Big Boss AI: Shift from Assistant to Chief of Staff Mindset

Noel said, "It's not about junior or senior, it's about good with AI or not good with AI." But, what does it mean

whether you're good with AI or not? How do you judge that? I think there are two things we have to look at and shift our

thinking around. Number one is fundamentally now what can you do with AI? And number two is how are you going

to do that thing with AI? The second you're using AI, you can now promote yourself to be the boss. And I'd argue

that you want to promote yourself to the big boss with a B. What I mean by that is I see lots of people online who are

saying, "Oh, you know, you can use AI agents as your assistant now." Or, "Hey, when you use LLM, it's like a smart

intern." And this way is, yes, thinking about I'm no longer the last person in the rank. I have an intern that I can

use. I have an assistant I can use. So, yes, you've promoted yourself. But, think about the difference of what you

can do. If instead of building an assistant, you decide I'm going to build a chief of staff for me. Yes, an

assistant helps you protect your time, your attention, right? This is 1,000% the limitation of knowledge workers is

that 1,000 people and 1,000 things are pulling us in different directions, and we need to be able to manage that energy

and manage that time, manage manage our calendars and our to-do's to get those things done. Yes, an executive assistant

can, to a certain extent, understand what should be prioritized, what are the things that are important, and kind of

fit pieces together. So, we do get leverage from an assistant. But, if you bring your big boss energy, instead of

assistant, you're thinking about my AI agent is going to be my chief of staff. Then you'll realize, okay, this agent is

not just helping me manage my time, it's not just helping me protect my energy. They are my thought partner. They are

going to turn the visions that I have into something that's executable. The chief of staff also manages a team of

people, team whether it's real or whether it's other AI agents who can then do more things. So, your leverage

just jumped from yes, the assistant can help me manage my calendar to yes, my chief of staff can then manage more

people to get more meaningful things done. And the type of conversation you're going to have with this chief of

staff is going to be fundamentally different than the ones you're going to have with your assistant. With the chief

of staff, you're mapping out priorities, mapping out opportunities, possibilities of which direction you should go, all of

the things that you can explore. And the type of work, the type of thinking is going to be higher order. Right, so

we're moving up the Bloom's taxonomy. A lot of people ask me, "Oh, so what do you think about

AI making you more stupid?" Like, yes, if you decide to use it at a lower-order thinking capacity. If you tell the AI,

"Hey, you are going to help me think through how to organize my calendar." Yes, you're That is the type of work

that you're going to be doing. And if you say, "Okay, yeah, whatever you say is right. I'm just going to use it."

Then, yes, you're not actually elevating your capacity of what you can think about. But, the AI is also a mirror. If

you say, "I want to do higher-order thinking, so I need a thought partner that has that kind of capacity." Then,

you're not going to become more stupid. You are going to think about, "Okay, for my chief of staff, they're going to need

all of this context. Plus, they're really smart and they have access to all of these other research that I can never

read through with my entire life on Earth." Then, you're naturally going to get smarter and think about problems in

a deeper way. Use that leverage in a different way. So, I'd argue bring that big boss energy into what you can do.

You can just say, "From today on, I'm the boss with a big B. I'm Bowser in Mario rather than one of the mini

bosses. Having that flexibility in who you see you can become is such an empowering part of this technology. And

I'm not saying that, you know, just by telling AI, "Hey, now you're my chief of staff." it can do all of these chief of

staff things and help make everything happen. No. But how you think about what you're going to provide it and how you

think about what kind of ideas you're allowed to work on fundamentally changes. That's not a

small feat. You are being empowered by the technology. And it doesn't limit you from having an assistant and a chief of

staff, right? Yes, AI is not quite there yet in order to perform all of the things that a human like experienced

chief of staff can do. Fine. But we also have to think a little bit further into the distance of the day when it is able

to you've already shifted your thinking to match that capability. And there is no one saying you cannot have an

assistant that is perfectly functional through a agent as well as a chief of staff. It's just that we have to change

what we think we can do. And that is the main thing. So, this is why seniority doesn't matter anymore, right?

Previously we can have people who are managers or, you know, execs and we look at them and we think, "Oh, I don't think

you actually can do all of these amazing things. I don't think your thinking is really that clear." So, yes, the role

that, you know, people are given doesn't matter anymore. If you have the brains, if you have the experience of that exec

that you think, you know, is not performing, well, now with AI you can just give yourself more capacity so you

can function and do the work that that person does. And in the workplace, the number one prerequisite for promoting

someone is they're already doing the work of that role. So, you can literally just go do that. But, this also means

that if you are a leader, you are someone who already manage teams, you're someone who knows how to delegate, how

to think at that more strategic level, higher order thinking. You know what are the things that you want to be working

on. Well, now you can just maximize that and scale that ability to not just the physical team that you have access to,

but also a team that doesn't stop 24/7. And that again exponentially expands what it is that you can actually do. So,

if you're searching for that meritocracy of, "Hey, I have ideas, I think clearly, I can get stuff done." Then, there is a

clear progression forward and you have a lot more resources than you had just a year ago.

All right. Now that you're bringing the big boss energy, how are you going to do the work? How is it going to change from

what it is that you do before you got promoted to boss? There are three things and for sake of simplicity, I think you

should spend 33% of your time on each. The first one is learning because uncertainty now is the only thing that

we know is certain for the next 10 years. Every single second, there seems to be news about something that's

fundamentally changed again. Oh my gosh, now we can do this. And we don't want to fall into this frantic AI guilt and AI

overwhelm and just chase and to keep up because that only creates more distraction for ourselves. Instead, I

suggest you learn to change your behavior. Uh so, what I mean by that is when I say learning, most people will

think, "Yeah, you know, I already learn. I consume a lot of information." But, that is only the first part of the

process, right? Getting input to understand something new, okay, that is part one. Learning is a process and at

the end what you want to do is change your behavior because if your behavior stays the same, then there was no point

in learning. And because we're thinking about that behavior, we're thinking about internally what is that change I

want to make. So, it's not about chasing the news of, "Oh, AI can now do this. AI can now do that. What do I want to make

sure I keep up on?" It's rather, "What is the behavior that I want to change now to get a different result? And let

me focus on that." Having that internal clarity then give you a better map of how you're going to learn. Okay. So,

basically, remember you're looking for that behavior change. Input is consu- Assuming everyone knows how to consume,

all right, but you have to take it further. It's a process, right? Then you map it to your current situation. Then

you find what is that behavior I'm going to change. You're going to make that change at least once, and you're going

to let feedback to tell you, "Oh, now that I made that change, is that the one I want? Is that the one that I don't

want?" And use that feedback to continuously learn and be your input. So, now you have a learning cycle, and

you're spiraling upwards. It's not, "Okay, I watched Vicky's video on why you have to take notes during AI." It

is, "I watched that video, okay, then I said, I'm going to start taking at least three notes today." You took the three

notes, and then you looked at them and said, "Okay, did this help me? Did I take the right notes? How do I want to

move on from this tomorrow? Do I want to take three more notes, or do I want to focus on something different?" You are

actively engaged in this process of learning. Take that feedback, make that into your input, then maybe it is you go

back to the video and say, "Okay, I actually don't know, you know, what kind of notes I should be taking." You look

at the video again, now something else pops out. Okay, I'm going to try it again, right? I change that behavior,

and I watch the feedback. And that way, you actually get something out of the learning. If you have not made a change,

whether it's a behavior, whether you you made a decision about something that you were undecided on, then you have not

learned anything. You have to change some sort of pattern in your brain in order for that to actually sink in and

be helpful to you. The whole process, the whole loop of learning has to happen over time.

And this brings us to the second thing you want to shift how you think about it, which is create. And again, I don't

mean the single act of creating something new or something original. Rather, again, it's a process from

identifying a problem, understanding that problem, coming up with a solution, create that solution, and then get

feedback of does it actually solve the problem? Is this the problem I actually want to be solving for? And you

continuously spiral upward again. Creativity is a proxy for intelligence, right? This is why everyone's going

crazy when all of the LLMs can generate things. They can generate you images. They can generate you one-pager

documents. They can do PowerPoints. They can do all of these things. That's why we think it's intelligent. And so, at a

fundamental level, you need to be able to create outputs. But the difference is, you don't want to be creating in the

pre-AI age, right? Pre-big boss age of someone told me to do this, and this is why I'm creating this output. No.

Creation is this whole thought process. Starting from identifying a problem, you are the

one who should own that problem. It's not that, oh, you know, the company's strategy team came out with this

problem, and now I'm just a cog in the system trying to fix it. No, you are the one looking for that problem, okay?

Then [snorts] you you shape it, you understand it, you start to create the solution for it. Let's bring it back to

what we did in the first um thing, which was learning. As you were learning, you know, how to use Obsidian to take notes

in the age of AI. And you realize, okay, wait a second. I find it really difficult because there are notes that

are about ideas and there are notes about practical to-do's. Perfect, right? Now you have a problem that you've

noticed and then you start to understand this. Okay, actually I realize that as I'm looking at a very practical to-do

versus my ideas, I notice that at the core of it, I'm thinking in very different ways. When it's about an idea,

I need to focus, I need to diverge, I need to think about all of the related ideas to go deeper, but when it's

something more practical like a to-do, I'm very much converging onto what is the next step and I'm pushing it forward

and I need to use other tools like, you know, the internet or some LLM and I'm moving all over the place. So now we

understand the problem, there's a different thought process happening between the two. Okay, so the solution

I'm going to do is to start two vaults. Yeah, one vault is about ideas, another vault is about the practical to-do's and

then I'm going to try it out. So then you go on your Obsidian and you create a new vault for all of the things that are

practical and then you test it, right? You say, "Okay, now I'm going to take notes for the next 3 days." And then you

might say, "Yay, it worked. Awesome." Right? You've created something that is your own. Or it's, "Oh gosh, it's not

working. I still need to refer to some of the stuff in the other vault." All right, right? So then you identify the

next problem of there still needs to be some integration. Then you go through the loop again. And by the end, you have

something that is uniquely yours, right? From my video of, "Hey, you should take notes

when you're using AI." to now, "Hey, here's a specific thing that I can't apply the knowledge to,

learn from, and then create something that's my own." And now, you know, if you want to, you create your own video

telling people, "Hey, if you also have this problem of not being able to separate idea space and the practical

space, here's what I tried and it worked. You are now reclaiming your thought process for yourself,

right? You It's not about Vicki's going to tell me what to do. Yes, she got you started, but because your situation is

uniquely yours and you want to achieve things that are uniquely yours, you have now We have now merged our

ideas to create something that works for you, right? You are in control of that higher-order thinking process, which is

awesome. All right, the third thing that you want to focus on based on these two steps is

now systematizing. Remember, you're the big boss. So, how do you get other people to achieve outcomes that you want

at a standard that you like is to systematize workflows. So, they don't have to start

from scratch. So, they can leverage your experience and what you like and what you found over the years and be able to

do that instead of always waiting for you to do. So, again, a higher-order cognitive ability is to take all the

knowledge in your brain and make it explicit. Take that tacit knowledge, make it explicit, and create workflows

for others that can follow it. And good news is, of course, AI is awesome at doing that. If you you already have

experience running teams, creating standard operating procedures, SOPs, or any other

workflows that you create into a little tool, a little SaaS, then you know what I'm talking about. This is how you get

scale. And this is why you see in every single agent harness, whatever it is, Cloud Code, you know, Codex, Manis, they

all have something called skills, a way of doing something for the agent to understand so that whenever it does that

thing, it goes and checks, oh, this is how I'm supposed to do this thing as opposed to make it up, right? because

you can leverage what you already know with your agent. This requires you to do two things. One is take something vague

and break it down into steps. If it's taking notes, it might be, "Oh, you have to have some metadata." And then you

want to write the core idea. Then you want to tag it and you want to link it to other notes. And so, whether it's you

or whether it's a human teammate or whether it's a AI agent can follow how you think through this process. Now, the

second thing you want to do, and most people stop at the first thing, so focus on the second thing, and is bringing out

emphasis and identify leverage points. What I mean by that is the problem with SOPs is, you know, if you got on boarded

and people just gave you a document with, you know, steps 1 to 100. And you look at it, you're just like, "Okay, I

I'm not really getting any of this. Sure, I may I'll try it and see what happens." But when you actually use a

process, what happens is the person explaining it to you usually tells you, "Oh, and this step is really because now

if you don't do this, it'll mess up something later." Or it's, you know, 80/20. In this process, the most

important thing is the research that goes in because garbage in, garbage out. If this process this part is not done

correctly, then everything else is not going to give you the right answers. So, you need to be able to clearly

articulate where is the emphasis, where are the leverage points that matter more than all of the other documents, right?

Which words have more idea density to them? Let me give you an example, right? One of the things that people always

make is some sort of document, right? And you say, "Okay, to make this document, you follow steps 1 to 10." And

then, where do you put the emphasis? One great one is quality. What are the things that you want to check in order

to ensure quality? So, you call out, "Always double-check numbers. Always double-check dates. Always double-check

the fonts that we use, whether they are consistent or not. Just by placing emphasis for this SOP of just, you know,

10 steps. Now, both you and your human teammate and your AI teammates can all understand, "Oh, I need to double-check

the quality parts of this." Because people usually mess up with numbers. People usually mess up with dates.

People usually mess up the font and the font sizes. Same idea, we can also apply something like 80/20 to the document,

right? Okay, steps 1 to 10, but focus on step number one and getting the right data and the research in. And spend most

of your time there because if we don't have good data, if we don't have good research, then the results are just

going to be a garbage in garbage out. And now the third thing you want to be able to do, while everyone stayed on the

first thing, is you want to see how you can connect this workflow and improve it over time. And you might have noticed

this pattern, right? Before big boss you, it was very much, "Okay, do this thing and then you say, 'Okay, I've done

it.'" It's a completion logic that most of us have of just, "Okay, here's a task, let me finish it." But, as the big

boss, you understand it's not about just finishing things. We want to be better at doing something because you

understand that there's multiple ways of doing something and getting to a result. And AI is really great here, right?

That, you know, most of the agents not only can write their own skills based on just going through

once doing a workflow with you, but as you're saying, "No, change this, change that." You can prompt it to say, "Hey,

now update that skill to make sure that we capture all of these nuances that we didn't capture before." This is how you

make your work compound. Thinking in loops is really important. And coincidentally, and probably not so

coincidentally, the difference between an LLM that works with chat versus the LLM that becomes an agent is that the

agent takes time to go through a loop. And we want to go from a completion logic, which takes, "Oh, you want A, let

me give you A." to, "Oh, okay, you want A, let me think through. What is the plan for getting you to A? What are all

of the things I need to take into consideration? What are all the tools and skills I have? Let me create it, and

then let me double-check. Does that actually get you to A?" And you know, this is the same with any human

teammates that you have. Right? When you have an intern, they come in, and yes, they give you I

you want A, I give you A. I don't think about, "Oh, how is A going to be used later? How can I make A better?" I just

create it and throw it back to you. You tell me if this is good or not. Right? And that's the frustrating part.

Whereas, having someone who think in loops, whether it's you, whether it's your AI agent, whether it's a human

teammate, by learning how to think in loops, they think for themselves. And they recognize, "Oh, actually these

things can be improved better. Actually, I need to think about, oh, how can I improve it?" And I can make suggestions.

"Hey, you said steps 1 to 10, but I think we should add and tweak some of these steps in order to make it better."

And the Japan Airlines CEO, Totori-san, had a really good analogy for this, which is they think about passing on the

baton the best way that they can between teams. Japan Airlines is known for their great quality in customer service in an

industry that have so many moving parts. So, how do they think about it is in loops. They think, "Okay, I am the

ground staff. When I pass the passengers to the flight attendants on the plane, I don't just say, 'Hey, here you go.'

Right? I think about, 'Oh, what are the customers going to need on the flight? What are the flight attendants going to

need to know when people are boarding?' And let me pass that information in a way that helps them. So that once

they're on board, they get the best service, and they're notified of the things that they should know. And the

crew on board then thinks about when the plane lands, what are the things that we need to communicate in order to make

sure the next team has some emphasis to Oh, what should they be focusing on as they're doing their part of the job? And

the cycle continues. So, if you apply it to a work place situation, it could be it's the marketing team thinking about

Okay, how are we going to talk about this product? Thinking that eventually all of the prospects get passed to the

sales team. The sales teams need to know 1 2 3 about each of the customers in order to close a sale. Sales team then

think about the fulfillment team. Okay, we've closed a sale, but the fulfillment team needs to know customer A is really

focused on this one part of the product. So, they need to know that to give that better experience to the prospect to the

buyer. And the fulfillment team then thinks about the next thing, customer service. Okay, these are the questions

we always get. So, let me make sure we pass it on to the customer service so that they can better help uh clients.

Whatever it is, by just thinking about the next part of the loop, you help create this loop thinking inside your

organization, whether it's human-run or AI-run or a hybrid of the two. The fundamental change you're making is to

go from get things done to actually having your point of view on what should be done and how to do them. You can use

AI just as the supercharging engine for getting things done. Right now, you know, everyone's expecting more of you

to do more with less time, so you just say, "Okay, AI, do this. I'm not going to bother, you know, really checking how

things are going. I just need to send these along." Yes, that's one way of doing it, but a better way of thinking

about this technology is to now exercise your point of view on what it is that's worth doing and how you're actually

going to do it, right? By what we talked about earlier, whether you choose AI to be your assistant versus your chief of

staff, what are you going to talk to it about is your point of view on what it is you want to spend your time and

energy on doing. Then is about how to do it. We talked about being able to learn, being able to create, being able to turn

them into workflows. You notice that the workflows are not standardized unless you let AI tell you how to do these

things. Every single step of the of the way we talked about how do you adapt it to your own situation using the loop.

So, what you're doing is exercising your point of view, building your point of view, your skills on how to create a

document might be different from my skills of how to create a document. We have different goals, we have different

experience, we have different of what we decide is right or wrong, good or bad. And so, that is

a very empowering and freeing idea, a very human one. Right? Is that you are at the center unless you give that power

away. You have that choice. When output becomes cheap, then we get a lot of quantity, which can be good and

bad. What we usually do is, "Hey ChatGPT, give me 10 ways of saying this difficult thing I've been thinking about

telling my boss." Or it's, "Hey Claude, give me 10 ways to get from LA to New York." Okay. Fine. This helps us decide

which one is better, which is a good thing. But also it's the same thing that leads to AI slop, right? Is, "Hey, just

create this output." But you notice what we're not thinking about. The next step is quantity begets quality. Yes,

currently AI slop is a problem, right? Because people are still stuck in that getting things done mentality, that

completion logic of I just need you to create output and put it out there. But you don't have to stay there. But you

have to think about the next step, at least for yourself, of how I'm going to use that quantity to get me to better

quality. And for me, quality comes from three things. If you're going from zero to one, it's about better exploration.

Previously, we were limited by our time, by our energy, by how much we can find, by who we have

access to to help us do these things, finding good research, finding good data, finding talking to people, right?

But, with AI, we can think about exploration in a whole new way, right? We can do so much more research that's

targeted, that's wasn't possible before. Quality comes from better loops. So, like what we

talked about, if you have something that you're optimizing towards, you can use feedback. Andrej Karpathy, one of the

founding members of OpenAI, he created this tool called Auto Research, which continuously loop as an agent until it

gets you better results, and it learns how to do all of those and build that feedback into its own skill, right?

That's now possible. And quality comes from better criteria, of your point of view of what's good and what's bad. This

is not an absolute, there is your taste, your curation, your judgment that comes into, okay, I like this red one versus

this blue one, or I like writing that sounds like Naval versus writing that sounds like Nassim Taleb. It doesn't

really matter, but it's for you to be actively involved in the whole process of thinking and creating. Developing

your point of view is crucial. So, coming back to it's not about whether you're senior or junior, it's about how

good you are with AI. And by good, it means how are you thinking about this partner? Where are you positioning

yourself? How much of your yourself are you bringing into this collaboration with this tool? I hope you will bring

that big boss energy. Let me know what you think in the comments below and I'll see you in the next video. Bye.