Subscribe
Share
Share
Embed
Do Good Work is not a label but a way of living.
It is the constant and diligent effort to achieve a new level of excellence in one’s own life.
It is the hidden inner beauty behind the struggle to achieve excellence.
It is not perfect but imperfect.
It is the effort, discipline and focus that often goes unnoticed.
The goal of this podcast is to highlight that drive.
The guests I have on this show emulate this drive in their own special way. You’ll be able to apply new ideas into your own life by learning from them.
We will also have 1on1 episodes with me where we’ll dive into my own experiences with entrepreneurship and leadership.
Every episode is designed to provide you with ideas that you can apply and grow in excellence in all areas of your life, business and career.
Do Good Work,
Raul
Raul: You can tell a lot about where
people are at when using AI by the
words that they use to describe what
they're doing, how they're presenting
what they're doing, or even what they're
actually delivering, because even a
lot of the AI transformation firms
might be out there still running on
manual back office and Google Docs.
I started taking AI seriously about 20,
like early 2023 to constantly improve
the way that I work, deliver value.
When I increase the value for I
deliver for clients and the value
that I can create from researching,
to auditing, to sales calls, to
combing through piles of information
to find the needle in the haystack.
And a lot of companies right
now, agencies, consultancies are
pressured to become AI native.
And a lot of people are talking about it.
So what I want to do in this pod is
peel back the curtain and actually share
firsthand experience of what I'm actually
doing in my own company, as well as what
I'm helping some clients go through.
As, because this is a journey, this
is not like a one and done thing.
It's not something else you do in one day.
It's a constant evolution and I think
there's a version of your company
that is worth building towards.
If you're working in a team or you're
actually leading the team or you're
running the company, getting out of the
trap of how do I avoid getting replaced
to create a new kind of company, the
one that delivers your value to your
customers even with more value or
faster speed to value or increasing
your margins potentially creating a
category of one where you can own and
really create new pricing or creative
pricing to help clients go further.
Now, the only way we can actually do this
is by embracing the technology for good.
So everything that I do inside
of my own company runs on AI.
And yes, the more that I work with it,
the more that I want to have local models
that I own and that work and run here.
We'll get to that in a second, but we've,
before we dive into the tech, what I
want to give you here is a framework
of migrating your own work to become AI
native so that you can build the firm of
the future using a systems perspective.
Instead of chasing the latest fancy
tool or automation or CloudMD, and
whatever you see online right now, by
the way, as a side note, whatever you
see online right now with other people
sharing, "Hey, get my skill, get this.
A lot of it is hot garbage.
If it is worth looking into, just have
one of your agents reverse engineer it.
Don't copy and paste things that people
are posting out there or skillsets or
uploading them to your to your, like
whatever OpenAI or, or Anthropic,
whatever you're using, don't automatically
upload that because literally you
can reverse engineer it and some of
those may or may not have malware.
All right, getting back to the original
point, there are four stages right now
to migrate your company to become more
AI native, and they are sequential.
Each builds on top of each other,
you can't just go from step
one or stage one to stage four.
But here's the thing, you wanna operate
at the stage that you're comfortable at.
So you don't just need to you don't
need to get to stage four if you're
comfortable at two and three.
But I would recommend knowing all
these, seeing the potential needs and
maybe having some of your com- part of
your company run at multiple stages,
but to the level that you can manage
and the level that you're comfortable.
The four stages in order are
enhancement, stage one, stage two,
augmentation, stage three, parallel
operations, and stage four autonomy.
So let's dive through each.
So in stage one, enhancement this is
where everyone actually starts, and
this is where you probably are where
I've been as well, but this is when
people say, "I use AI in my work."
there's a lot happening in the
market right now and people using
AI in their work, even your team
is either telling you they're using
it or they're not telling you.
That's called Shadow AI, where team
members are using it to operate
based on your company data or your
client's data without you knowing.
So it is important for you to
orchestrate that and have everything
dialed in in one location.
But the key thing here is everyone, every
person right now is using the technology
to enhance what they're currently doing.
This is a simple stage.
You are already doing your daily
workflow, but you're just leveraging and
topping AI on top of what you're doing
to do things faster, to go a little
deeper or to create real efficiencies.
So if you're a leader, it's up to
you to make sure that you can help
the team and empower them to use the
infrastructure that you have created
that is safe, that's not training the
actual LLMs or the frontier models on
your proprietary datasets to make sure
you have the right plans and make sure
that your team is using them accordingly.
What doesn't change in this
stage for your company is how you
structure your operations, your
offerings, or your price points.
You're not redesigning your work,
you're just improving the work
that you're currently doing.
For example, in the past, I would
audit for teams for clients.
I would audit their sales calls.
I would go through their Gong recordings
and listen to as much as I could in a day.
Now that's a lot.
I couldn't listen to the
last three months in one day.
However, with AI, obviously, with my
frameworks, I've been able to train AI
on that, look at the call transfers,
help identify what I wanted to find, and
find the needles in the haystack within
a day, and be able to audit literally
three, six months, within a short
timeframe, and leverage my frameworks to
get to the truth or get to the answers
that I wanted for the insights that
I needed to work on for my clients to
get value quickly, because I needed to
deliver value in a short amount of time.
So I'd use my judgment, my frameworks,
my way of auditing and analyzing calls,
and using AI to make work more efficient.
This is stage one, and you're
probably doing that right now.
Maybe with client prep, maybe for
research, maybe for cleaning up your CRM.
And a lot of the tools that you currently
use now are adding their copilots.
They're beca- becoming
quote unquote AI native.
So there's nothing new here.
The only thing is that you might be
doing what used to take hours in 20
minutes, but the dark side of things
but the dark side of this is that you
might be doing more 20 minute work.
So instead of doing one thing for four
hours, you might be doing 12 or 15
things for 20 minutes, so you're using
up the same amount of time, you're just
being more productive or doing more.
But burnout can be real hair.
So maybe you're not really doing a
lot more to get time back and you,
but you are getting more work done.
The only caveat is to be careful is
burnout because burnout is real here.
Now, enhancement gives you more capacity,
but now it's up to you for how you use it.
And this stage, stage one and
enhancement has its own ceiling.
You can't cross even further
because you can't just do more
in the same amount of time.
You can be more effective.
Your margins can improve and you might
do more by the hour, but it doesn't
fundamentally change how you run
your business to become AI native.
It doesn't change your operations,
your team structure, your margin
structure, like your actual profit
margin, your actual leadership
structure, or your go to market.
Enhancement, I think, is an
important stage to be aware of.
This is most, where most people
are at, but it's the foundation.
It's not the actual infrastructure
change that's going to change how
you work and how you deliver value.
It's a prerequisite for stage
two, which is augmentation.
So this is where things start to
change in phase two or stage two.
It's less about delivering more outputs.
It's less about doing more and focused
on packaging your intellectual property,
how you deliver value to clients
and deliver that value at scale.
And one of the biggest internal lessons
for me, and it took me over two years
to develop this daily habit, was
to write every day, to think about
my own thinking, to write something
about it, and to publish that, even
internally or just in my notes app.
And to actually get into that
practice took me two years to
do that, thanks to Seth Godin's
program and Ship 30's program.
But thinking about your own thinking is
just another fancy way of saying writing.
I need you to write.
I still need you to think because the
difference between doing more work and
scaling your IP, your methodology, how
you deliver value is when you define your
methodology of how you deliver value to
clients, and when you can define that
and articulate it, you can train AI or
agents or workflows to help you compound
what you already do to either do it more
effectively, to go deeper with clients,
to delivering your intellectual property
at scale without, you need to be there
at every step, but most importantly
radically changes how you see your
work and how you execute that work.
Now, a lot of the things that you're
doing right now is based on a method.
It's based on a protocol.
It's based on the things that have
gone success for clients in the past.
You just have to codify it.
If you haven't had it documented, you
need it documented because you, if you're
not documented, it's like the old trap.
You're either going to keep doing it
yourself, have someone in your team
do it based on the culture, what they
actually think needs to be done, or your
team plus technology is going to do it.
But then now, you can start leveraging
AI and agents, like actual agents with
the right tooling, the right process,
the right methodology to fill in the
gaps to help your clients go further.
And I talked about this and I wrote about
this in the services stack where you can't
train AI to your level of judgment, okay?
Your particular insight, your nuance
take on client deliverables, but you
can train AI on simple execution with
things that needs to get done, with
templatized strategy or just if then
conditional strategy, you can do that.
One important thing that I want
you to do is there's a point where
AI ends and your thinking begins.
And when I'm, when I'm encouraging you
here to write about your own thinking, to
write, to put down your methodology, to
actually codify it, is to understand what
is the 10% that you intuitively drive?
What's the nuance that
you bring to the table?
'Cause if you know how these things are
trained or how they actually developed,
they don't have every single piece
of data point in the universe because
they don't have experiential knowledge.
They don't have relational knowledge.
There's a lot of things that
you still bring to the table.
All right.
Without diving into that, I just want
you to think about what is the last 10%
of your intuitive judgment of what a
client actually needs and the strategy and
direction that you need to take a client?
How do you codify that so that you know
this is where AI ends, everything else the
AI can do, or the series of agents can do,
and I'll, and I'll share more about that.
But here is where I come in.
Here's the nuance takes that I can do.
Because once you're able to do this, and
I'm speaking from experience here, you're
able to take work that used to take 20
to 40 hours, and in some cases, for me,
on average 20 hours, and condense that
down to half a day, and that's what, two
and a half hours, two hours, three hours?
I'm not exaggerating, that's just
something that I'm personally doing, and
I know where the AI ends, I know where
I need to step in, where I need to be,
a co-creator, a thought partner, and
also directing it of what we need to do.
For example, and I'll give
you a clear example here.
I have, I've had this advantage, and
you might have this advantage too, if
you've written about your own thinking.
I've written down my methodologies
and published several books.
None of them made The Wall
Street Journal or New York Times.
They helped clients get real revenue.
That's all I really care about.
But the whole key thing is I was able
to train my agents from day one based
on all my writings and methodologies.
Over 566 different thought pieces, like
how I think the frames that I have, the
approaches that I have in the world.
And it's, again, it, this is what
it told me, what my agent told me.
It's over indexed over 3,000 documents
of our conversation strategies and
actual things that drive the chain.
However, with that, I've been codifying
my methodology, not into just one
master skill, but a series of nine to
12 different skillsets and different
agents and toolings that can run
and help me execute for clients, not
just faster, but higher quality in
a shorter amount of time, and now
I can add net new value to clients.
All the parts that I know AI is going
to do better for me, deep research,
competitive landscaping, surfacing
insights from specific questions that
I'm looking for based on specific
context or situations that we're facing.
I always know and feel where I'm
gonna take the client, how we're
gonna craft their offer, what we need
to trans- transform their business,
what the mindset of the client is.
Because I've literally trained thousands
of entrepreneurs, I've seen behind
the scenes of hundreds of businesses.
From my perspective, I do know what's
going to work, what's not going to
work, but I'm leveraging AI here as a
co-partner to help me co-create things
that I couldn't do at this level of speed.
I am working and executing now at
the speed of my thoughts, right?
That's a real thing.
You can probably do that as well, and
you're probably experiencing that too, if
you've been delving into the technology,
actually using it to its full potential.
But here, I want to discuss a few things.
Speed to value is the first metric
that you're going to feel, and
speed to value is the first metric
that you're going to measure.
Like, how quickly can you get to value?
I think that's great.
We're always thinking about
efficiencies, but I think on the
split side, we need to start talking
about what's the net new value.
Because once you understand efficiencies
and you're able to get to value faster,
then it really begs the question, what
is net new that I couldn't do before?
What are the things that I can add value
to my clients and help them achieve
that I couldn't do before or because
I didn't have the, the competencies in
terms of technology or the team members
or because I didn't have the time and my
bandwidth was stuck in the things that
again, used to take 20 hours, now take
about two and a half to three hours.
Either your bandwidth, your energy,
your money was sucked into other
areas, and now what is the net new?
I, this is all a Substack
post, by the way.
If you want, go to my Substack,
do goodwork.substack.com.
And I have a link here to a whole essay
on what net new value and also another
YouTube around what net new value is.
All right.
So anyways, the key thing for this stage
or stage two, the prerequisite is for
you to codify your intellectual property.
Now, I know a lot of people and I've
talked to these founders and a lot
of them were former clients as well.
They're thinking if I train the thing,
isn't the thing just gonna steal my IP?"
Yes and no.
Yes and it's like, it's like saying if
I connect to the internet, is it going
to steal on my entire bank account?
You can't bank on the internet.
Everyone banks on the internet now, right?
So just think about that.
There are some features and some level
of plans that you can buy where you're,
like, the data's more protected,
you don't train it on your own data,
plus where everyone is trusting,
even the government is trusting.
It's not trading on its data, all right?
So all I'm saying is that if you
want to move forward and move fast,
do the right things compliantly,
but also think about this.
How do you codify your
intellectual property?
Now, if you haven't actually written
down your own thoughts in terms
of writing, that is the number one
strategy you can think, do right now.
Think about your own thinking.
Just te- spend 10 to 15 minutes every
day, use that organ in your head, that
organ in your head that didn't cost you
a hundred billion dollars or whatever the
number is that can do significantly more.
Use that organ in your head,
think about your own thinking,
reason through, and know where
does AI end in my thinking begin?
If you can't do that, I'm not here to
scare you, but legitimately, if you
can't do that, you will be replaced.
You are in real trouble.
All right, stage three.
Now that we do that, now we
talk into stage three where
it's parallel operations.
So similar to stage two is an agent
working with you to, to deliver value,
but now in, in parallel operations,
you're also looking at the other
areas of your business model to help
create agents, to help you become
more effective and more valuable.
Now, when I talk about business models,
a lot of people are like, "Okay,
what, what is a business model?"
The nice thing is that there's
a formula to a business model.
A lot of these things are formularic.
One of my favorite approaches to a
business model is the business model
canvas, been using this tool for, I don't
even know, over 12, 12, 14, over 14 years.
It's an amazing tool, but essentially
here are the different criteria
that make up your business model.
Your value proposition, how you keep
client relationships, how you reach
clients, who your ideal client profile
is, your ICP, how you make money, your
pricing structure, the key activities,
your key vendors, partnerships and
resources, and your cost structure,
meaning your expenses and your overhead.
That sounds like a lot, but I
codified this into three parts.
Every single business in the world runs
on these three parts, codifying the
business model, and you operationalize
it with, number one, operations,
administration, finance, strategy team.
Number two, product, client
fulfillment, lifetime value,
retention, R&D, inventing new ways
to help clients go further, product.
So you have your operations, you have
your product, which is your service or
products that you help clients go further.
Then you have marketing and sales, how you
get and keep clients, rinse and repeat.
I combine marketing and sales.
That's a different conversation
for a different time, but those
are the three essentials that every
single business in the world has.
Operations, product, marketing and sales.
Yes, your people live under operations
because everything runs with either
you, another human or a machine.
That's pretty much it.
I'm trying to not
oversimplify, but showing you.
Now, if you've codified how you've
deliver value under product, now, what
if we codify how you operate, how you
get and retain clients, your key customer
relationships, your key activities.
You just need to know, again, where AI
falls short and where you need to come in.
And you also need to know okay, this
is like the beginning of philosophy.
What is real?
Based on what is real,
what are other people?
How do I treat other people
based on what I know?
Some companies right now are
treating people like machines.
Do you want to be that company?
I don't think so.
If you're listening to
me, you probably don't.
I think that humans are at the forefront,
that we have to help humans flourish.
The technology has to be for us to enable
flourishing, for our communities to
flourish, for you to flourish, for your
business to flourish, for your customers
to flourish, for your impact to flourish.
You can hear the voice, tone change.
Might be a little passionate about this.
Key thing, how are you
codifying your operations?
How are you going to execute that?
Another key note is that the key operators
that I've been working with and friends
that I have or tech leaders or even
clients who are literally leaders in
the space, CTOs, leaders at other tech
firms, at other big name companies
that you might know, the conversations
and then what I'm hearing, the best
operators right now with this technology
still keep the human at the loop or
in the loop, just involved with the AI
where the agents, the AI reports back
to you and re- works alongside you.
That's stage three.
You're running parallel operations,
helping you execute, they execute,
they report back to you, you iterate,
you think, and then you move forward.
So you're literally staffing a small
team or a small set of functions that
you've programmed based on what you
know needs to happen for the business
to grow internally for your clients
to thrive in all three core areas,
product, marketing, operations.
And This is exactly what
Jensen Huang described.
We're entering the era where it's
not the best model that wins.
It's the system that harnesses
you and the models that you use.
It's essentially the system's thinking,
the strategy of how are you using these
tools and directing them to an end goal
and an outcome that you can utilize to
flourish, ROI to help people go further.
So you can leverage literally any AI
model, LLM, free, open source, frontier
to create value for yourself, for your
company, for your clients, for the
marketplace, again, for you to thrive.
I'm not overstating this.
It's not an SEO strategy.
I just really care.
How do you make this tech pay
for itself so that you can
help other people go further?
Now . A lot of what I've
spoken isn't just theoretical.
Let me give you some real
examples and Agentic workflows.
I think I got about about nine
examples I'm gonna share with you.
And some of these, I am
going to give you overviews.
I have literally copy and paste templates
that you can read yourself and audit,
that it's clean, it's completely in
in marked down that you can audit
and read at do goodwork.substack.com.
You can take a look at it, and then
from there, you'll be able to know
how these things work and also codify
it for yourself, or even hand it over
to your agent so your agent can start
building this out with you, and you
don't have to copy anything that I do.
Just "Hey, this is inspiration.
What can we do?
All right, so the first thing
that I would recommend you build
right now for parallel operations
is building your security agent.
It's probably the most important things
that you can do, and it's not sexy.
It's critical though, because if
you're still downloading skills from
the internet, pulling repositories
from GitHub to enhance your
agents, or using repositories,
assuming they're safe, don't do it.
Some of them might have valuable.
Like you might be uploading skills
that are potentially dangerous.
So here's the ingredients of what
my actual security agent does.
Please feel free to copy and
paste this into your own agents.
Again, I don't have anything
for you to download.
I don't have any repository for
you to go and just swipe this.
I'm giving this to you in plain
human language in English.
All right, so this is what my security
agent does for me, and I'm just going
to give you a quick summary of that.
It reviews any code before
I actually install it.
It audits any repo before I ever touch
it, and we actually have a rule where
it actually never auto downloads or
auto runs or auto executes without
one, auditing it for security.
Number two, it builds it from scratch,
meaning if someone has code written
on the repository, we actually don't
download the code and install it.
We just rewrite the code based on
our filter to make sure that we're
writing it clean with a fresh build.
We don't download, we build.
I also audit any API co- connection.
And if you're using Mac, which I am,
it stores any API key chain into,
into the key chain, never in plain
text, and never in environment files.
Like that's, that's super critical right
now that you might have your API keys
literally hiding in plain text, that if if
something malicious were to happen and you
would get hacked or you visited the wrong
website, they were able to find all your
API keys, which means that they could run
all the tools that you run on your behalf
and cost you a lot or even steal from you.
It also watches out for anything
I do in my workflows that touches
sensitive business information and
knows what to store what not to
store, which is really important.
And it runs before other agents run.
So if I tell it to do something, we're
building something new, I always have
it go through the security agent so
it's like a security layer the second
thing that I think is useful for you to
build out for your operations, parallel
operations was what I call a review chain.
So this is the heart of a lot of
the things that can happen at stage
three, is that if you just run a
single agent, they can make mistakes.
They hallucinate, they miss context,
they make things up to make you happy.
They can get lost inside their own prompt
instructions, like they can forget to run.
And the fix isn't just let's make
this agent super strong and super
smart and hardened the agent.
The fix, for me at least, is a chain
so the chain is actually a
worker agent that does the grunt
work and produces the draft.
It could be a research, a
deliverable, or an analysis.
And then above it, I have a domain expert
agent, which we trained based on my
domain expertise in that specific thing.
So I have nine agents working
for me, like a sales agent, a
content agent ops agent, et cetera.
And they review the output.
And also the worker agent
has to show its work.
It has to prove its work.
It can't just say I did the thing.
It has to prove the logic.
The domain expert reads it and reviews it.
If it's wrong, it sends
it back to do the work.
And then the chief agent, the top of
the chain is a cross domain agent.
It's the chief that manages all
my agents, and it works with me to
collaborate based on what the agents
did, and then I'm with the agent make
the final call, and then we execute.
So why this matters is that
everything has its own homework,
and you can grade that homework.
The domain expert catches any errors that
the worker can see, and the chief agent
catches errors upon multiple domains,
and works with me to make sure that we're
in alignment and make sure that we can
run it, and then most of the time with
that, honestly, it's pretty spot on.
So you don't need three layers
for every piece of your work, like
light, simple work "Hey, search
this domain or search this topic."
But for full chain work, like you're
actually doing specific strategy
decisions or client facing outputs or
anything with an execution cost, you
should probably run it through the chain.
You never want to trust a worker's
agent to self-report on quality
because sometimes workers say,
"Hey, I did an amazing job.
I worked for five hours.
Here's my output."
And you can't just trust that
you ha- it has to prove its work.
It can't hide from the output.
So you always want to press your agents
to do better and review, and I've just
codified that with a systems review.
So going back to stage three,
I gave you two simple examples.
Again, what I would do in parallel
operations, so actually what I
did in parallel operations so that
you can model it and work with
your agents to build this out.
But the prerequisite for this stage
is that you know how to build agents
or know enough to direct someone else
or your, your AI to build it for you.
You can articulate your methodology and
approach well enough so that you know
exactly where AI falls short and where
your nuanced thinking and strategy and
brain come in, either yours or your teams.
And the only way to do that is to
codify everything, to understand
the nuances of your, and your team's
actual thinking and judgment and the
execution that needs to happen, and
you leverage the agents to parallel the
work with you as, as actual co-workers,
not just tools, but jobs to be done.
So that's the prerequisite to
actually fully utilize stage three.
Now, stage four is autonomy, and this
is where we need to be careful, either
because one, it's been oversold, like
with OpenClaw, or it's been misunderstood.
Now, don't get me wrong, I'm
a huge fan of open source.
I love everything about the open
source community when it's done
right and with the right intent.
But what I do want to make
clear is a lot of these are
pre-programmed to feel autonomous.
No A- AI agent thinks on its own.
And there's a lot of people with a
lot of investment, a lot of money that
have to show the world or convince the
world that it actually does because
they have to make the money back.
But right now, no AI agent
actually thinks on its own.
It's pre-programmed to execute because
they use cron jobs, they run on the
schedule, or they use heartbeats.
They wake up and they run
something that's pre-programmed.
So your job is to design a
system that anticipates your
needs without asking for it.
How do you anticipate your needs without
asking for it, for the needs that you
know you're gonna have and the needs
that you don't know you're gonna have?
It's really interesting.
But you can do that.
You can design it.
So for example, I moved a lot of my
operational H- HQ into Obsidian because
it's a simple markdown file editor,
but for there, I can do everything.
I can sync it to the cloud.
I can have team members
access parts of the vault.
I can connect applications to it.
I have a history and everything that
I work with my agents gets logged.
Now with Obsidian, it's, it's
like a daily planner for me.
It's like an execution for me.
It's like it, it has everything
preset for me every single day.
How do I do that?
Because I have a heartbeat that goes
throughout the day, it logs my progress,
it tracks everything that I'm doing
based on what I report to it, from sales
to product delivery, to any innovation,
and it also does some things on its own,
and I'll share that in just a second,
to my own ideas, to the content or the
things that I wanna get done that day.
And then also reads my schedule and
based on who I'm meeting, it creates
an agenda in alignment to my goals
and my purpose for my work, and it
has my ethos baked into it, and I'll
share more about that in a second.
So by the time that I show up at
my desk, it has everything ready
for me to go in a daily note.
And for you, this might sound, or
me for some, but not for everyone.
This might sound like some genius
AI that thought about it on its own.
To be honest with you, I just
designed this based on what
I actually know that I need.
And plus, I've actually run teams
and scaled multiple agencies.
I've helped lead others and I know
what it takes to run a company.
I know what I need.
I know what other people need.
I know how, the infrastructure
that's needed and it just,
now it just does it for me.
And honestly, my computer
never sleeps anymore.
I might turn it off over the weekends
just for my own peace of mind.
But every day I show up, I have a daily
note that tells me, "Hey, here's the date.
Here's who you're talking to today."
And every single person has their
own CRM card with their own agenda,
with their own key thing that
I'm, what I'm gonna talk about.
It also has my goals based on what
I've trained at SoulMD, and I'll give
you the example of the SoulMD in the
Substack where essentially the main
orientation is human flourishing, so
that I flourish, my clients flourish,
and whoever I speak to flourishes.
And it also shares with me the key actions
I need to do today based on those goals.
Then it goes through my key habits,
the people that I talk to the writing
that I would wanna do that day, or the
content that I need to think about.
It also goes through value ads that
it though about and quote unquote
dreamt about overnight for clients or
agendas that I have to run that day.
And it also helps me plan out things
that I should build for the future, like
redesigning an, an agent or changing an
approach here or changing a workflow.
And that structure I've trained teams
on, like leadership teams at retreats
or one-on-ones in my 80 / 20 training.
And that's like a free 40 minute
training that you can watch.
It's called 8020 training.
Just Google Raul Hernandez, 80 / 20
training, it'll show up, it's a 40 minute
video and it literally, that structure
is what I use and now my agents prefill
everything for me and do work for me.
So if you go to Substack as
well, do goodwork.substack.com,
you'll be able to find a
template of my daily note.
It's a really robust marked down template.
You can tell your agent to do the
same thing, and it also tracks My
accomplishments, so I actually list and
I final, like I have a final list at the
end of the day what I got done that day.
I also on that same Substack have a
templatized version of my SoulMD and this
is how you instruct your agent to behave
when working with you and other agents.
So I'm not gonna read the soul, the
whole SoulMD because it's very long.
But I wanna dive through the structure
and actually read a few things for
you that I think would be helpful.
So when you go to the Substack, you
can just copy and pa- you can read
it first so you can see that there's
nothing malicious about it, but read
it first and you can copy and paste
this and, and, and review it later.
But the SoulMD has a template
section of how to use it.
So the Soul MD is not the
system, it's who the system is.
It's describing here's who you are,
here are your values, and here's how
what we're never going to do, and here's
how you're going to talk to me, here's
how you're gonna execute, here's how
you're gonna think about things when
we talk about these things, here are
the boundaries that you have, here's
the relationship that we have together,
here's the vision that we have, here's
how you're going to evolve on your own.
And from there, it runs every single time.
You there's clear instructions in that
sole MD that I have templatized for you,
for you to see how this is going to run.
One of the things that I think
is important that I want to read
verbatim here is the human flourishing
part, because I think this is
really, really, really important.
And this is, this has fundamentally
changed how my technology works, so
I think it is worth reading, is that
everything in the system that we build
must help other humans flourish, and
it must help Raul, me, in this case you
replace it with your name, flourish.
If an optimization serves metrics,
but not people, it fails the test.
If a shortcut saves time, but
undermines trust, dignity, or
real value, it fails the test.
This means never fabricate data,
dishonesty degrades trust, never
automate away the human relationship,
the network is the greatest asset,
and never optimize for efficiency
at the cost of the soul of the work.
End quote.
The reason I wanted to share that with you
is a lot of the slop and a lot of the crap
and a lot of the things that we're seeing
from AI is because people don't know how
to actually orient it, their technology
to harness it in the right direction.
And I wanted to share that with you
because those simple lines, simples have
radically changed my interaction with
technology, the things we're able to
build, the decisions we work through,
the strategies I'm able to co-create,
and the ideas I'm able to drive forward.
Obviously, there's still times
where I direct it because I know
what needs to work, or sometimes
it deviates, because, you're using
providers or you're having it local.
Regardless, it has some, some
presets, like working with any human.
It has some presets, some pre-orientations
already set up, but still, that has
fundamentally changed the approach, and I
think it's very it's been very fruitful,
so I wanted to share that with you.
Now, going back to, to what else
my agents do is that it takes
note on what happens overnight.
So it has quote unquote dream mode,
and again, I'm not using the open
claw right now, but you can really
set this up yourself, is that, because
essentially it's just Python scripts.
You're setting up automated scripts to
do this, but in quote unquote dream mode,
it consolidates its own memory, and it
lists the accomplishment of the day and
tracks and tracks for me personally,
am I being more productive
or am I just doing busy work?
And just to give you some transparency
that, because again, that's like
the number one KPI, productivity.
On a regular day, pre-AI, pre-native
AI, I would get maybe three big
things or six big things done.
That's a pretty good day.
And like on a, on a weekend or on a
day with no calls, maybe 13 things.
And that's, those are pretty good days.
Like for, to be honest with you, if
you're focusing on high level strategic
things or executing on those things,
now at the end of the day now, like
with native AI, my average day is
like 30 to 40 things accomplished.
In a peak day, like my latest peak day
was close to 80 things accomplished.
So efficiency, again,
is a easy measurement.
It's a real measurement because these
are things that are getting done on
my behalf or things I'm co-leading or
things I'm executing by myself, but
the real measure is net new value.
The value I'm creating for myself, for
my clients ... things that I couldn't
do before or wasn't able to do at the
highest capacity that I wanted, and
the easiest measure for me for net
new value is revenue produced based
on work we co-produced with my agents.
And I don't like putting numbers to
it to sound hype, but I'll just say
just in the last couple of months,
it's been north of six figures for
myself and clients included, and
this is outside of clients whose deal
sizes are like five to 20 million.
So that's, that's a real number.
Those are real things that are
happening, and these are just
like, this is the beginning.
There's some things that are in
the pipeline that haven't even
closed yet for myself and clients.
And this isn't because AI did all
the thinking by itself, it just
amplified what we were already
doing for myself and for my clients.
Now, here are some more examples of
autonomous pieces that are running in
the background for you to get a taste
of what phase four can look like and
here's actually what's running for me.
So the first one is a pre-call
research and agendas automatically.
Now this one's pretty simple, but for
every call on my calendar, there's a
research docsier, a draft on the agenda
before I ever show up to the call, and the
draft in the research is based on my goal.
So every time I sit down with
someone, I know who I'm meeting and
what our goal for that conversation
is based on who they are.
My goal is because it's trained on my
goals, what the next strategic move
is, and also maybe vetting who I can
connect them with, because I have
my, my network literally connected
with my, in my, in my database.
So I can literally think
this happened for real.
Someone reached out to me, a
friend of mine, and's "Hey,
I'm hiring this position.
Who do you know?
" And usually for something like that,
I would ship it on a Friday or check
it later that week but literally I was
just in passing hey, my age, I told
my agent, "Hey, can you just check our
network and see who would make sense,"
and then draft an email back to her.
Literally within five minutes of that
email, she got five names of po- people
who are qualified for the role she
was hiring, and then I made intros
to, I think, either one or two of
them, like directly right after that.
And again, that's very simple,
but it's very proactive and it's
helpful because if the system works
for you, it anticipates your needs.
Another key thing that runs is an
auto research that runs overnight.
So while I sleep, it's a Caparthi
style agent that self-improves, so
this thing literally self-improves my
own agents and skill sets every night.
And we set parameters to it so it doesn't
take the average mean, but essentially if
there's a skillset or an agent that I'm
running, it auto research it and finds the
best way to do what it's already trained
to do based on what actually works.
And then from there, it runs
synthetic tests, synthetic inputs
against itself to improve itself.
And then from there, it reports back
to me and either, A, I have to test
it or B, I have to qualify, do I want
to accept the changes or do we want
to keep the existing skillset intact?
Because the reason it's able to do
this, it's the methodology that I baked
into it stays logged in, that's not
changing, it's just getting sharper
in how it executes that methodology.
So it's pretty cool to see
a report every morning.
My agents are getting slightly
better, and there is plateaus, like
there's only so much you can get
but it plateaus like when we change
something and it plateaus in two days.
That's amazing.
So you take a skill, you take something
you're already running, you can run
auto research to it and you can get
better literally within a week and
instead of spending months of training.
So anyways, most AI agents
and skillsets are static.
If you do something like this, again, on
phase four, you can improve automatically
overnight and get to a plateau.
Then you take real world input,
test it, and then from there
the testing begins again.
Another key thing that runs overnight
is per client value intelligence.
So this runs, depending if it's
systemized, like automatic or if
we run it based on a sequence.
But essentially if this value client value
intelligence tries to find new ways to add
value to clients and creates a game plan
for that, now it doesn't automatically run
it, it automatically runs what we could
do for clients and it brings those ideas
to the table so that I can review and
then from there I know what is gonna make
sense in the real world if we should act
on it, discard it or plan it for later or
adapt it based on where the client is at.
Another key thing that runs
every week is an 80 / 20 cycle
and a monthly content cycle.
So for 80 / 20 every Thursday to draft
my Friday daily note, it reviews,
what did we get done that week?
What were the 20% of accomplishments
that drove 80% of the results?
How do I plan my week to align
only to just do that 20%?
And then we also review my content
calendar, like what are the
performance from last month's content?
How do we improve?
What are some of the hooks and
angles that we should dive into?
And then similar to how I came up
with this podcast and this, this
Substack, I just think about it.
I think about it, I dictate it
and then we work through ideas and
sometimes it interviews me, but this
is essentially my way of creating more
content that's valuable based on my
own thinking to deliver to you and
the system runs for me automatically.
So essentially stage four requires
you to bake everything that we
learned in stage one to three and
then ask your AI agents, where can
you be pro- more proactive for me?
If you don't know, ask the agents,
codify it, bake it into the system
when you do, you'd have what I
call anticipatory intelligence.
And that's a play on words and yes,
my AI agent did tell me to say that
and I did say that, so there you go.
All right, so diagnostic time.
Where are you at?
So here's the simple diagnostics that
we're gonna dive into so you can self
identify where you're at in stages
and what you need to do to evolve
to the next stage, because we just
went through a lot of thing, maybe
30 or 40 minutes of content here.
So use this to self-identify and just
to, to recap what we've learned here.
So in stage one is enhancement.
You're in enhancement if you're using
AI for specific tasks, but it's not
operationally fundamentally changed the
way that you run price or service clients.
You're saving time in this stage where
you're not creating things that you
couldn't do before, like net new value.
Your offerings and your pricing,
your team structure look the
same before AI showed up.
That's when you know you're in stage one
and the prerequisite to move to stage
two is your IP has to be codified and
you need years or at least hundreds of
repetitions of real practice and writing
down your own thinking about how you
actually operationalize what you do,
your frameworks and methodologies and
decision trees so that you can see what
AI can amplify and where it falls short.
Then you move into stage two augmentation.
You know you're in this stage.
If AI is delivering your methodology at
scale, not just your tasks, not just your
things, it's actually delivering value.
Things that used to take 20 to 40
hours are now happening in half a day.
And now you're getting to the point
of like, how, you're seeing the line
and you're saying, "What is the new
thing that I could do or the new
offering that I can offer my clients
or something that I couldn't do
before but I always wanted to get to?
That's when you know you're in stage two.
The prerequisite in stage two before
you can move into stage three is to
know how to build agents and to know,
or either direct someone who does.
And you can articulate your methodology
clearly enough so that you exactly
know here's what AI can do, here's what
it can't do, here's where I come in.
And then also looking at your, your
business itself, what are the three
core areas of your business and
operations, product, marketing and
sales that you wanna help systemize
with AI and you can actually build
and run and execute agents there.
Now in stage three, we're
in parallel operations.
You know you're in parallel operations
if you have multiple agents running
alongside you, executing across
product operation, marketing and sales.
These agents report to you and
operate with human in the loop so
they're working with you hand in hand.
They feel, and if they're autonomous,
they might feel autonomous, but they're
not, but you know that they're working
there or you're directing them and you're
working with them day in and day out.
You're staffing essentially a small
team or a small functions of programs
that you've pre-programmed yourself
or your team has and your capacity
constraint by thinking, not by doing,
especially if you're really focusing
on your time because I know that
you can get overwhelmed and burned
down by doing way too much, but in
your new capacity is creativity.
It's what am I not thinking about that
I should be thinking about that we could
be doing that my AI could do for me.
That's your real capacity.
That's actually the real unlock to
know that you're in stage three.
I said unlock, that is not
what AI told me to say.
I, that's just what I thought.
So the prerequisite to move to stage
four is a clean, consistent data flowing
from your system and reliably knowing
what is working, what is not working
and knowing where your agents can be
more proactive for you so that it can
anticipate your needs that you know
and be proactive to create systems
to anticipate your unknown needs.
And then you're in stage four, autonomy,
where the system now anticipates what
you know you need and it can also,
the system itself can help anticipate
that you don't know that you need.
It's oh, that was actually helpful.
Like sometimes my agents, when I come
to it with a problem, it solves the
problem and then it goes further and it
patches the problem and it goes further
and creates it I'm like, wow, they, it
did everything I wanted and more that I
didn't even know the system is baked in.
It's something that's designed,
but it also helps you go further.
It might even help you surface things that
you didn't even know you needed or gaps.
So again, that's what I just mentioned.
Your daily plan, and this is just
for me, but your daily plan, meeting
prep content or decisions are being
assembled on your behalf and synced
back to a single source of truth.
So your single source of truth has, is
a database that could be how, wherever
you want to set up your database.
Right now I have a local repo
and I also have Obsidian.
And again, that can sync to the
cloud, that can be like your notion.
And you're spending a majority of your
day sitting on judgment and thinking,
not just executing, but directing, and
that's when you know you're in stage four.
So where is all of this going?
So from my vantage point of the firm
of the future, if you can codify all of
this and become AI native, I think this
is going where you can take more risks.
So here's what I mean by that.
If you can help clients go further,
then you can participate more in the
upside of the value you create for them
in a more bold way and proactive way.
Now, co-risk offers aren't new,
like this has always existed, but
I think we're gonna be seeing more
outcome-based pricing and outcome-based
deliverables in the agentic era.
And the, the key thing is you can now
afford to take more risk if you actually
solidified your methodology, you know
this is what works, you just have to
find the right type of client, the
right person, the right offering, the
right marketplace, the right mindset
to, to make your methodology work.
'Cause if your thing actually works, if
your methodology actually works and you
have agents to make sure that it works and
you believe in what your client does, so
there has to be that framing of belief.
You believe in the work your client does
or the services that they deliver, the
products that they sell, so much so that
you're willing to take a risk with them to
get a bigger potential upside, as opposed
to just saying, "Here are my normal
retainer or hourly fees," then I think
that's the firm of the future to be able
to take more risks because right now when
you're selling project fees or retainer
fees or hourly fees, don't sell hourly, by
the way, but if you're doing that, you're
telling the client, "You take the risk."
Yeah, there's risk reversals and we can
make it work, of course, but I think it's
even cutting more into co-risk offers
with the client where I am taking more
risk because I know this thing works.
That does require, one, you
to be a real expert, two,
actually believe in your client.
Number three, for you to become AI native,
to radically redesign how your company
operates, delivers value, packages your
offerings, prices them, goes to market
and sells because it's not the same.
It literally isn't the same.
I'm living proof of that, and
that's where I see this is going.
And this is what I'm betting on.
This is what I'm doing myself.
So if any of this resonates with
you and you want to co-design how
you operate and sell And scale your
company in the Agentic era, there's
two ways that you can reach me.
Number one, you can book a call find a
link somewhere here or the Substacks, you
can D me on, DM me on LinkedIn, or just
reply back to any email that I sent out.
Either way, this is Rod
Hernandez, do good work.