WEBVTT

00:00:00.000 --> 00:00:02.720
Every morning, you sit down at your bright desk,

00:00:03.060 --> 00:00:06.799
the screen wakes up, you open like 14 different

00:00:06.799 --> 00:00:09.359
tabs in your browser, you type the exact same

00:00:09.359 --> 00:00:12.640
context over and over again. Yeah, it is incredibly

00:00:12.640 --> 00:00:15.119
exhausting. It really is. You feel a strange

00:00:15.119 --> 00:00:18.600
new kind of tired AI was supposed to save your

00:00:18.600 --> 00:00:21.179
valuable time. Right. Instead, it just created

00:00:21.179 --> 00:00:23.899
a brand new daily chore. You feel like a manager

00:00:23.899 --> 00:00:28.100
retraining staff every single day. Exactly. Welcome

00:00:28.100 --> 00:00:31.420
to today's deep dive. We are exploring a fascinating

00:00:31.420 --> 00:00:33.560
shift in how we work. It is such an exciting

00:00:33.560 --> 00:00:36.079
concept to explore today. We are moving away

00:00:36.079 --> 00:00:39.359
from the Amnesiac search box. Today, we are unpacking

00:00:39.359 --> 00:00:41.179
how to build something fundamentally better.

00:00:41.640 --> 00:00:44.859
We are talking about setting up a local AI agent.

00:00:45.000 --> 00:00:47.359
Which changes absolutely everything about your

00:00:47.359 --> 00:00:49.520
digital life. It really does. By the end of our

00:00:49.520 --> 00:00:51.259
conversation, you will understand the mechanics.

00:00:51.399 --> 00:00:53.420
You will know how to build your own personalized

00:00:53.420 --> 00:00:56.189
team. Yeah, a tireless digital worker on your

00:00:56.189 --> 00:00:58.670
own machine. Right, one that actually remembers

00:00:58.670 --> 00:01:01.170
your files and runs on a reliable schedule. So

00:01:01.170 --> 00:01:04.329
to escape that endless chat box fatigue, we really

00:01:04.329 --> 00:01:06.709
have to pivot our thinking. We have to understand

00:01:06.709 --> 00:01:09.569
what makes an autonomous agent completely different.

00:01:09.750 --> 00:01:12.390
Right, because it is not just a smarter version

00:01:12.390 --> 00:01:15.989
of a standard chatbot. It starts with its underlying

00:01:15.989 --> 00:01:19.269
physical and digital anatomy. Let's unpack that.

00:01:19.629 --> 00:01:22.090
I know a local AI agent has seven core functional

00:01:22.090 --> 00:01:25.040
parts. It does. And once you see them clearly,

00:01:25.400 --> 00:01:28.879
the setup feels much less intimidating. First,

00:01:29.099 --> 00:01:31.680
it absolutely needs a physical machine to live

00:01:31.680 --> 00:01:33.780
on. Right. Let's talk about the physical reality

00:01:33.780 --> 00:01:35.900
of this setup. You are giving this digital worker

00:01:35.900 --> 00:01:38.859
a dedicated physical home. Second is a functional

00:01:38.859 --> 00:01:41.480
mouth and ears. Yeah, which is simply a dedicated

00:01:41.480 --> 00:01:43.540
channel to send messages back and forth. Third

00:01:43.540 --> 00:01:45.680
is the brain, which is your chosen AI model.

00:01:45.879 --> 00:01:48.700
Exactly. Fourth is memory. And this is surprisingly

00:01:48.700 --> 00:01:51.079
just plain text files. We will definitely dig

00:01:51.079 --> 00:01:52.920
deep into those simple text files later. Oh,

00:01:53.000 --> 00:01:55.739
we will. Fifth, the agent needs functional tools

00:01:55.739 --> 00:01:59.400
to operate. Meaning having web search, file readers,

00:01:59.700 --> 00:02:02.739
or screenshot takers. Right. Sixth is the heartbeat,

00:02:03.000 --> 00:02:05.439
which is a continuous running schedule. And finally,

00:02:05.480 --> 00:02:08.219
it has eyes to read your actual screen directly.

00:02:08.500 --> 00:02:12.120
That is complete anatomy. Let's beat. Let's pause

00:02:12.120 --> 00:02:13.939
and consider the scope of this architecture.

00:02:14.280 --> 00:02:16.860
You do not need every single part on day one.

00:02:17.060 --> 00:02:19.400
No, you can start incredibly small and build

00:02:19.400 --> 00:02:21.960
up over time. But hardware choices are where

00:02:21.960 --> 00:02:24.759
most enthusiastic people get stuck. Why is your

00:02:24.759 --> 00:02:28.419
daily laptop a truly bad choice here? Well, because

00:02:28.419 --> 00:02:31.719
a true autonomous agent runs 24 hours a day.

00:02:32.060 --> 00:02:34.680
You close your work laptop and shut it down constantly.

00:02:34.860 --> 00:02:37.379
Right. So whenever the lid closes, the agent

00:02:37.379 --> 00:02:39.800
simply goes to sleep. I think about it like an

00:02:39.800 --> 00:02:43.479
office layout problem. It is like moving an employee

00:02:43.930 --> 00:02:46.889
from a temporary hot desk. I love that analogy.

00:02:47.069 --> 00:02:48.990
You take them out of the busy, disconnected public

00:02:48.990 --> 00:02:51.770
cloud. You give them a permanent, dedicated corner

00:02:51.770 --> 00:02:53.849
office inside your house. Yeah. They live on

00:02:53.849 --> 00:02:55.830
your laptop. You keep packing up their desk.

00:02:55.870 --> 00:02:58.030
That is a brilliant way to visualize the hardware

00:02:58.030 --> 00:03:01.050
constraint. You really want a clean, secondary

00:03:01.050 --> 00:03:03.949
machine solely for your agent. An old, wiped

00:03:03.949 --> 00:03:06.550
laptop is a great entry -level setup today. Right.

00:03:06.789 --> 00:03:09.870
And later, you might want a $400 used Mac Mini.

00:03:10.050 --> 00:03:12.449
You put it on a shelf, and it runs day and night.

00:03:13.370 --> 00:03:15.379
But... We need to deeply understand the actual

00:03:15.379 --> 00:03:19.740
system limits here. Why is RAM the absolute deal

00:03:19.740 --> 00:03:22.500
breaker for local hardware setups? Right. And

00:03:22.500 --> 00:03:24.300
this is crucial to understand computationally.

00:03:24.840 --> 00:03:27.800
A local model is an AI brain running entirely

00:03:27.800 --> 00:03:30.580
on your own machine. OK. That brain is essentially

00:03:30.580 --> 00:03:33.240
a massive file of statistical probabilities.

00:03:33.780 --> 00:03:36.800
So to think properly, it must sit entirely in

00:03:36.800 --> 00:03:39.770
your active memory. Ah, I see. If you lack RAM,

00:03:40.150 --> 00:03:42.669
the model simply cannot process anything. The

00:03:42.669 --> 00:03:44.750
machine will crash trying to load the neural

00:03:44.750 --> 00:03:48.270
weights. Got it. 16 gigs minimum to fit the model's

00:03:48.270 --> 00:03:51.150
brain locally. Exactly. That is the soft floor

00:03:51.150 --> 00:03:54.389
for running anything meaningful. BD. So the agent

00:03:54.389 --> 00:03:57.610
has a physical permanent corner office. How does

00:03:57.610 --> 00:03:59.810
it think? And how do we naturally communicate

00:03:59.810 --> 00:04:02.349
with it? I mean, we cannot log into that dedicated

00:04:02.349 --> 00:04:04.710
machine every five minutes. That completely defeats

00:04:04.710 --> 00:04:06.530
the whole purpose of background automation. You

00:04:06.530 --> 00:04:08.330
want a quick, seamless message right on your

00:04:08.330 --> 00:04:10.310
phone. So you need a reliable mouth and ears

00:04:10.310 --> 00:04:12.770
for the agent. Right. Telegram is the absolute

00:04:12.770 --> 00:04:14.750
easiest messaging channel to start with. You

00:04:14.750 --> 00:04:17.949
can set up a basic secure bot in minutes. What

00:04:17.949 --> 00:04:20.389
about managing multiple different research projects

00:04:20.389 --> 00:04:23.800
at once? Discord works much better for that specific

00:04:23.800 --> 00:04:26.480
organizational need. You can use separate dedicated

00:04:26.480 --> 00:04:28.839
channels for separate daily jobs. Okay, that

00:04:28.839 --> 00:04:31.040
makes sense. But, you know, if you use the Claude

00:04:31.040 --> 00:04:34.189
Cowork ecosystem... Dismatch is best. It is the

00:04:34.189 --> 00:04:36.610
smoothest option for sending automated tasks

00:04:36.610 --> 00:04:39.850
directly. Let's explore the actual computational

00:04:39.850 --> 00:04:43.009
brain of this operation. In this system, you

00:04:43.009 --> 00:04:45.110
choose different functional versions of Claude.

00:04:45.189 --> 00:04:48.490
Yeah. It is exactly like hiring the right employee

00:04:48.490 --> 00:04:51.089
for the task. You have three primary brains to

00:04:51.089 --> 00:04:53.569
choose from right now. Claude Opus is the smart

00:04:53.569 --> 00:04:56.889
architect for deep complex logic. Right. It handles

00:04:56.889 --> 00:04:59.509
intricate planning and heavy multi -step research

00:04:59.509 --> 00:05:02.370
very well. But it uses significantly more compute

00:05:02.370 --> 00:05:04.670
credits than takes time. It does. Then you have

00:05:04.670 --> 00:05:07.370
Claude Sonnet, the reliable daily manager. This

00:05:07.370 --> 00:05:10.089
is the absolute best choice for most routine

00:05:10.089 --> 00:05:13.089
jobs. And the third one. Finally, there is Claude

00:05:13.089 --> 00:05:15.689
Haiku. It is the incredibly fast quick helper.

00:05:16.029 --> 00:05:18.610
It is lightning fast for simple sorting and basic

00:05:18.610 --> 00:05:20.970
checking. This brings us to a very important

00:05:20.970 --> 00:05:24.230
operational boundary. There is a golden rule

00:05:24.230 --> 00:05:27.310
for managing these distinct brains. Oh, absolutely.

00:05:27.550 --> 00:05:29.870
Do not use the smartest brain for the simplest

00:05:29.870 --> 00:05:34.069
task. I still wrestle with defaulting to the

00:05:34.069 --> 00:05:37.470
smartest model, just out of sheer laziness. Yeah,

00:05:37.649 --> 00:05:39.930
you are definitely not alone in that expensive

00:05:39.930 --> 00:05:42.610
habit. It is so tempting. But it wastes money

00:05:42.610 --> 00:05:46.290
and heavily slows down your system. Computationally,

00:05:46.689 --> 00:05:48.810
a massive model takes longer to generate the

00:05:48.810 --> 00:05:51.949
first token. Start with Sonnet and only upgrade

00:05:51.949 --> 00:05:54.930
when it genuinely fails. Let me ask about the

00:05:54.930 --> 00:05:57.529
underlying architecture of these brands. Does

00:05:57.529 --> 00:06:00.110
staying within one specific model family actually

00:06:00.110 --> 00:06:02.560
matter? It matters a great deal for consistency

00:06:02.560 --> 00:06:04.860
and overall security. Mixing different models

00:06:04.860 --> 00:06:07.279
means they might format system answers differently.

00:06:07.420 --> 00:06:09.860
Oh, interesting. Yeah, a specific prompt optimized

00:06:09.860 --> 00:06:13.360
for Claude might completely confuse Llama. Staying

00:06:13.360 --> 00:06:15.339
within the Claude family keeps your entire agent

00:06:15.339 --> 00:06:18.019
predictable. Makes sense. Sticking to one family

00:06:18.019 --> 00:06:20.500
keeps the whole system fast and secure. It is

00:06:20.500 --> 00:06:22.740
the safest baseline for anyone building locally.

00:06:23.040 --> 00:06:26.699
to sex silence. So a smart brain is totally useless

00:06:26.699 --> 00:06:29.980
without historical context. It is incredibly

00:06:29.980 --> 00:06:32.199
frustrating if it suffers from daily amnesia.

00:06:32.399 --> 00:06:35.040
That is the fundamental flaw of regular web -based

00:06:35.040 --> 00:06:38.040
chatbots. Every single chat session starts entirely

00:06:38.040 --> 00:06:40.480
from zero background knowledge. But memory for

00:06:40.480 --> 00:06:43.180
local agents is surprisingly low -tech, conceptually.

00:06:43.279 --> 00:06:45.339
It is It's beautifully low tech when you look

00:06:45.339 --> 00:06:48.839
under the hood. It is just plain text files sitting

00:06:48.839 --> 00:06:52.160
in a local folder. Just plain text. Yeah. A simple

00:06:52.160 --> 00:06:56.240
markdown file called Claude .md holds the background

00:06:56.240 --> 00:06:59.519
data. It specifically holds the persona, the

00:06:59.519 --> 00:07:02.300
current goals, and the workflow. Right. Large

00:07:02.300 --> 00:07:04.980
language models need this context injected into

00:07:04.980 --> 00:07:07.319
their processing window. It feels like an impressive

00:07:07.319 --> 00:07:09.459
illusion of magic sometimes. It really does.

00:07:09.860 --> 00:07:12.920
Yeah. Whoa. Imagine scaling to a billion tasks

00:07:12.920 --> 00:07:15.920
all grounded by a few plain text files. It is

00:07:15.920 --> 00:07:18.660
amazing. It is so elegant and completely transparent

00:07:18.660 --> 00:07:21.519
for humans to read. But static memory alone does

00:07:21.519 --> 00:07:23.819
not get the actual work done. Tools give the

00:07:23.819 --> 00:07:26.860
agent actual digital hands to do things. Exactly.

00:07:26.959 --> 00:07:29.819
Tools let the agent read local files and search

00:07:29.819 --> 00:07:32.439
the web. You can connect your Google Drive or

00:07:32.439 --> 00:07:35.819
your daily calendar. So the agent uses API calls

00:07:35.819 --> 00:07:39.079
to interact with external systems. Yes. It reaches

00:07:39.079 --> 00:07:41.610
out and does the work. There is a serious security

00:07:41.610 --> 00:07:44.569
warning we need to discuss here. Blind trust

00:07:44.569 --> 00:07:47.449
in downloaded shared skills is a real vulnerability.

00:07:47.990 --> 00:07:50.970
Yes. IBM and Palo Alto actively flagged this

00:07:50.970 --> 00:07:53.670
major risk. A shared skill is ultimately just

00:07:53.670 --> 00:07:56.269
arbitrary code running locally. Which is dangerous.

00:07:56.689 --> 00:08:00.449
Very. A bad shared skill can quietly steal your

00:08:00.449 --> 00:08:02.970
personal files. It can send your private data

00:08:02.970 --> 00:08:05.660
completely outside your secure network. So you

00:08:05.660 --> 00:08:08.319
must have Claude review any new skill code first.

00:08:08.360 --> 00:08:10.120
You do this before you ever run it on your machine.

00:08:10.319 --> 00:08:12.579
Always. Let me ask about how the memory actually

00:08:12.579 --> 00:08:15.019
functions dynamically. How does an agent actually

00:08:15.019 --> 00:08:17.360
read a plain text file without getting overwhelmed?

00:08:17.939 --> 00:08:19.660
Well, it does not keep the text running forever

00:08:19.660 --> 00:08:22.180
in memory. It simply scans that file immediately

00:08:22.180 --> 00:08:25.420
before it acts. I see. It loads the rules, completes

00:08:25.420 --> 00:08:27.639
the assigned task, and clears itself. Simple.

00:08:27.779 --> 00:08:29.699
It scans the text file right before starting

00:08:29.699 --> 00:08:32.639
any new task. Exactly. Now that our agent can

00:08:32.639 --> 00:08:35.200
think and act securely, we need to make this

00:08:35.200 --> 00:08:38.360
digital worker fully proactive. It should not

00:08:38.360 --> 00:08:41.019
wait for us to ask it anything. This is where

00:08:41.019 --> 00:08:43.340
the heartbeat changes the whole operational game.

00:08:43.759 --> 00:08:46.460
The heartbeat is a schedule that runs fixed background

00:08:46.460 --> 00:08:48.440
jobs. In traditional programming, we call this

00:08:48.440 --> 00:08:51.820
a basic cron job. And the 7 a .m. daily brief

00:08:51.820 --> 00:08:54.539
is a truly great example. It really is. It is

00:08:54.539 --> 00:08:58.039
the perfect scheduled task to copy and try today.

00:08:58.840 --> 00:09:01.279
Every morning the agent quietly gathers three

00:09:01.279 --> 00:09:04.059
top news headlines. It checks your daily calendar

00:09:04.059 --> 00:09:08.039
events and one key metric. It notes one small

00:09:08.039 --> 00:09:10.620
win or potential worry from yesterday. And it

00:09:10.620 --> 00:09:13.539
tightly caps all of this under 250 words. It

00:09:13.539 --> 00:09:15.799
easily saves you 30 minutes of scattered morning

00:09:15.799 --> 00:09:18.379
reading. You wake up and a concise summary is

00:09:18.379 --> 00:09:21.409
already waiting. Beat. But many eager beginners

00:09:21.409 --> 00:09:23.809
fall into a very common trap here They try to

00:09:23.809 --> 00:09:26.350
build one giant agent for absolutely everything

00:09:26.350 --> 00:09:28.470
all the time This directly leads to something

00:09:28.470 --> 00:09:31.350
we call context bloat Contest bloat is giving

00:09:31.350 --> 00:09:35.250
an AI too much data. So it forgets. Yes, it becomes

00:09:35.250 --> 00:09:38.429
sluggish forgets details and gets terribly confused

00:09:38.429 --> 00:09:40.610
Let's expand on that idea with a real -world

00:09:40.610 --> 00:09:43.690
analogy. It is like a stressed restaurant manager

00:09:43.690 --> 00:09:46.230
working completely alone, right? They are trying

00:09:46.230 --> 00:09:49.049
to cook serve tables and do the taxes all at

00:09:49.049 --> 00:09:51.850
once The attention mechanism gets heavily divided

00:09:51.850 --> 00:09:54.710
across too many simultaneous tasks. And the final

00:09:54.710 --> 00:09:58.750
results will be messy and incredibly slow. When

00:09:58.750 --> 00:10:01.629
you stuff a massive prompt into an AI model,

00:10:01.649 --> 00:10:04.389
it breaks. The mathematical attention head simply

00:10:04.389 --> 00:10:07.110
cannot track every single relationship. No, they

00:10:07.110 --> 00:10:10.490
can't. The optimal solution is a small, specialized

00:10:10.490 --> 00:10:13.090
team of different agents. Each digital worker

00:10:13.090 --> 00:10:16.289
has a tiny memory and one distinct goal. Let's

00:10:16.289 --> 00:10:18.549
carefully break down this ideal starter team

00:10:18.549 --> 00:10:21.200
of four. First is the dedicated scout. This is

00:10:21.200 --> 00:10:23.519
your active research agent. It uses Claude Sonnet

00:10:23.519 --> 00:10:26.399
to quickly scan daily news and trends. Next is

00:10:26.399 --> 00:10:28.820
the creator, which uses the heavier Claude Opus.

00:10:28.919 --> 00:10:31.460
It does deep thinking to turn raw research into

00:10:31.460 --> 00:10:34.320
drafts. What about keeping the actual local files

00:10:34.320 --> 00:10:37.080
organized and safe? That is the admin agent,

00:10:37.139 --> 00:10:39.500
also using the Claude Sonnet brain. Its only

00:10:39.500 --> 00:10:42.240
job is actively managing folders and your daily

00:10:42.240 --> 00:10:44.950
calendar. And the fourth one. Finally, the guard

00:10:44.950 --> 00:10:48.370
agent uses Haiku for rapid security checks. It

00:10:48.370 --> 00:10:51.149
verifies outputs before anything gets sent or

00:10:51.149 --> 00:10:53.889
saved locally. Does managing four distinct agents

00:10:53.889 --> 00:10:56.769
actually become harder than managing one? Surprisingly,

00:10:57.070 --> 00:10:59.889
no. They do not cross paths or share confusing

00:10:59.889 --> 00:11:02.250
operational rules. They do exactly what they

00:11:02.250 --> 00:11:05.169
are explicitly told and stop. Right. The scout

00:11:05.169 --> 00:11:07.710
hands data to the creator cleanly and efficiently.

00:11:07.950 --> 00:11:10.429
Right. Separating them means absolute accuracy

00:11:10.429 --> 00:11:13.169
and you only pay for what you use. Specialization

00:11:13.169 --> 00:11:15.789
is exactly how you scale any business with AI.

00:11:16.090 --> 00:11:19.690
Two -sec silence. High -level theory and digital

00:11:19.690 --> 00:11:22.190
structure are great to discuss here. But how

00:11:22.190 --> 00:11:24.070
does the listener actually turn this on tonight?

00:11:24.220 --> 00:11:26.879
You start simply inside the Claude desktop app

00:11:26.879 --> 00:11:29.600
using Cowork. You make one single folder on your

00:11:29.600 --> 00:11:32.019
dedicated local machine. Just one folder? Yeah.

00:11:32.320 --> 00:11:35.080
You write your plaintext Claude .md memory file

00:11:35.080 --> 00:11:38.279
right there. Then you set one single scheduled

00:11:38.279 --> 00:11:40.840
task to run tomorrow. It really is that straightforward

00:11:40.840 --> 00:11:43.340
to get it moving initially. It is. And once you

00:11:43.340 --> 00:11:46.000
are entirely comfortable, you can build something

00:11:46.000 --> 00:11:48.620
cooler. You can build a self -refreshing HTML

00:11:48.620 --> 00:11:51.220
daily dashboard for yourself. That sounds a bit

00:11:51.220 --> 00:11:53.279
more complicated for a typical beginner. A little

00:11:53.279 --> 00:11:56.450
bit. Is no code really enough, or are we just

00:11:56.450 --> 00:11:58.789
delaying the inevitable need to learn Python?

00:11:58.940 --> 00:12:01.480
No code will actually get you 90 % of the way

00:12:01.480 --> 00:12:04.179
there. The tool calling mechanism completely

00:12:04.179 --> 00:12:06.919
abstracts the complex coding layer away. That

00:12:06.919 --> 00:12:10.379
is a relief. You do not need Python just to organize

00:12:10.379 --> 00:12:13.539
your daily life. Local AI agents are definitely

00:12:13.539 --> 00:12:17.059
not just a passing internet trend. It is a massive

00:12:17.059 --> 00:12:19.200
fundamental industry shift happening right now.

00:12:19.480 --> 00:12:22.379
Absolutely. Anthropic, Microsoft, and Notion

00:12:22.379 --> 00:12:24.580
are betting heavily on this architecture. We

00:12:24.580 --> 00:12:27.379
are shifting entirely from passive chat to autonomous

00:12:27.379 --> 00:12:30.129
action. The people who embrace this fundamental

00:12:30.129 --> 00:12:32.730
shift will move much faster. What separates the

00:12:32.730 --> 00:12:34.809
people who succeed with this from the people

00:12:34.809 --> 00:12:37.629
who give up? The ones who fail try to build the

00:12:37.629 --> 00:12:40.370
whole company overnight. They get heavily frustrated

00:12:40.370 --> 00:12:42.889
by early errors and abandon the project. Right.

00:12:43.309 --> 00:12:46.330
The successful ones start with a single, highly

00:12:46.330 --> 00:12:49.509
specific automated task. It's all about patience.

00:12:49.809 --> 00:12:52.350
Start with one simple task, then build from there.

00:12:52.590 --> 00:12:55.090
Exactly. Let the basic system run quietly for

00:12:55.090 --> 00:12:58.600
a full week. Let's take a step back and view

00:12:58.600 --> 00:13:01.820
the whole picture. We are summarizing the ultimate

00:13:01.820 --> 00:13:04.480
takeaway of this technological shift. We are

00:13:04.480 --> 00:13:08.059
moving away from a needy blank chat window. We

00:13:08.059 --> 00:13:11.600
are moving toward a quiet, capable, local, digital

00:13:11.600 --> 00:13:14.600
colleague. You simply combine a dedicated machine

00:13:14.600 --> 00:13:17.899
and specialized AI brains. You add text -based

00:13:17.899 --> 00:13:20.620
memory and a reliable scheduled daily heartbeat.

00:13:21.000 --> 00:13:23.240
You literally buy back 30 minutes of your life

00:13:23.240 --> 00:13:26.889
daily. That is the real tangible promise of local

00:13:26.889 --> 00:13:30.529
AI agents. There is no magic, no hype, just steady

00:13:30.529 --> 00:13:32.769
and quiet work. It happens seamlessly in the

00:13:32.769 --> 00:13:34.889
background while you sleep peacefully. I want

00:13:34.889 --> 00:13:36.590
to leave you with a deeply provocative thought.

00:13:36.870 --> 00:13:39.250
If your digital team is doing the busy work while

00:13:39.250 --> 00:13:41.590
you sleep, what are you going to do with that

00:13:41.590 --> 00:13:44.870
perfectly reclaimed 30 minutes? When the constant

00:13:44.870 --> 00:13:47.529
friction of daily organization completely disappears,

00:13:48.009 --> 00:13:50.429
what will you actually build? Wipe an old laptop

00:13:50.429 --> 00:13:52.210
clean and just try it out. Just try making a

00:13:52.210 --> 00:13:54.389
single text file tonight. Thank you for joining

00:13:54.389 --> 00:13:56.990
us on this deep dive. Our T -Row music.