation journey where you can build already your own app using AI. You can hook in your AI into our course and let it build through the course for you. You learn all the things that you need to know about building your own app and make it work and make it reliable. It is related to the input control output refine part. But one more thing that happened now is inside the advanced section and in here in refine we added the AI assistant here just a simple one that that's where holista came up that's

[51:39] just a simple assistant helping interacting with your back end and uh then down here there was then advanced courses building the myka the orchestration layer the work streams and all this And this was based on what we built in the past and outdated in some regards. The concepts were still the same as I just explained in the Crockpot video. They are one to one the same thing now in Crockpot what we built already on on this level. But it made no sense to have the AI layer based on what

[52:17] I showed you now in this overview here. This is the AI layer right inside build your own tool because people think they need to have it all and all in one tool. Now you understand the difference between the back end and front end. You can build your own back end and your own front end. You can decide how you want to interact with it and then you can decide do I want to even go down the rabbit hole and build now my own agent harness which would be Grockbot as an

[52:47] agent harness for example. And you can we provide it all. You have all the lessons. You have all the the resources. You will hear the interactive final version that you will get out of it. Even access through your phone all this and you can do that. But there's also the new Grockbot course and we will provide more tool courses too to also catch these people up who don't care about building their own thing. And that's where this is the uh nerdy zone,

[53:18] let's say. And that's the max level. That's where you want to really have highly customized things. That's where we recommend if you have experience using tools and you always said, man, I'm so annoyed that they are doing it this way and so on. That's why we ended up building all this and we had the capabilities of doing so and it will just get easier in the future. We I just shared in a community about something the um self-building UI on the fly. The moment you start talking, it creates the

[53:46] UI interface and so on. I don't know. See, I love that there's so much engagement that I have to keep scrolling to get my to my post. Perfect. Here it is. Okay. But, um, this is this is just important to point out. And what happened here for anybody who was building in in this main course and there was this section down here and you were in this uh, nothing changed. It moved into this new course and it got updated. And if you connect your AI with it, your AI will know about the changes and it will pick up from there and it will just be the more efficient version

[54:23] that you have here. But this is it guys. These are the two. It's not one course. It's not two courses. It's a series of courses that helps you to build your own app and all the additional features like voice agent your even your own email app. And we are so amazed by all the people sharing their tools and making videos about it. I really love the the style of the videos and all that comes up here where you guys build your own apps and please keep using it. I still have my anic uh app. It doesn't mean that anything shifts now from one end to the other. You not need just to know

[55:00] understand where at the level you are. And this is peak guys. If you build your own tool and it makes work and you are productive with it, more productive than before, then you made it. Then you made your custom thing and if you're happy with it, keep tinkering around with it. That can become your own thing. But others we can pick up earlier now too with the out ofthe-box tools and this is where we keep iterating. We are still not affiliated with any company because uh if next month it is claw do it be prepared that we might have another session talking about oh my god claw do

[55:37] it is the next big thing we don't know but the thing is and that's why I'm so glad that Paco is a co-founder here of my I core as a multi- business owner with four businesses a brickandmortar business with 70 people managing and he does all this thing too okay the shift the decision making but you see if You have the fundamental system below. You are not affected so much. The business keeps running. You're still productive and you see immediately the impact of any changes that you make. Does it actually make you more efficient? Does it make it remove friction or does it actually add more friction? And that's

[56:13] for example the maintenance of the tools or build keep building more and more features like this fluent hand over into your phone and talking to it and then handing it over to another or maintenance of uh VPS and VM virtual machines and all these things that most people don't want to even care about. this is now going away that and I think we both think and that's what we concluded today that Grogbot is really the first one that shows that we are entering now finally this phase where we have something out of the box that you can just talk with but also on a professional level that is that you can

[56:51] use on a professional level too and uh yeah because nobody would tell nobody trusts me as now full-time YouTuber here that I have any idea about business also. Uh those who know my history know that there's a lot more to it. But yeah, here we go. Okay. So, I will always point to Paco if he can do it. Anybody else in our community can do it and you

[57:16] don't even need to. Correct. And I have shared an article that I created the the other day uh explaining okay how uh you can have a a a blueprint to tool independence okay it's an article that I think that it takes 45 minutes to read 23 sections and I put my heart on it so that you get is the last one Tom and you can get an overall understanding of how you should approach a proity system to ensure okay that you become really tool independent okay and I think that the people who get these points are the ones who will never complain about changing tools because the core the essence is

[58:07] always the same and I think that that's important especially nowadays more than ever because the pace of software solutions will be insane insane as we are seeing okay not only about new tools emerging but the improvement of the existing tools okay and by the way I cannot wait after the amazing in my opinion it was criticized obviously because everything can be criticized in this life okay but many people criticized for me the amazing job of the new cloud desktop UX that for me it's amazing what they have done and how they have simplified to the minimum something that is really complex but really complex I I couldn't imagine such a

[58:53] beautiful implementation with such so many features okay on it and I'm looking forward to do it okay to see what it delivers because uh for me anthropic is the the company that was always focused on professionals and that's why it's the one that I we have used the most okay but let's pay attention also to other options because they can be interesting tool and also because they can be complimentary. People tend to just bet on you know a aggressive streams instead of just seeing if you can grab from here. Again we have done the exact same approach that we have done so many times. Take the best features out of

[59:39] this tool. If I consider that for my chats, my interacting, my operational issues, Grubbook works fantastically well until I will see uh how do it does it. Okay. But I know that I can perfectly code and code and code and develop things using code code without hitting the limits because by the way I was surprised because there was a week that I only achieved 40% of my weekly limits. And of course talking about the five hours limit of clo code. I never achieve those limits. Well, that's a an amazing tool being able to develop whatever the crazy idea comes to my mind at that price ensuring that I never hit

[1:00:24] the limits. For me, that was a a sensation and a feeling that it completely disappeared when I started using robots and rem reminded me the old fable time. Okay. when you just created a task and it started consuming and you were seeing the percentage of the week completely uh going down uh after every minute. Okay. So that's why again if you know the architecture if you understand what a tool delivers to you if you understand the specific features and you say well this is the gap that my system have and this is the feature that I will use that's it that's the always as you see the concept behind and the approach

[1:01:08] to tools it's exactly the same because at the end we are talking about software products correct amen that's it for today everybody and we catch you up tomorrow in a coaching session. Most of you see you there. Bye-bye. Take care. Thank you so much.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt