September 2, 2025 • 38 mins
Join the Tool Use Discord: https://discord.gg/PnEGyXpjaX
Unlock massive productivity gains with AI and get hours of your life back. In Episode 55, after 10 years as a software developer with the last three focused on AI, I'm sharing the secrets to my personal productivity system. Learn how to leverage multiple LLMs, build a "digital brain" with Obsidian that goes beyond a simple second brain, and create hyper-custom automations to focus on what truly matters.
I'll walk you through my complete setup, from the AI models I use daily (like Claude and local models) to the hardware and software I use to interface with my computer, including powerful voice transcription workflows. We'll cover how to use AI coding assistants to build your own scripts, essential terminal utilities that will speed up your workflow, and the specific media tools I use for tasks outside my expertise. This is the ultimate guide to leveling up your productivity with AI.
Mike's GitHub: https://github.com/MikeBirdTech
Obsidian: https://obsidian.md/
Mike's Stack: https://github.com/MikeBirdTech/stack
Dataview Plugin: https://obsidian.md/plugins?id=dataview
Connect with me
https://x.com/ToolUseAI
https://x.com/MikeBirdTech
00:00:00 - Intro
00:00:52 - Which LLM Should You Use?
00:04:17 - How to Build a "Digital Brain" in Obsidian
00:12:40 - Voice Transcription: The Biggest Productivity Unlock
00:17:59 - Building Custom Automations with AI Coding
00:32:06 - Using AI as a Thought Partner
Subscribe for more insights on AI tools, productivity, and automation.
Tool Use is a weekly conversation with the top AI experts, brought to you by ToolHive.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
The most common question people ask me these days is how I use
AI to become more productive. Those real roductivity gains,
not the demo where the things that give me hours back every
single day. Well, it's time to share some
secrets. In 10 years of being a software
developer with the last three focused on AI tooling, I've
tested out hundreds of differenttools and built out some great
workflows and automations to improve my productivity.
(00:21):
On top of the dozens of tool useconversations, I've been hired
as a productivity coach and I'vebuilt out the AI Academy of Box
1 Ventures, which has upscaled the entire team on improving the
productivity. So in episode 55 of Tool Use,
brought to you by Tool Live, I'mgoing to teach you how I've
levelled up my productivity withthe use of AI.
These workflows will give you hours of your life back.
The automations are going to allow you to focus on what
(00:43):
really matters. I'll be going over my digital
brain set, how I interface with my computer, how I build these
Hypercustom automations and a bunch of my tools.
Let's get into it. So obviously you can't have a
conversation about leveraging AIto become more productive until
you start talking about which models you use, which LLM
providers get you the best gains.
It's the one thing people go back and forth with all the
(01:03):
time. So for me, the model that I use
the most is all of them, which Iknow is kind of a non answer,
but there's a good reason for it.
You don't want to get locked into building on a specific
platform because once you get locked in, it makes it a lot
harder to shift. And as these models start
changing the way that prompts affect them and changing the
ways interface with tooling, there will be a divergent where
we're not going to have the samelevel of systems being model
(01:25):
agnostic as we used to. So it's going to be very
important that you start gettingused to the way that they act in
different situations form that intuition, start realizing which
models work better in some situations because not only do
new models come out, that's going to keep changing, but the
providers of the models actuallywill change the amount of
compute that they give towards resources.
They might nerf them at certain times or change the way that
(01:46):
their test time compute works. So it's very important to just
get a general intuition. When you give an input, what
output do you expect? How does it change across
providers? So if you get locked into using
one single provider for everything that you build,
you're going to lose that flexibility and observe how
different models were. So it's important to think of it
less as choosing a team that you're on and more of aligning
(02:08):
it with in your real life. Who would you go to for advice?
Depends on the topic. Maybe it's a friend or mom and
dad or all of the above. And you really do not have a
static answer for who's the bestperson to go for for advice.
And now we have access to intelligence that's just an API
Callaway, a chat message away. And this is something that's
brand new to us. We don't know how to best use
(02:29):
it. We're still figuring it out.
And everyone has different preferences and different
methods of interacting with it. So there is no best answer.
And the spectrum of possibilities changes every
single day. And even though that might feel
over whelming, reflecting on it really shows how incredibly
miraculous it is that we've beenable to turn math and
electricity into intelligence and we want everyone to be able
(02:49):
to benefit from it. And there is no silver bullet.
So the best of next to have is use it, play with it, explore,
experiment, immerse yourself in it and see what happens.
So my general setup is I have Clawed desktop and Jan running
on my computer. Clawed is for the big things,
Jan is for the easy questions. I personally think it's worth
the extra effort to direct your questions to the appropriate
model because do you really needthat massive model to get a
(03:11):
definition or translation? Local models use way less
electricity and for the rest of the LMS I use web interfaces.
I don't really have a reason, it's just kind of fall into my
habit. Like I'll still use Open AI,
Grok, Gemini, they all are in the regular routine of models I
check in on. Sometimes I use Deepseeker
Hermes, but the open source onesI tend to use locally and I
really don't have justification for this yet.
(03:32):
It's just by falling into. I do try to stay on top of
testing new things so that'll probably change as time goes on.
I encourage you to play as well.I will say though, one big piece
of advice I have is to set up a system so that you can easily
switch between these models and test them out in comparison.
Create your eval suite, create your AI service.
Check out my GitHub for some examples of those.
But if you're able to take control over the experimentation
(03:54):
process to actually efficiently see when a new model comes out,
how it works with your workflow,how it compares to the other
ones, you're going to be more rapidly able to adapt these
changes. So don't just think about which
models you use, think about which systems you need to set up
to compare the models. So at the snapshot in time for
your specific use case, you can easily discover what the best
(04:15):
model is. But if I had to choose, I'd
probably say Claude. The next thing I wanted to touch
on is the aspect of my digital brain.
So you might have heard people talk about the second brain,
which will be a system setup to store and retrieve their notes.
So they're able to get everything consolidated in a
digital way so they can easily search and retrieve things.
On the age of AI, we can take itso much further.
So I've really encourage everyone, and you've probably
(04:36):
heard me say this 100 times before in the show to try out
Obsidian. Obsidian has many perks and
especially the age of AI, it becomes very powerful.
Before I get into the specifics of how I use Obsidian to create
a digital brain, I just want to go over 3 principles.
Why I've chosen Obsidian over all the rest of the note taking
apps. So the first is the file over
app principle. What this means is you control
(04:57):
your data. The A is simply an interface for
you to interact with that data. But if you didn't like Obsidian,
you could get rid of it today and still keep all of your
information and easily port it over to a new system.
There's some tools like Notion which become quite popular,
especially with their AI integration, but it's very
limited. And if you try to export your
Notion data, it is a mess. Good luck importing anywhere
(05:17):
else. They really try to keep you
locked in. And like I mentioned earlier,
you don't want to be locked. In this day and age, everything
moves so fast and evolves so aggressively that you want that
flexibility of controlling your data and being able to take with
you wherever you want. The next one is the ease of AI
to be able to read Markdown. Markdown is nicely structured, a
very simple syntax, so the learning curve is very small and
(05:38):
it just allows things to be structured in a way that AI can
easily understand. There is benefits to using
things like XML tags and whatnot, but those are so hyper
specific that they really works better for prompts rather than
structuring your data. And the third aspect of Obsidian
I love is the plugins. Plugins give you so much
capability. There is templates that you can
go on GitHub to download from, from the Obsidian team to build
(05:59):
your own. I developed the Obsidian plugin
generator. We can have a simple chat with
AI to develop your own plugin, but these plugins allow
functionality not just within the vault, but within your
computer as a whole. And it exposes different APIs so
you can really easily interact with things.
And I got a few exciting ones toshow you.
The first thing that I want to touch on with my Obsidian setup
is the dashboard. So I have a dashboard that loads
whenever I open Obsidian and it dynamically generates the things
(06:21):
that I need to see most. So we have tasks, what's today's
task? What's overdue, what are
priorities. This is dynamically generated
using things like data view plugin and tasks plug in, which I'll
link both to. So by leveraging these, I'm able
to create tasks in my daily notes, which I use with the
calendar plug in and have a collection of of tasks that I
(06:44):
get from meetings or that I meetthroughout the day.
And then I have it centralized in a specific dashboard so I can
quickly see at the start of my day or the start of a workout
what I need to prioritize most. Also, because I use my vault for
personal as well as professionalstuff, I have my health and
fitness pages linked to it, so Ican easily see where I'm at.
My rowing program, I have a section for the next two years
episode. We'll link to the notes and
(07:05):
that's dynamically generated based on what's coming up next.
I have stats system, health important links.
I have things that are developedfor me.
And The thing is, with these dashboards, you can make it
completely tailored to your use cases.
What things do you want to see? What would help you prioritize
your day? By having a dynamically
generated dashboard, you're ableto really quickly get a
(07:26):
glanceable view at what you haveto do and then be able to feed
those inputs like checking off atask or adding something.
You do a list, it'll feed through your system to where it
needs to be. 1 community plug inthat I really wanted to
emphasize is data view. I think data view is so powerful
it might be a little intimidating because you are
writing code. It supports both its own unique
query language as well as Java, JavaScript, it just which makes
(07:48):
it so powerful. If you struggle with it, you can
have clawed code running in yourObsidian Vault, looking at the
different dashboards and updating it for you.
So you don't have to necessarilydive too deep into the docs,
Although getting an idea of how it works will be helpful.
So I still encourage you to. But clawed code will be able to
help get you most of the way there.
And then you can literally go back and forth to see what
(08:09):
works. Maybe you want to pull tasks
just from your daily notes, or maybe you also want to include
it from your meetings, your people, your company lists.
All of that is possible with data.
And then on top of these community plugins that I
recommend, I actually have developed more custom plugins
than what I found on their community listing.
And that's because of my Obsidian plugin generator.
I can really quickly spin these things up.
(08:29):
So I use my Obsidian Vault as a CRM.
Every meeting I have, every person that I meet, every
company they work for that is stored in my Obsidian Vault and
it's linked to the different meetings and daily notes and
whatnot because it's very valuable for me to be able to
quickly go through the flow of information and with it can
follow these links. So I have one set up to do
(08:51):
automatic meeting stuff, whetherit's pulling in data from
granola or generating to do's that we can follow up with later
or even being able to feed out into e-mail systems.
All that is hosted in meetings. And then of course with tool
use, I have some podcast automation.
So every time there's a new episode, it'll generate the new
directory along with the required documents and it'll get
that set up. I've taken it one step further
(09:11):
to when I finish editing and I export the transcript, the
podcast automations are actuallyable to generate AI suggestions
for titles, which are always terrible.
We're not getting good at titlesunfortunately.
These two descriptions decent, but even things like
manipulating the thumbnail so I can take it from the regular 16
by 9 thumbnail and make it square for Spotify or other
(09:34):
podcast show art. My Obsidian is able to do that.
So you're not necessarily stuck with doing automations within
Obsidian. You can get system processes or
interact on files and systems outside of your app outside of
Obsidian to make yourself more automated.
So that that's why I'm considering more of a digital
brain than a second brain because you're able to induce
actions. It's not just storing and
(09:56):
retrieving memories. You're able to actually act upon
it, get intelligence from it, doa whole lot of things.
So it's phenomenal. And one thing I definitely want
to give a shout out to Steph Ango or also know online is Pano
is the creator of Obsidian and he has a phenomenal writing.
Check out his website so you canread more.
There's a lot of great things, but he has some principles that
he talks about with the way he sets up his Obsidian vault.
(10:16):
And I really like a lot of them.So he likes to use links and
categories and tags instead of nested directories.
So you're able to really deeply link different files to each
other with Obsidian with wiki link style.
So you can just simply make a reference to a person and link
to the person's page so you're able to more rapidly get context
on that person with nested directories.
It's nice for a simple UI to click around, but you can
(10:38):
rapidly, with Command O on Mac, open up the notes that you're
looking for. So you don't necessarily need to
use the UI for everything where your fingers are already on the
keyboard. You can just learn a few
keyboard shortcuts and rapidly switch from the notes that you
need to be focused on. Another concept he mentioned,
this fractal journey and random revisiting, which I think is
great. So you can really quickly spin
up a partial note on Obsidian and then pull them together.
(11:00):
So you can say all the notes made on this day can go into a
centralized location and then later on I can review those
notes with his random revisitingand then figure out where these
things actually belong to live. So you don't have to be
constantly going through your vault to figure out what goes
where and where information should live.
You can randomly visit it later at a better time for you and
just simply move things to the right location.
(11:21):
Properties and templates to set up new nodes.
I have a few templates set up. I definitely want to get more
but it makes it easy to pre populate different bits of
information and later on as you gather more information, say you
have a meeting with someone in abrand new company.
So you just simply have the company name, quick description
of what they do later on, you find out their website or their
LinkedIn page, all their other socials.
You're able to add that into thetemplate later on.
(11:42):
So as I said, check out his website to get more of his
writing. It's excellent.
He does a lot of design stuff aswell, which I think is
phenomenal and it really reflects in how powerful,
flexible and enjoyable it is to use Obsidian.
The way you interact with a computer might be the biggest
opportunity for improving your productivity.
We've been using computers the same way for ages, keyboard,
mouse, typing, and it's really not optimal for bandwidth. 10
(12:05):
fingers cannot not beat speakingspeed for an example.
So what you'll want to do is reflect on the way that you use
a computer, figure out what tools exist out there that might
better align your optimal workflow with the way you
interact with the computer, and then just start exploring and
playing with it. Because we don't have to be
locked into these old paradigms.There is so much more
opportunity for us to explore now.
(12:26):
So I'm going to go over a few the ways that I interface with
my computer, both software and hardware, and then just give you
a little inspiration, things youcan explore because it might
work for you, it might not. Everyone's is different, but the
purpose is to make sure that youexplore the opportunities to
find out what works best for you.
So by far the biggest unlock that I've had has been voice
transcription. It's made such a huge difference
(12:46):
in my productivity. I can speak 10 times faster than
I can type. When I get excited, it's
probably like 20 times faster. And these modern transcription
tools are just so good. You can add custom vocabulary,
you can have corrections. I even have it where I say
insert my YouTube channel and it'll put the URL instead of
just verbatim transcribing it. And now I know that voice
transcription is not appropriatefor every situation.
(13:06):
If you work in an open space, open concept office, it's not
going to be great for you just rambling at your computer all
day. But you should try to leverage
it in the situations that do work for you.
Maybe you go for a walk and you just like take notes as you go.
Or you can book a meeting room. When you have this big piece of
work you have to get through with speaking, you take
advantage of different neural pathways than when you're
typing. It really gives you a different
output. And if you start prompting an
(13:27):
LLM with speaking, sometimes typing, you can start seeing how
you approach it differently and how you interact with it
differently. And I think it can be massively
beneficial. So right now I use Aqua voice.
It's good. I like it.
I don't think I'll use it forever.
It has had a few flaws, but for the most part it's good.
I used to use super whisper alsogood.
It's Mac OS only Check that out if it works for you.
There's an open source 1 handy from CJ Pie who is on an earlier
(13:49):
episode and back on our episode 50.
Support them, check it out and really just see if it will work
for you because even if nothing else, if you're just able to at
the start of your day do a big brain dump into an Obsidian dock
into any markdown file I guess and just say what is on your
mind. And then being able to take that
output and have an AI structure it, maybe prioritize it.
(14:10):
I know Ty. Back in the day made a tool AI
prioritize when you can just do that exact workflow to help
structure a day to give you a bit more direction.
It's really going to help unlockdifferent aspects of
organization and productivity, but you need to see what works
for you. But the most important thing is
just trying on top of voice transcription of me just
speaking at my computer, I also leverage granola during meeting
(14:31):
notes. So when I have a meeting I have
granola running, it'll transcribe and then make notes
based on that while still allowing me to have my own
notes. I like granola.
Also not perfect. My current workflow is having
granola and Zapier work togetherbecause I want my meeting notes
directly into my Obsidian for some reason.
There's no single export to markdown or export to local
directory function in granola. But I made a Zapier flow where
(14:52):
it takes my notes, does some manipulation to the title, saves
it to Google Drive. Google Drive is installed on my
computer SIM link to a city involved.
It's an unnecessarily complicated workflow just for
taking meeting notes, pulling them over.
Which is why I'm not totally bullish on granola.
But the concept is there. Being able to have AI actively
do a task for you so you don't have to be worried about meeting
(15:13):
notes. You can be fully engaged in the
conversation. But it will augment the notes
that you do take rather than just record a video of a call
that's kind of questionable. This is just transcription.
So it works a lot smoother and it doesn't have the same privacy
concerns. But like I said, granola is not
perfect, but it's something you use now and then for hardware.
So I'm not going to go too much into hardware, but one thing I
(15:34):
do want to give a shout out to is my moon Lander keyboard.
It's fully programmable, fully customizable.
So I have workflows triggered bydifferent keys.
You can have different layouts depending on what app is open.
Or if I want to do things like music control or window control,
I can do all that. And if you're on a Mac like I
am, you can pair it with Hammer Spoon.
Hammer Spoon is a free and open source app and it exposes the
(15:55):
Apple API's and it makes it so powerful.
You can do windows control, workflow automations, custom
interfaces, so much more. I'm a huge fan of Hammer Spoon,
being able to take either hotkeys or different types of
keyboard shortcuts and then turnthem into automations and with
my moon Lander keyboard and be able to very rapidly trigger it.
My other hardware device that I want to give a shout out to is
(16:16):
the Activity Dice. So it's a new device I've been
using and it really is a nice, cool tactile way of tracking my
time. I have not had success with
software trackers, but this is agood looking device that just
sits on my desk and I'm able to just easily rotate it to trigger
tracking another activity. And this is just another data
point that comes into my digitalbrain.
(16:36):
So I'm able to start seeing how am I spending my time?
Am I spending too much time making two of these videos
versus building up the AI Academy versus making tooling
itself so I'm able to get the data to make a decision rather
than just go off of vibes? And the more data you can
collect, this should be your data hoarding period.
Store the data you can in a way that's somewhat structured.
And as your tooling suite gets built out more, you can actually
(16:59):
start leveraging the benefit of that data.
We are accelerating towards a future where the way that we
interface with compute is going to be entirely different.
The things I touched on today are just intermediary steps.
This isn't the end goal. Some people predict a future
where there'll be no screens. Other people say, you know, a
slight little voice input. Maybe the glasses form factor is
going to be all that you need. We don't know how this is going
(17:20):
to play out, but what we do knowis right now there are better
ways to capture your input and interact with your computer than
the ways of old. You don't want to be typing at
110th the speed you can be speaking, if that's an option
for you. Maybe there's a webcam feature
that can do posture checks or other things to make sure that
not only are you maximizing yourwork, but you're maximizing your
(17:41):
health so that you're able to work at an optimum pace.
Anyways, there are so many new and novel ways for us to
interact with the computer that the best thing you can do, and
I'll say it a million times in this video, I imagine, is
explore, experiment, and play. See what's out there, see what
recognizes your accent the best,and just try to implement things
that work best for you. A lot of automations need to
(18:01):
leverage deterministic code rather than AI.
The reason being is it's faster,it's cheaper, it's more
efficient, and you're going to get more predictable results.
But that doesn't mean you have to learn how to be a coder to
benefit from this. One of the reasons I love
promoting cursor and cloud code is it reduces the bearer to
people to be able to create their own tools.
This includes automations, scripts, workflows.
(18:23):
Anything that can be done in code should probably be done in
code. There is the benefit to having
LMS be a fuzzy interface when you can get unstructured input
and then structure it for processing.
Or maybe you need to customize the output for a given user or
situation. So having LLMS at the ends of
these pipelines makes a lot of sense.
But throughout the central process of it, the main happy
(18:43):
path, you want to minimize the use of LLMS because then you'll
get a more predictable output and the other benefits that I
mentioned. So for a lot of the scripts
automations I made, I actually just leverage AI coding.
So I'll briefly touch on the type of tools that I use in
order to generate these scripts.I do a pretty even split in my
AI coding assistance between Cloud Code and Cursor these
(19:04):
days. So for Cloud Code, you have to
see the deep dive with Ray and Eric.
It was an absolute master class on how to use it.
They did such a phenomenal job. So I'm not going to get into too
many details here as to how I use it, but I tend to use it
when I want to do a fire and forget where I give it.
A task that I can come back to and review later, or anything
that operates in my Obsidian Vault, I'll leverage Cloud Code
(19:24):
for, and that's been really valuable for me.
Cursor is still a big part of myworkflow because I do review all
the AI generated code. Just last week a buddy caught AI
trying to run a command with an incorrect filter.
That would cause massive damage to a system.
And I don't think AI will intentionally try to nuke your
work, but mistakes do happen. You have to review your code.
So I don't tend to use cloud code as often with production
(19:47):
systems. Not because I don't think it's
capable. I'll still use Opus 4.1 in
Cursor, but I view the interfaceof dealing with the terminal
that Cloud produces to be easierfor tasks that I don't have to
babysit it for. With Cursor, you still get the
central pane, the main editing window as my review board if you
will, where I can actually look at the changes, see if it enters
(20:08):
any new variables, if the file structure was changed.
And that's just easier for me toconsume than is with Cloud Code
1. Service I've been exploring a
little bit lately is AMP. I really need to do a deep dive
on it. Actually, if you want to see
that, ping Ryan Carson, tag him on Twitter, ask if he'll come
back on to teach us because he'sthe go to guy for AMP right now.
I think it has a lot of potential.
I like that it's LM agnostic andthey're not incentivized to
(20:30):
either increase or decrease token usage, which you could
argue Cursor and Claude have as a vulnerability to their true
performance. Maybe Claude wants to increase
the token usage with Ultra, think maybe Cursor wants to
decrease the token usage to helptheir margins.
But with AMP there's no financial incentive unless you
want to go with a business plan.But for regular people, you can
just install it and use it, and it's strictly API use and
(20:52):
they're incentivized to optimizethe performance of it.
So I think there's potential there.
Still haven't explored it, but something I want to look into a
bit more. Since I'm pushing people towards
the terminal, I do want to make sure everyone has the latest and
greatest Unix utilities. So these are a dozen things that
I've come across throughout my experimentation with all this
terminal workflows that I've found made a meaningful impact
to my productivity. And I'll link everything down
(21:14):
below. I'll just speed run them right
now, but check them out and see if you will benefit from any of
these. I think you will.
A lot of these are direct replacements for built in
utilities, but so much more performant and this will just
help optimize your workflow. So a 2IN is a replacement for
History. It's a great service.
I've been a fan of it for a while.
That replaces CAT. ESA is a replacement for LS.
(21:37):
FZF is a fuzzy search. JQ lets you operate on Jason and
that gets really valuable when you start transmitting data back
and forth. RIP grip is a faster version of
grip. Starship is a shell prompt
customizer. You got to make your terminal
pretty. TLDR is like man pages but has
examples so this will help you learn from examples for
commands. T mux is a terminal multiplexer.
(21:58):
This just allows you to have multiple panes open and you can
go back to them later. This is really helpful if you're
connecting to external systems. Might not be need for everyone
but I want to make sure you knewabout it.
WTF Back from the open interpreter days.
Corrects errors and console commands and we'll rerun them
correctly. Zoxide replaces CD.
That's actually really cool because you can just simply
press Z and then the name of thedirectory without having to go
(22:18):
through all the pass to teleportthere.
Makes it a lot faster and delta is an improved get DIF.
I'm sure there are tons. Please let me know if you use
any others. I love experimenting with these,
but these are a dozen that I think everyone would benefit
from. Now I know not everyone wants to
use the terminal, especially notfor everything.
There are some workflows where transitioning from GUI to
terminal make a lot of sense, soI wanted to make sure that you
(22:40):
knew about this one workflow that I find great.
So Mac OS has this built in application called Automator.
It's for automating things. Get used to it.
It's actually pretty cool. One cool thing is you can wrap
scripts like bash scripts or Python scripts in an app that
starts making things a lot more powerful.
I also learned through talking with Claude that you can drag
applications into the menu bar of Finder.
(23:02):
So my workflow was I'd be offering in Finder doing
something, something in the GUI,and I want to take that exact
directory and open it in the terminal.
So then I can start processing it or doing more complex things
directly in the terminal. So I created open in Alacrity.
Alacrity is my terminal terminalemulator of choice.
You can use Ghosty, You can use that term.
True. There's lots of options.
So what I wanted was when I'm ina certain directory in Finder, I
(23:24):
want to one click and open in the terminal already in that
directory. So I used a cursor to generate
the code that would do this. I use Automator trapped in
application. And then I learned, like I said,
you can drag it into the menu bar.
So now I have this one button that you can click and it'll
open up the directory directly in the terminal.
It's a small little thing, but it actually saved me a few
(23:46):
seconds every time I had to do it, and it was such a frequent
thing. And now, even though this itself
might have probably saved the exact amount of time it took to
make, I have the knowledge that I'm actually able to start
adding UI elements to transitionto my ideal workflow so I don't
have to always move things over in a certain way.
I have the setup, and now I knowhow to use Automator better.
(24:06):
I know what's capable through the GUI, and all you have to do
is start experimenting and thinkhypothetically, how could I make
this better? So this has been one of those
aspects to a real world use caseof bridging the gap between the
GUI and the terminal that AI hasenabled for me, and I hope it
enables stuff for you as well. One other coding tool that I
really like to use is MCP. But in order to get real
(24:27):
benefits from that, you have to give it access to your real data
and systems and that can be kindof scary.
So that's why I've been testing out Tool Hive.
Tool Hive makes it simple and secure to use MCP.
It includes a registry of trusted MCP servers.
It lets me containerize any other server with a single
command. I can install it in a client in
seconds and secret protection and network isolation are built
in. You can try Tool Live 2.
(24:48):
It's free and it's open service.You can learn more at Tool Live
dot dev the back to the productivity hacks.
While I always prefer to build my own scripts and automations
and workflows, tools are a wholedifferent beast because there
are so many brilliant people building brilliant tools that
you don't have to build everything yourself.
You can take something off the shelf and see if it works for
you. So I wanted to go over a few
tools that I use in areas that Ihave no expertise in or areas
(25:11):
that I have no desire to build in, but that I get actual
benefit from. And I also want to talk about a
couple of tools or more like directions, which haven't quite
met my needs yet, but I really hope to see more innovation in
the space. So to cover a few media tools,
the first one that I make my thumbnails with and some other
design work for is Canva. I am artistically challenged
(25:32):
through and through. I'm no good at it and it's OK,
you don't have to be. And Canva makes it pretty easy.
It is a good way for me to be able to get a template for a
thumbnail and kind of iterate onit, make a few different copies,
share with people. I've been happy with it.
They have a few built in AI features like background removal
which makes adding the person tothe thumbnail easy.
And admittedly I'm not a designer, I'm probably missing
(25:53):
out a bunch of features, but it does make things relatively easy
for me to use and I've been veryhappy with it.
The other graphics work I do is on Ding Board.
If you are on Twitter, you're probably familiar with Ding
Board and it's how I make memes or anytime I want to make more
fun images. What's really cool with Ding
Board is most of the AI processing, if not all, runs
directly in your browser. So you fire up an instance and
(26:14):
then you can drag and drop in different images.
You can cut, you can tweak, you can manipulate it, squish it all
together and then download the output and that works really
well. For quick things that don't have
to be totally polished, you can like clean up the edges a little
bit, but being able to put some together relatively quickly, I
don't think it can be beat. It's just so much fun to be able
to generate these fun little images through Ding board.
(26:35):
For more serious AI image generation, I use Ideogram.
I don't know if it's the best. I've been happy with it.
It's one of those things where because I don't have the optimal
eye for making great images, it seems better than most.
Maybe there's something better. I know Nano Banana just dropped
and that's probably going to take the supreme seat for image
generation, but I've been happy with ID Gram.
(26:57):
I don't know if I recommend it just because I'm not familiar
enough with the landscape, but I've had no issues at all.
For music generation, I use Suno, including the music that's
in these episodes. If you remember the episode with
Young Flo, he compared the different AI music generators
and I think Suno came out on top.
There's probably different preferences for different
situations which would cause youto choose different ones, but
(27:18):
I've never had an issue with Suno.
It was the tool that I got my dad into AI stuff so I always
have a soft for it. Anytime there's a family
occasion someone will make a song and just adds a little
flair to it. I'm starting to hear more of the
very clearly AI generated sound and for some situation that
doesn't matter, lyrics really show the AI aspect to it.
For instrumentals I find it can blend in more compared to real
(27:41):
music or should I say human generated music.
The King undertake. But I've been happy with Suno.
I would say check it out. And finally, for my video
editing, I use D script. I have mixed feelings about
Dscript. The best feature by far is that
it'll automatically transcribe the videos you drag into the
project and then you get edited like a word doc.
I'm so happy with that. They're features like eye
(28:02):
contact and whatnot I think are terrible.
I think they're putting too mucheffort into niche unpolished AI
features just to say they have these AI features, but the core
offering is really good. I just wish they improve
stability. I can't quite recommend Descript
even though I really want to because it's been helpful to me
as someone with no video editingexperience to be able to produce
over a year of podcast episodes.But they just don't quite get
(28:25):
the level of Polish that I feel you need to replace a more
seasoned experience. Video editing software.
But if you're looking to quicklyput together podcast episodes or
demo videos, things where it's for video editing is outside
your area of expertise. Dscripts work quite well for me.
One area that a lot of people talk about, and I don't have a
suggestion, there's no tool that's stuck in my workflow yet
(28:47):
is browser use. So I'm sure you heard of Comet
by perplexity seemed to have a lot of promise at first, but
then Brave pointed out that there's a lot of vulnerabilities
with it. We had people from Strawberry
browser on before as well as plug insurance like Retriever,
and there's promise there. We want these agents to be able
to operate on the web on our behalf, but we're still early
and they make mistakes and we don't have the same control to
(29:09):
rollback or undo changes on the open web rather than on our our
local machines. So I personally don't trust a
lot of these agents to operate on my e-mail.
There can be things like Superhuman that'll help organize
your e-mail. I used Shortwave before, which
is like a Gmail wrapper that helps you draft emails without
actually send it. Like it'll get you part of the
way there. But I found there's very few
(29:31):
instances where I don't want to write my own emails because
writing is akin to thinking. And it's very important that we
don't let AI take away too much of our writing.
If it's a generic thing that I'll send over and over, sure, I
can let AI take a template and run with it, do a little
customization. But for the most part, if I'm
writing something to someone, I want it to be from me.
So AI can't quite succeed there in my workflow yet for people
who want to do things like booking a flight or all those
(29:53):
classic demos, that really mightsave a bit of time.
I am not comfortable with the AImaking decisions that involve
like transactions. I don't want it purchasing
anything on my behalf. We've run experiments before and
it's just, it's not great yet. There's no promise.
I'm still optimistic, still bullish, but I don't currently
use anything to operate the web on my behalf.
There are however tools like browser base or Exa or fire
(30:18):
crawl which will actually do webbehaviors behind the scenes like
searching, crawling, retrieving information and that I'm all for
pull in as much as you can. Let's automate it, let's get it
so it's in just for our systems.But to actually do the actions.
I haven't found one that I like yet, so if you know any let me
know. I'd love to try out more, but
for now web browsing is just an input to my more robust, mostly
(30:43):
deterministic systems rather than let it run off on its own.
One caveat I will say, if you leverage a system like E2B or
you're able to dockerize environments for to have the AI
operate in, that can be safer. But again, it's not necessarily
what you're worried about on on your machine.
It's it's the web and the One Direction aspect of it.
(31:04):
It's also a risk where people can embed hidden text in ASCII
characters or invisible text on websites, things that can lead
an agent down a wrong path that you won't see.
And the web is the Wild West. There's so much, it is pretty
much Infinity opportunity for malicious behavior on the
Internet. If you let an AI get some
credentials, if you give it yourcredit card and just release on
(31:24):
the Internet, you have to be willing to accept the
consequences of that risk. And that's just somewhere where
I'm not at yet. But being able to build a system
that can ping a URL, pull back all the data, process it, and
present me with some informationthat's helpful to my workflows,
very bullish on that. So explore those tools, but be
careful when you're using AI on the open Internet.
(31:46):
There's just a lot of risks involved with it.
So one tool that I guarantee everyone here is using, but I
want to touch on a few differentopportunities that it presents
is just a chat interface, a chatbot.
Whether you go on a ChatGPT or use Claude or whatever other LM
provider you want, the things that you do with chat can be
very impactful in your productivity.
And I want to go over two big use cases.
(32:06):
So the first is using as a thought partner when you're
starting your day, planning yourday, planning a project,
planning a, it's very easy to just ask it to give you a prompt
that you can give to a coding LLM or just go directly to the
coding LLM. But if you take a step back and
put more energy into the planning process, you're able to
come to a better result faster. And it doesn't just have to be
(32:28):
when generating an output. You can talk about strategizing
and this garries across the entire spectrum of your life.
It could be a professional or personal If you want to
strategize a weight loss plan, having a thought partner as to
what is realistic expectations for both your calorie
restrictions, your exercise plan, your dietary habits.
Being able to bounce it off of an entity to get extra insight
(32:50):
is going to be valuable. If you're playing out a script,
thinking about different perks to your operating system or
different benefits to using one language over another, it might
be a dead end, but it might actually expose something to
help you think about things moredeeply.
So I encourage people who eitherare in my AI Academy or just
hire me for coaching is just have the little chat interface
(33:12):
open at all times. Go over, ask a little question.
If it's super basic, ask Jan, ask your favorite local model,
and then it can just bounce my ideas off you.
If it's something more complex and you need some more planning,
ask Claude. Enable thinking so we can go
deep into it. Get web search so you can get
the most updated information. But you want to get into a habit
of getting a second opinion on things.
And this isn't to say the AI should drive your decision
(33:34):
making, it absolutely should not.
Remember, LMS are probabilistic,so it is going to give you the
probably best answer. Very middle of the road, very
unopinionated in many ways. Plus whatever censorship or
guardrails that the provider puts on will definitely bias his
decision but or recommendation. But if you are driving the
decision and you're simply asking for an extra set of eyes,
(33:57):
someone to give you a bit of feedback on it, the AI can help
fill that gap. The other one is actually as AAI
therapist and I had mixed reviews if you look back on old
videos. I was very opposed to it, but
eventually I came around to seeing it as a viable tool not
to replace human therapist, but to actually enable a more open
and transparent conversation with the things bounced around
(34:19):
in your head, to get to a more comfortable place, and to mostly
establish strategies and tools that you can have to think
through things clearly. High Neum was on recently and
the first time he came on, he was talking about how going to a
therapist is understanding how you think and can give you
superpowers that way. And while an AI can't go to the
same extent, I have found that if you're willing to be
(34:41):
transparent and open with the thoughts that are going on your
head, it can help you process them.
And one big thing with productivity is can you be
focused? And if you're distracted because
of the hardships of life, because everyone deals with a
whole bunch of hard stuff, you're able to become more
present on your task and be ableto focus more deeply if you can
even give yourself temporary relief from the stresses that is
(35:03):
life. So if you're able to just brain
dump in the morning, the things that are bothering you, the
things that are concerning you, the things that you know you
have to deal with, it just gets them out of your head.
It's like writing them down, butverbally, I still encourage you
write things down. And if you can do so in a way
where that can be an input to your system, all the better.
But it's very important to get what's inside out in a safe
manner that you're comfortable with.
(35:24):
And if you're able to give some of that to an AI just to get a
little bit of feedback, it'll help you get through that
moment. It'll help you take your next
step, be a little more productive for the next period
of time until you can get eitherhelp you need, guidance you
need, or even just feedback fromother human to help you get the
right place. But sometimes you'll be lying in
bed unable to get to sleep because the stresses of the work
(35:45):
day are bouncing around. And if you just say it out loud
to an NT and have something go back being like I understand
here is some strategies to fall asleep faster, it might actually
help you get a better night's sleep, making tomorrow an even
productive day. So I would say before you Goff
at the idea of using an AI as a therapist, just have a
conversation with it the next time something's bringing you
down. You might be surprised with the
(36:05):
suggestions it has, and they canbe pretty thoughtful and
insightful and might give you more tools and insight to
understand how your brain works,what what is helping you escape
from a spiral. And then you'll be able to be
more productive just because you're in a better state of
mind. Well, that was a lot.
It also feels like it wasn't that much at the exact same
time. I can go deeper into detail on
any of these topics and I'm really curious to see what
(36:26):
resonated most with you. What do you want to see more on?
What can I double Click to explore more deeply?
Please let me know. There's a few things I do want
to bring to your attention just to make sure you're aware of it.
Things like the sycophancy or the irrational disagreeability
of LMS. If you tell them, be critical.
You can't always trust the outputs, and I encourage you to
not trust the outputs. Validate, double check.
(36:47):
Make sure there's a human in theloop.
You might want to optimize productivity and just say AI go
crazy and just full throttle ahead, no guardrails, but you're
going to end up spending more time fixing things if you go
down the wrong path or if there's mistakes along the way.
So these are fully intended on augmenting you, not replacing
you. And it goes the same with your
(37:08):
workflows. Augment your workflows, don't
replace it. If you can use deterministic
code that's reliable, it will produce the correct output more
frequently. So make sure that you use the LM
to generate code rather than just all the LLM to do a certain
tasks. The one example I have is my
thumbnail generator, The square thumbnail generator.
If you ask the AI to do every single time, it'll work 9 times
(37:29):
out of 10. But then sometimes it'll go off
the rails. It might install the wrong
library. But if you have the AI generate
the code to do it, it happens instantly every single time.
And then you can trigger that from your Obsidian vault.
So you get start getting these interconnected pathways.
There's so much more that we need to explore, but I wanted to
give you the foundation so that you understand how I've become
massively more productive with the help of these systems and
(37:50):
get inspiration. What works for me might not work
for you, but the principles remain.
Figuring out your processes, thestep by step instructions that
you go through in your head automatically or consciously,
and then being able to start offloading that to technology.
There's so much more to cover. I want to give a quick shout out
to Tool Hi for supporting the show.
(38:11):
So I can do these nerdy deep dives and teach you the things
that I've learned over the years.
But I want to go a lot deeper. So please let me know what you
want to see and then I'll see you next week.
Thank you.
The most common question people ask me these days is how I use
AI to become more productive. Those real roductivity gains,
not the demo where the things that give me hours back every
single day. Well, it's time to share some
secrets. In 10 years of being a software
developer with the last three focused on AI tooling, I've
tested out hundreds of differenttools and built out some great
workflows and automations to improve my productivity.
(00:21):
On top of the dozens of tool useconversations, I've been hired
as a productivity coach and I'vebuilt out the AI Academy of Box
1 Ventures, which has upscaled the entire team on improving the
productivity. So in episode 55 of Tool Use,
brought to you by Tool Live, I'mgoing to teach you how I've
levelled up my productivity withthe use of AI.
These workflows will give you hours of your life back.
The automations are going to allow you to focus on what
(00:43):
really matters. I'll be going over my digital
brain set, how I interface with my computer, how I build these
Hypercustom automations and a bunch of my tools.
Let's get into it. So obviously you can't have a
conversation about leveraging AIto become more productive until
you start talking about which models you use, which LLM
providers get you the best gains.
It's the one thing people go back and forth with all the
(01:03):
time. So for me, the model that I use
the most is all of them, which Iknow is kind of a non answer,
but there's a good reason for it.
You don't want to get locked into building on a specific
platform because once you get locked in, it makes it a lot
harder to shift. And as these models start
changing the way that prompts affect them and changing the
ways interface with tooling, there will be a divergent where
we're not going to have the samelevel of systems being model
(01:25):
agnostic as we used to. So it's going to be very
important that you start gettingused to the way that they act in
different situations form that intuition, start realizing which
models work better in some situations because not only do
new models come out, that's going to keep changing, but the
providers of the models actuallywill change the amount of
compute that they give towards resources.
They might nerf them at certain times or change the way that
(01:46):
their test time compute works. So it's very important to just
get a general intuition. When you give an input, what
output do you expect? How does it change across
providers? So if you get locked into using
one single provider for everything that you build,
you're going to lose that flexibility and observe how
different models were. So it's important to think of it
less as choosing a team that you're on and more of aligning
(02:08):
it with in your real life. Who would you go to for advice?
Depends on the topic. Maybe it's a friend or mom and
dad or all of the above. And you really do not have a
static answer for who's the bestperson to go for for advice.
And now we have access to intelligence that's just an API
Callaway, a chat message away. And this is something that's
brand new to us. We don't know how to best use
(02:29):
it. We're still figuring it out.
And everyone has different preferences and different
methods of interacting with it. So there is no best answer.
And the spectrum of possibilities changes every
single day. And even though that might feel
over whelming, reflecting on it really shows how incredibly
miraculous it is that we've beenable to turn math and
electricity into intelligence and we want everyone to be able
(02:49):
to benefit from it. And there is no silver bullet.
So the best of next to have is use it, play with it, explore,
experiment, immerse yourself in it and see what happens.
So my general setup is I have Clawed desktop and Jan running
on my computer. Clawed is for the big things,
Jan is for the easy questions. I personally think it's worth
the extra effort to direct your questions to the appropriate
model because do you really needthat massive model to get a
(03:11):
definition or translation? Local models use way less
electricity and for the rest of the LMS I use web interfaces.
I don't really have a reason, it's just kind of fall into my
habit. Like I'll still use Open AI,
Grok, Gemini, they all are in the regular routine of models I
check in on. Sometimes I use Deepseeker
Hermes, but the open source onesI tend to use locally and I
really don't have justification for this yet.
(03:32):
It's just by falling into. I do try to stay on top of
testing new things so that'll probably change as time goes on.
I encourage you to play as well.I will say though, one big piece
of advice I have is to set up a system so that you can easily
switch between these models and test them out in comparison.
Create your eval suite, create your AI service.
Check out my GitHub for some examples of those.
But if you're able to take control over the experimentation
(03:54):
process to actually efficiently see when a new model comes out,
how it works with your workflow,how it compares to the other
ones, you're going to be more rapidly able to adapt these
changes. So don't just think about which
models you use, think about which systems you need to set up
to compare the models. So at the snapshot in time for
your specific use case, you can easily discover what the best
(04:15):
model is. But if I had to choose, I'd
probably say Claude. The next thing I wanted to touch
on is the aspect of my digital brain.
So you might have heard people talk about the second brain,
which will be a system setup to store and retrieve their notes.
So they're able to get everything consolidated in a
digital way so they can easily search and retrieve things.
On the age of AI, we can take itso much further.
So I've really encourage everyone, and you've probably
(04:36):
heard me say this 100 times before in the show to try out
Obsidian. Obsidian has many perks and
especially the age of AI, it becomes very powerful.
Before I get into the specifics of how I use Obsidian to create
a digital brain, I just want to go over 3 principles.
Why I've chosen Obsidian over all the rest of the note taking
apps. So the first is the file over
app principle. What this means is you control
(04:57):
your data. The A is simply an interface for
you to interact with that data. But if you didn't like Obsidian,
you could get rid of it today and still keep all of your
information and easily port it over to a new system.
There's some tools like Notion which become quite popular,
especially with their AI integration, but it's very
limited. And if you try to export your
Notion data, it is a mess. Good luck importing anywhere
(05:17):
else. They really try to keep you
locked in. And like I mentioned earlier,
you don't want to be locked. In this day and age, everything
moves so fast and evolves so aggressively that you want that
flexibility of controlling your data and being able to take with
you wherever you want. The next one is the ease of AI
to be able to read Markdown. Markdown is nicely structured, a
very simple syntax, so the learning curve is very small and
(05:38):
it just allows things to be structured in a way that AI can
easily understand. There is benefits to using
things like XML tags and whatnot, but those are so hyper
specific that they really works better for prompts rather than
structuring your data. And the third aspect of Obsidian
I love is the plugins. Plugins give you so much
capability. There is templates that you can
go on GitHub to download from, from the Obsidian team to build
(05:59):
your own. I developed the Obsidian plugin
generator. We can have a simple chat with
AI to develop your own plugin, but these plugins allow
functionality not just within the vault, but within your
computer as a whole. And it exposes different APIs so
you can really easily interact with things.
And I got a few exciting ones toshow you.
The first thing that I want to touch on with my Obsidian setup
is the dashboard. So I have a dashboard that loads
whenever I open Obsidian and it dynamically generates the things
(06:21):
that I need to see most. So we have tasks, what's today's
task? What's overdue, what are
priorities. This is dynamically generated
using things like data view plugin and tasks plug in, which I'll
link both to. So by leveraging these, I'm able
to create tasks in my daily notes, which I use with the
calendar plug in and have a collection of of tasks that I
(06:44):
get from meetings or that I meetthroughout the day.
And then I have it centralized in a specific dashboard so I can
quickly see at the start of my day or the start of a workout
what I need to prioritize most. Also, because I use my vault for
personal as well as professionalstuff, I have my health and
fitness pages linked to it, so Ican easily see where I'm at.
My rowing program, I have a section for the next two years
episode. We'll link to the notes and
(07:05):
that's dynamically generated based on what's coming up next.
I have stats system, health important links.
I have things that are developedfor me.
And The thing is, with these dashboards, you can make it
completely tailored to your use cases.
What things do you want to see? What would help you prioritize
your day? By having a dynamically
generated dashboard, you're ableto really quickly get a
(07:26):
glanceable view at what you haveto do and then be able to feed
those inputs like checking off atask or adding something.
You do a list, it'll feed through your system to where it
needs to be. 1 community plug inthat I really wanted to
emphasize is data view. I think data view is so powerful
it might be a little intimidating because you are
writing code. It supports both its own unique
query language as well as Java, JavaScript, it just which makes
(07:48):
it so powerful. If you struggle with it, you can
have clawed code running in yourObsidian Vault, looking at the
different dashboards and updating it for you.
So you don't have to necessarilydive too deep into the docs,
Although getting an idea of how it works will be helpful.
So I still encourage you to. But clawed code will be able to
help get you most of the way there.
And then you can literally go back and forth to see what
(08:09):
works. Maybe you want to pull tasks
just from your daily notes, or maybe you also want to include
it from your meetings, your people, your company lists.
All of that is possible with data.
And then on top of these community plugins that I
recommend, I actually have developed more custom plugins
than what I found on their community listing.
And that's because of my Obsidian plugin generator.
I can really quickly spin these things up.
(08:29):
So I use my Obsidian Vault as a CRM.
Every meeting I have, every person that I meet, every
company they work for that is stored in my Obsidian Vault and
it's linked to the different meetings and daily notes and
whatnot because it's very valuable for me to be able to
quickly go through the flow of information and with it can
follow these links. So I have one set up to do
(08:51):
automatic meeting stuff, whetherit's pulling in data from
granola or generating to do's that we can follow up with later
or even being able to feed out into e-mail systems.
All that is hosted in meetings. And then of course with tool
use, I have some podcast automation.
So every time there's a new episode, it'll generate the new
directory along with the required documents and it'll get
that set up. I've taken it one step further
(09:11):
to when I finish editing and I export the transcript, the
podcast automations are actuallyable to generate AI suggestions
for titles, which are always terrible.
We're not getting good at titlesunfortunately.
These two descriptions decent, but even things like
manipulating the thumbnail so I can take it from the regular 16
by 9 thumbnail and make it square for Spotify or other
(09:34):
podcast show art. My Obsidian is able to do that.
So you're not necessarily stuck with doing automations within
Obsidian. You can get system processes or
interact on files and systems outside of your app outside of
Obsidian to make yourself more automated.
So that that's why I'm considering more of a digital
brain than a second brain because you're able to induce
actions. It's not just storing and
(09:56):
retrieving memories. You're able to actually act upon
it, get intelligence from it, doa whole lot of things.
So it's phenomenal. And one thing I definitely want
to give a shout out to Steph Ango or also know online is Pano
is the creator of Obsidian and he has a phenomenal writing.
Check out his website so you canread more.
There's a lot of great things, but he has some principles that
he talks about with the way he sets up his Obsidian vault.
(10:16):
And I really like a lot of them.So he likes to use links and
categories and tags instead of nested directories.
So you're able to really deeply link different files to each
other with Obsidian with wiki link style.
So you can just simply make a reference to a person and link
to the person's page so you're able to more rapidly get context
on that person with nested directories.
It's nice for a simple UI to click around, but you can
(10:38):
rapidly, with Command O on Mac, open up the notes that you're
looking for. So you don't necessarily need to
use the UI for everything where your fingers are already on the
keyboard. You can just learn a few
keyboard shortcuts and rapidly switch from the notes that you
need to be focused on. Another concept he mentioned,
this fractal journey and random revisiting, which I think is
great. So you can really quickly spin
up a partial note on Obsidian and then pull them together.
(11:00):
So you can say all the notes made on this day can go into a
centralized location and then later on I can review those
notes with his random revisitingand then figure out where these
things actually belong to live. So you don't have to be
constantly going through your vault to figure out what goes
where and where information should live.
You can randomly visit it later at a better time for you and
just simply move things to the right location.
(11:21):
Properties and templates to set up new nodes.
I have a few templates set up. I definitely want to get more
but it makes it easy to pre populate different bits of
information and later on as you gather more information, say you
have a meeting with someone in abrand new company.
So you just simply have the company name, quick description
of what they do later on, you find out their website or their
LinkedIn page, all their other socials.
You're able to add that into thetemplate later on.
(11:42):
So as I said, check out his website to get more of his
writing. It's excellent.
He does a lot of design stuff aswell, which I think is
phenomenal and it really reflects in how powerful,
flexible and enjoyable it is to use Obsidian.
The way you interact with a computer might be the biggest
opportunity for improving your productivity.
We've been using computers the same way for ages, keyboard,
mouse, typing, and it's really not optimal for bandwidth. 10
(12:05):
fingers cannot not beat speakingspeed for an example.
So what you'll want to do is reflect on the way that you use
a computer, figure out what tools exist out there that might
better align your optimal workflow with the way you
interact with the computer, and then just start exploring and
playing with it. Because we don't have to be
locked into these old paradigms.There is so much more
opportunity for us to explore now.
(12:26):
So I'm going to go over a few the ways that I interface with
my computer, both software and hardware, and then just give you
a little inspiration, things youcan explore because it might
work for you, it might not. Everyone's is different, but the
purpose is to make sure that youexplore the opportunities to
find out what works best for you.
So by far the biggest unlock that I've had has been voice
transcription. It's made such a huge difference
(12:46):
in my productivity. I can speak 10 times faster than
I can type. When I get excited, it's
probably like 20 times faster. And these modern transcription
tools are just so good. You can add custom vocabulary,
you can have corrections. I even have it where I say
insert my YouTube channel and it'll put the URL instead of
just verbatim transcribing it. And now I know that voice
transcription is not appropriatefor every situation.
(13:06):
If you work in an open space, open concept office, it's not
going to be great for you just rambling at your computer all
day. But you should try to leverage
it in the situations that do work for you.
Maybe you go for a walk and you just like take notes as you go.
Or you can book a meeting room. When you have this big piece of
work you have to get through with speaking, you take
advantage of different neural pathways than when you're
typing. It really gives you a different
output. And if you start prompting an
(13:27):
LLM with speaking, sometimes typing, you can start seeing how
you approach it differently and how you interact with it
differently. And I think it can be massively
beneficial. So right now I use Aqua voice.
It's good. I like it.
I don't think I'll use it forever.
It has had a few flaws, but for the most part it's good.
I used to use super whisper alsogood.
It's Mac OS only Check that out if it works for you.
There's an open source 1 handy from CJ Pie who is on an earlier
(13:49):
episode and back on our episode 50.
Support them, check it out and really just see if it will work
for you because even if nothing else, if you're just able to at
the start of your day do a big brain dump into an Obsidian dock
into any markdown file I guess and just say what is on your
mind. And then being able to take that
output and have an AI structure it, maybe prioritize it.
(14:10):
I know Ty. Back in the day made a tool AI
prioritize when you can just do that exact workflow to help
structure a day to give you a bit more direction.
It's really going to help unlockdifferent aspects of
organization and productivity, but you need to see what works
for you. But the most important thing is
just trying on top of voice transcription of me just
speaking at my computer, I also leverage granola during meeting
(14:31):
notes. So when I have a meeting I have
granola running, it'll transcribe and then make notes
based on that while still allowing me to have my own
notes. I like granola.
Also not perfect. My current workflow is having
granola and Zapier work togetherbecause I want my meeting notes
directly into my Obsidian for some reason.
There's no single export to markdown or export to local
directory function in granola. But I made a Zapier flow where
(14:52):
it takes my notes, does some manipulation to the title, saves
it to Google Drive. Google Drive is installed on my
computer SIM link to a city involved.
It's an unnecessarily complicated workflow just for
taking meeting notes, pulling them over.
Which is why I'm not totally bullish on granola.
But the concept is there. Being able to have AI actively
do a task for you so you don't have to be worried about meeting
(15:13):
notes. You can be fully engaged in the
conversation. But it will augment the notes
that you do take rather than just record a video of a call
that's kind of questionable. This is just transcription.
So it works a lot smoother and it doesn't have the same privacy
concerns. But like I said, granola is not
perfect, but it's something you use now and then for hardware.
So I'm not going to go too much into hardware, but one thing I
(15:34):
do want to give a shout out to is my moon Lander keyboard.
It's fully programmable, fully customizable.
So I have workflows triggered bydifferent keys.
You can have different layouts depending on what app is open.
Or if I want to do things like music control or window control,
I can do all that. And if you're on a Mac like I
am, you can pair it with Hammer Spoon.
Hammer Spoon is a free and open source app and it exposes the
(15:55):
Apple API's and it makes it so powerful.
You can do windows control, workflow automations, custom
interfaces, so much more. I'm a huge fan of Hammer Spoon,
being able to take either hotkeys or different types of
keyboard shortcuts and then turnthem into automations and with
my moon Lander keyboard and be able to very rapidly trigger it.
My other hardware device that I want to give a shout out to is
(16:16):
the Activity Dice. So it's a new device I've been
using and it really is a nice, cool tactile way of tracking my
time. I have not had success with
software trackers, but this is agood looking device that just
sits on my desk and I'm able to just easily rotate it to trigger
tracking another activity. And this is just another data
point that comes into my digitalbrain.
(16:36):
So I'm able to start seeing how am I spending my time?
Am I spending too much time making two of these videos
versus building up the AI Academy versus making tooling
itself so I'm able to get the data to make a decision rather
than just go off of vibes? And the more data you can
collect, this should be your data hoarding period.
Store the data you can in a way that's somewhat structured.
And as your tooling suite gets built out more, you can actually
(16:59):
start leveraging the benefit of that data.
We are accelerating towards a future where the way that we
interface with compute is going to be entirely different.
The things I touched on today are just intermediary steps.
This isn't the end goal. Some people predict a future
where there'll be no screens. Other people say, you know, a
slight little voice input. Maybe the glasses form factor is
going to be all that you need. We don't know how this is going
(17:20):
to play out, but what we do knowis right now there are better
ways to capture your input and interact with your computer than
the ways of old. You don't want to be typing at
110th the speed you can be speaking, if that's an option
for you. Maybe there's a webcam feature
that can do posture checks or other things to make sure that
not only are you maximizing yourwork, but you're maximizing your
(17:41):
health so that you're able to work at an optimum pace.
Anyways, there are so many new and novel ways for us to
interact with the computer that the best thing you can do, and
I'll say it a million times in this video, I imagine, is
explore, experiment, and play. See what's out there, see what
recognizes your accent the best,and just try to implement things
that work best for you. A lot of automations need to
(18:01):
leverage deterministic code rather than AI.
The reason being is it's faster,it's cheaper, it's more
efficient, and you're going to get more predictable results.
But that doesn't mean you have to learn how to be a coder to
benefit from this. One of the reasons I love
promoting cursor and cloud code is it reduces the bearer to
people to be able to create their own tools.
This includes automations, scripts, workflows.
(18:23):
Anything that can be done in code should probably be done in
code. There is the benefit to having
LMS be a fuzzy interface when you can get unstructured input
and then structure it for processing.
Or maybe you need to customize the output for a given user or
situation. So having LLMS at the ends of
these pipelines makes a lot of sense.
But throughout the central process of it, the main happy
(18:43):
path, you want to minimize the use of LLMS because then you'll
get a more predictable output and the other benefits that I
mentioned. So for a lot of the scripts
automations I made, I actually just leverage AI coding.
So I'll briefly touch on the type of tools that I use in
order to generate these scripts.I do a pretty even split in my
AI coding assistance between Cloud Code and Cursor these
(19:04):
days. So for Cloud Code, you have to
see the deep dive with Ray and Eric.
It was an absolute master class on how to use it.
They did such a phenomenal job. So I'm not going to get into too
many details here as to how I use it, but I tend to use it
when I want to do a fire and forget where I give it.
A task that I can come back to and review later, or anything
that operates in my Obsidian Vault, I'll leverage Cloud Code
(19:24):
for, and that's been really valuable for me.
Cursor is still a big part of myworkflow because I do review all
the AI generated code. Just last week a buddy caught AI
trying to run a command with an incorrect filter.
That would cause massive damage to a system.
And I don't think AI will intentionally try to nuke your
work, but mistakes do happen. You have to review your code.
So I don't tend to use cloud code as often with production
(19:47):
systems. Not because I don't think it's
capable. I'll still use Opus 4.1 in
Cursor, but I view the interfaceof dealing with the terminal
that Cloud produces to be easierfor tasks that I don't have to
babysit it for. With Cursor, you still get the
central pane, the main editing window as my review board if you
will, where I can actually look at the changes, see if it enters
(20:08):
any new variables, if the file structure was changed.
And that's just easier for me toconsume than is with Cloud Code
1. Service I've been exploring a
little bit lately is AMP. I really need to do a deep dive
on it. Actually, if you want to see
that, ping Ryan Carson, tag him on Twitter, ask if he'll come
back on to teach us because he'sthe go to guy for AMP right now.
I think it has a lot of potential.
I like that it's LM agnostic andthey're not incentivized to
(20:30):
either increase or decrease token usage, which you could
argue Cursor and Claude have as a vulnerability to their true
performance. Maybe Claude wants to increase
the token usage with Ultra, think maybe Cursor wants to
decrease the token usage to helptheir margins.
But with AMP there's no financial incentive unless you
want to go with a business plan.But for regular people, you can
just install it and use it, and it's strictly API use and
(20:52):
they're incentivized to optimizethe performance of it.
So I think there's potential there.
Still haven't explored it, but something I want to look into a
bit more. Since I'm pushing people towards
the terminal, I do want to make sure everyone has the latest and
greatest Unix utilities. So these are a dozen things that
I've come across throughout my experimentation with all this
terminal workflows that I've found made a meaningful impact
to my productivity. And I'll link everything down
(21:14):
below. I'll just speed run them right
now, but check them out and see if you will benefit from any of
these. I think you will.
A lot of these are direct replacements for built in
utilities, but so much more performant and this will just
help optimize your workflow. So a 2IN is a replacement for
History. It's a great service.
I've been a fan of it for a while.
That replaces CAT. ESA is a replacement for LS.
(21:37):
FZF is a fuzzy search. JQ lets you operate on Jason and
that gets really valuable when you start transmitting data back
and forth. RIP grip is a faster version of
grip. Starship is a shell prompt
customizer. You got to make your terminal
pretty. TLDR is like man pages but has
examples so this will help you learn from examples for
commands. T mux is a terminal multiplexer.
(21:58):
This just allows you to have multiple panes open and you can
go back to them later. This is really helpful if you're
connecting to external systems. Might not be need for everyone
but I want to make sure you knewabout it.
WTF Back from the open interpreter days.
Corrects errors and console commands and we'll rerun them
correctly. Zoxide replaces CD.
That's actually really cool because you can just simply
press Z and then the name of thedirectory without having to go
(22:18):
through all the pass to teleportthere.
Makes it a lot faster and delta is an improved get DIF.
I'm sure there are tons. Please let me know if you use
any others. I love experimenting with these,
but these are a dozen that I think everyone would benefit
from. Now I know not everyone wants to
use the terminal, especially notfor everything.
There are some workflows where transitioning from GUI to
terminal make a lot of sense, soI wanted to make sure that you
(22:40):
knew about this one workflow that I find great.
So Mac OS has this built in application called Automator.
It's for automating things. Get used to it.
It's actually pretty cool. One cool thing is you can wrap
scripts like bash scripts or Python scripts in an app that
starts making things a lot more powerful.
I also learned through talking with Claude that you can drag
applications into the menu bar of Finder.
(23:02):
So my workflow was I'd be offering in Finder doing
something, something in the GUI,and I want to take that exact
directory and open it in the terminal.
So then I can start processing it or doing more complex things
directly in the terminal. So I created open in Alacrity.
Alacrity is my terminal terminalemulator of choice.
You can use Ghosty, You can use that term.
True. There's lots of options.
So what I wanted was when I'm ina certain directory in Finder, I
(23:24):
want to one click and open in the terminal already in that
directory. So I used a cursor to generate
the code that would do this. I use Automator trapped in
application. And then I learned, like I said,
you can drag it into the menu bar.
So now I have this one button that you can click and it'll
open up the directory directly in the terminal.
It's a small little thing, but it actually saved me a few
(23:46):
seconds every time I had to do it, and it was such a frequent
thing. And now, even though this itself
might have probably saved the exact amount of time it took to
make, I have the knowledge that I'm actually able to start
adding UI elements to transitionto my ideal workflow so I don't
have to always move things over in a certain way.
I have the setup, and now I knowhow to use Automator better.
(24:06):
I know what's capable through the GUI, and all you have to do
is start experimenting and thinkhypothetically, how could I make
this better? So this has been one of those
aspects to a real world use caseof bridging the gap between the
GUI and the terminal that AI hasenabled for me, and I hope it
enables stuff for you as well. One other coding tool that I
really like to use is MCP. But in order to get real
(24:27):
benefits from that, you have to give it access to your real data
and systems and that can be kindof scary.
So that's why I've been testing out Tool Hive.
Tool Hive makes it simple and secure to use MCP.
It includes a registry of trusted MCP servers.
It lets me containerize any other server with a single
command. I can install it in a client in
seconds and secret protection and network isolation are built
in. You can try Tool Live 2.
(24:48):
It's free and it's open service.You can learn more at Tool Live
dot dev the back to the productivity hacks.
While I always prefer to build my own scripts and automations
and workflows, tools are a wholedifferent beast because there
are so many brilliant people building brilliant tools that
you don't have to build everything yourself.
You can take something off the shelf and see if it works for
you. So I wanted to go over a few
tools that I use in areas that Ihave no expertise in or areas
(25:11):
that I have no desire to build in, but that I get actual
benefit from. And I also want to talk about a
couple of tools or more like directions, which haven't quite
met my needs yet, but I really hope to see more innovation in
the space. So to cover a few media tools,
the first one that I make my thumbnails with and some other
design work for is Canva. I am artistically challenged
(25:32):
through and through. I'm no good at it and it's OK,
you don't have to be. And Canva makes it pretty easy.
It is a good way for me to be able to get a template for a
thumbnail and kind of iterate onit, make a few different copies,
share with people. I've been happy with it.
They have a few built in AI features like background removal
which makes adding the person tothe thumbnail easy.
And admittedly I'm not a designer, I'm probably missing
(25:53):
out a bunch of features, but it does make things relatively easy
for me to use and I've been veryhappy with it.
The other graphics work I do is on Ding Board.
If you are on Twitter, you're probably familiar with Ding
Board and it's how I make memes or anytime I want to make more
fun images. What's really cool with Ding
Board is most of the AI processing, if not all, runs
directly in your browser. So you fire up an instance and
(26:14):
then you can drag and drop in different images.
You can cut, you can tweak, you can manipulate it, squish it all
together and then download the output and that works really
well. For quick things that don't have
to be totally polished, you can like clean up the edges a little
bit, but being able to put some together relatively quickly, I
don't think it can be beat. It's just so much fun to be able
to generate these fun little images through Ding board.
(26:35):
For more serious AI image generation, I use Ideogram.
I don't know if it's the best. I've been happy with it.
It's one of those things where because I don't have the optimal
eye for making great images, it seems better than most.
Maybe there's something better. I know Nano Banana just dropped
and that's probably going to take the supreme seat for image
generation, but I've been happy with ID Gram.
(26:57):
I don't know if I recommend it just because I'm not familiar
enough with the landscape, but I've had no issues at all.
For music generation, I use Suno, including the music that's
in these episodes. If you remember the episode with
Young Flo, he compared the different AI music generators
and I think Suno came out on top.
There's probably different preferences for different
situations which would cause youto choose different ones, but
(27:18):
I've never had an issue with Suno.
It was the tool that I got my dad into AI stuff so I always
have a soft for it. Anytime there's a family
occasion someone will make a song and just adds a little
flair to it. I'm starting to hear more of the
very clearly AI generated sound and for some situation that
doesn't matter, lyrics really show the AI aspect to it.
For instrumentals I find it can blend in more compared to real
(27:41):
music or should I say human generated music.
The King undertake. But I've been happy with Suno.
I would say check it out. And finally, for my video
editing, I use D script. I have mixed feelings about
Dscript. The best feature by far is that
it'll automatically transcribe the videos you drag into the
project and then you get edited like a word doc.
I'm so happy with that. They're features like eye
(28:02):
contact and whatnot I think are terrible.
I think they're putting too mucheffort into niche unpolished AI
features just to say they have these AI features, but the core
offering is really good. I just wish they improve
stability. I can't quite recommend Descript
even though I really want to because it's been helpful to me
as someone with no video editingexperience to be able to produce
over a year of podcast episodes.But they just don't quite get
(28:25):
the level of Polish that I feel you need to replace a more
seasoned experience. Video editing software.
But if you're looking to quicklyput together podcast episodes or
demo videos, things where it's for video editing is outside
your area of expertise. Dscripts work quite well for me.
One area that a lot of people talk about, and I don't have a
suggestion, there's no tool that's stuck in my workflow yet
(28:47):
is browser use. So I'm sure you heard of Comet
by perplexity seemed to have a lot of promise at first, but
then Brave pointed out that there's a lot of vulnerabilities
with it. We had people from Strawberry
browser on before as well as plug insurance like Retriever,
and there's promise there. We want these agents to be able
to operate on the web on our behalf, but we're still early
and they make mistakes and we don't have the same control to
(29:09):
rollback or undo changes on the open web rather than on our our
local machines. So I personally don't trust a
lot of these agents to operate on my e-mail.
There can be things like Superhuman that'll help organize
your e-mail. I used Shortwave before, which
is like a Gmail wrapper that helps you draft emails without
actually send it. Like it'll get you part of the
way there. But I found there's very few
(29:31):
instances where I don't want to write my own emails because
writing is akin to thinking. And it's very important that we
don't let AI take away too much of our writing.
If it's a generic thing that I'll send over and over, sure, I
can let AI take a template and run with it, do a little
customization. But for the most part, if I'm
writing something to someone, I want it to be from me.
So AI can't quite succeed there in my workflow yet for people
who want to do things like booking a flight or all those
(29:53):
classic demos, that really mightsave a bit of time.
I am not comfortable with the AImaking decisions that involve
like transactions. I don't want it purchasing
anything on my behalf. We've run experiments before and
it's just, it's not great yet. There's no promise.
I'm still optimistic, still bullish, but I don't currently
use anything to operate the web on my behalf.
There are however tools like browser base or Exa or fire
(30:18):
crawl which will actually do webbehaviors behind the scenes like
searching, crawling, retrieving information and that I'm all for
pull in as much as you can. Let's automate it, let's get it
so it's in just for our systems.But to actually do the actions.
I haven't found one that I like yet, so if you know any let me
know. I'd love to try out more, but
for now web browsing is just an input to my more robust, mostly
(30:43):
deterministic systems rather than let it run off on its own.
One caveat I will say, if you leverage a system like E2B or
you're able to dockerize environments for to have the AI
operate in, that can be safer. But again, it's not necessarily
what you're worried about on on your machine.
It's it's the web and the One Direction aspect of it.
(31:04):
It's also a risk where people can embed hidden text in ASCII
characters or invisible text on websites, things that can lead
an agent down a wrong path that you won't see.
And the web is the Wild West. There's so much, it is pretty
much Infinity opportunity for malicious behavior on the
Internet. If you let an AI get some
credentials, if you give it yourcredit card and just release on
(31:24):
the Internet, you have to be willing to accept the
consequences of that risk. And that's just somewhere where
I'm not at yet. But being able to build a system
that can ping a URL, pull back all the data, process it, and
present me with some informationthat's helpful to my workflows,
very bullish on that. So explore those tools, but be
careful when you're using AI on the open Internet.
(31:46):
There's just a lot of risks involved with it.
So one tool that I guarantee everyone here is using, but I
want to touch on a few differentopportunities that it presents
is just a chat interface, a chatbot.
Whether you go on a ChatGPT or use Claude or whatever other LM
provider you want, the things that you do with chat can be
very impactful in your productivity.
And I want to go over two big use cases.
(32:06):
So the first is using as a thought partner when you're
starting your day, planning yourday, planning a project,
planning a, it's very easy to just ask it to give you a prompt
that you can give to a coding LLM or just go directly to the
coding LLM. But if you take a step back and
put more energy into the planning process, you're able to
come to a better result faster. And it doesn't just have to be
(32:28):
when generating an output. You can talk about strategizing
and this garries across the entire spectrum of your life.
It could be a professional or personal If you want to
strategize a weight loss plan, having a thought partner as to
what is realistic expectations for both your calorie
restrictions, your exercise plan, your dietary habits.
Being able to bounce it off of an entity to get extra insight
(32:50):
is going to be valuable. If you're playing out a script,
thinking about different perks to your operating system or
different benefits to using one language over another, it might
be a dead end, but it might actually expose something to
help you think about things moredeeply.
So I encourage people who eitherare in my AI Academy or just
hire me for coaching is just have the little chat interface
(33:12):
open at all times. Go over, ask a little question.
If it's super basic, ask Jan, ask your favorite local model,
and then it can just bounce my ideas off you.
If it's something more complex and you need some more planning,
ask Claude. Enable thinking so we can go
deep into it. Get web search so you can get
the most updated information. But you want to get into a habit
of getting a second opinion on things.
And this isn't to say the AI should drive your decision
(33:34):
making, it absolutely should not.
Remember, LMS are probabilistic,so it is going to give you the
probably best answer. Very middle of the road, very
unopinionated in many ways. Plus whatever censorship or
guardrails that the provider puts on will definitely bias his
decision but or recommendation. But if you are driving the
decision and you're simply asking for an extra set of eyes,
(33:57):
someone to give you a bit of feedback on it, the AI can help
fill that gap. The other one is actually as AAI
therapist and I had mixed reviews if you look back on old
videos. I was very opposed to it, but
eventually I came around to seeing it as a viable tool not
to replace human therapist, but to actually enable a more open
and transparent conversation with the things bounced around
(34:19):
in your head, to get to a more comfortable place, and to mostly
establish strategies and tools that you can have to think
through things clearly. High Neum was on recently and
the first time he came on, he was talking about how going to a
therapist is understanding how you think and can give you
superpowers that way. And while an AI can't go to the
same extent, I have found that if you're willing to be
(34:41):
transparent and open with the thoughts that are going on your
head, it can help you process them.
And one big thing with productivity is can you be
focused? And if you're distracted because
of the hardships of life, because everyone deals with a
whole bunch of hard stuff, you're able to become more
present on your task and be ableto focus more deeply if you can
even give yourself temporary relief from the stresses that is
(35:03):
life. So if you're able to just brain
dump in the morning, the things that are bothering you, the
things that are concerning you, the things that you know you
have to deal with, it just gets them out of your head.
It's like writing them down, butverbally, I still encourage you
write things down. And if you can do so in a way
where that can be an input to your system, all the better.
But it's very important to get what's inside out in a safe
manner that you're comfortable with.
(35:24):
And if you're able to give some of that to an AI just to get a
little bit of feedback, it'll help you get through that
moment. It'll help you take your next
step, be a little more productive for the next period
of time until you can get eitherhelp you need, guidance you
need, or even just feedback fromother human to help you get the
right place. But sometimes you'll be lying in
bed unable to get to sleep because the stresses of the work
(35:45):
day are bouncing around. And if you just say it out loud
to an NT and have something go back being like I understand
here is some strategies to fall asleep faster, it might actually
help you get a better night's sleep, making tomorrow an even
productive day. So I would say before you Goff
at the idea of using an AI as a therapist, just have a
conversation with it the next time something's bringing you
down. You might be surprised with the
(36:05):
suggestions it has, and they canbe pretty thoughtful and
insightful and might give you more tools and insight to
understand how your brain works,what what is helping you escape
from a spiral. And then you'll be able to be
more productive just because you're in a better state of
mind. Well, that was a lot.
It also feels like it wasn't that much at the exact same
time. I can go deeper into detail on
any of these topics and I'm really curious to see what
(36:26):
resonated most with you. What do you want to see more on?
What can I double Click to explore more deeply?
Please let me know. There's a few things I do want
to bring to your attention just to make sure you're aware of it.
Things like the sycophancy or the irrational disagreeability
of LMS. If you tell them, be critical.
You can't always trust the outputs, and I encourage you to
not trust the outputs. Validate, double check.
(36:47):
Make sure there's a human in theloop.
You might want to optimize productivity and just say AI go
crazy and just full throttle ahead, no guardrails, but you're
going to end up spending more time fixing things if you go
down the wrong path or if there's mistakes along the way.
So these are fully intended on augmenting you, not replacing
you. And it goes the same with your
(37:08):
workflows. Augment your workflows, don't
replace it. If you can use deterministic
code that's reliable, it will produce the correct output more
frequently. So make sure that you use the LM
to generate code rather than just all the LLM to do a certain
tasks. The one example I have is my
thumbnail generator, The square thumbnail generator.
If you ask the AI to do every single time, it'll work 9 times
(37:29):
out of 10. But then sometimes it'll go off
the rails. It might install the wrong
library. But if you have the AI generate
the code to do it, it happens instantly every single time.
And then you can trigger that from your Obsidian vault.
So you get start getting these interconnected pathways.
There's so much more that we need to explore, but I wanted to
give you the foundation so that you understand how I've become
massively more productive with the help of these systems and
(37:50):
get inspiration. What works for me might not work
for you, but the principles remain.
Figuring out your processes, thestep by step instructions that
you go through in your head automatically or consciously,
and then being able to start offloading that to technology.
There's so much more to cover. I want to give a quick shout out
to Tool Hi for supporting the show.
(38:11):
So I can do these nerdy deep dives and teach you the things
that I've learned over the years.
But I want to go a lot deeper. So please let me know what you
want to see and then I'll see you next week.
Thank you.
Popular Podcasts
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!
NFL Daily with Gregg Rosenthal
Gregg Rosenthal and a rotating crew of elite NFL Media co-hosts, including Patrick Claybon, Colleen Wolfe, Steve Wyche, Nick Shook and Jourdan Rodrigue of The Athletic get you caught up daily on all the NFL news and analysis you need to be smarter and funnier than your friends.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.