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
I want to share with you a workshop
that I did here in San Diego, and what I
wanna do with this podcast is make it as
interactive as possible, going through a
few areas of lessons, insights, learnings,
plus some reflection questions for you.
And then hopefully that you can play
this either back with your team, or
you can do this with yourself and your
executive leadership team, or just, like,
literally planning on your own where you
can leverage AI from a strategic point of
view, not just a technology point of view.
Through this, I'm gonna be sharing
something that I built for myself, an
AI COO harness, where essentially I've
solved problems for my own company, and
that has been useful for other people.
And I, I'm only doing that for the
reason to give you a real example thatâ¦
so that you can reflect and see where
you're at in your business, in your
growth, so that you can apply what you
need to and disregard what you d- what
you don't need, uh, in your business.
Then I'm gonna dive into ingredients
so that you can see what makes up
the soup of what we're, what I'm
using here with AI, so that you can
identify what you need to be using.
And then lastly, real use cases
or just real examples that I
currently do for the business.
Simple use cases and a little bit more
complex ones so that you can kinda
get an idea of how to package your
IP, how to package your execution and
make your workflows more streamlined.
And then I'm gonna finish with
some reflection questions.
Cool.
So the title of the workshop is Work
and Grow at the Speed of Your Thoughts.
So my goal for you here is to
become empowered by the current
state where you can leverage AI
in your company so that you can
actually own it and not just rent it.
And I also want you to think about AI
as a strategy from multi dimensions,
from ethics to ownership to pricing
to development and net new value that
you can create for your business.
And my goal here is to give you the
ingredients, so we're gonna dive into
the core ingredients that you'll need.
We also dive into what AI
really does for your business.
Then I'm gonna give you some live
examples so that you can reflect
and just see, just inside, peek
inside without looking at the code.
And then I also wanna dive into bigger
questions, like how does this change your
pricing strategy and your business model?
And I'll give you a couple
reflection questions to end with.
So I built something for myself
where I call it the AI COO.
It's a simple harness that I built
to solve problems in my company.
I needed to solve my own issues with
strategy development, with offer
creation, with go-to-market, with
sales, CRM, operations, content.
These are things that I have to do
anyways, so I was able to package these
things together to support me through a
chief orchestrator and sub-agents and a
memory layer and a context layer to help
me literally work at the speed of my
thoughts, to be able to execute, grow, and
help focus with clients where I need to be
focused, where AI and technology can't be.
Where I'm focusing on belief and
transcending the future and figuring
out what can we create together?
Where do you wanna take your
business, and how can I support you
and your team in that shared belief
to co-create that future together?
AI's gonna help you get to a point.
It's not gonna help you actually do and
fulfill everything, especially holding
that same belief with your clients.
Regardless of what you're doing
for your clients, you are holding a
shared belief, and you're being held
responsible for execution of that belief
as well as execution and performance.
So AI is not gonna take the blame.
You will take the blame or the
responsibility or the reward.
So I knew what to build for my own
company, thankfully, because I've
had experience growing multiple
teams globally and remote, and I
know what teams I need to build.
I know what roles, what execution.
This doesn't mean I don't have people.
I have assistants helping me.
Uh, and if you're watching this,
you've been amazing supporting me in
helping this podcast come to life.
But there's also other areas
that I know that it's like admin
work or busy work that actually
doesn't help a human flourish.
So I, I know what to build, I know
what needs to happen, thankfully,
from my experience, so I'm
compounding that here for you.
And let's dive into some
of the core ingredients.
So in my belief, I'm a big
proponent of AI sovereignty,
which means I want you to own it.
So what are the-- what
are we actually owning?
In a simple mental model, I want
you to think about three components.
Right now you have your
data You have your own data.
My goal, my hope is
that you own your data.
You don't just rent it out
or give it away for free.
The second is the model.
Now, right now, you do
not own frontier models.
Right now we're renting them, which is
okay, but you can actually have them
hosted on your computers or on your own
servers, and you can run them locally and
post-train them with your own intellectual
property or your own whatever process
that you're doing, and it actually
doesn't have to touch the internet.
And then the final is the harness.
And I'm gonna double-click on the
harness in just a minute, but I wannaâ¦
Those are the three areas that I
want you to focus on, the data,
the model, and the harness.
Right now, everyone is thinking about,
"I need to pick the right model.
I need to see which one
is gonna work for me.
Uh, what is it gonna do for me?
What are the capabilities?
What are the benchmarks?"
That's okay, but that
doesn't really add value.
Like, a good model is
just a brilliant stranger.
I think the key thing that I want you
to think about is the model that you're
using paired with the right harness.
And the harness is essentially code around
the model to make it more effective.
There's an orchestration, there's the
skills that you use, there's the memory
layer, there's context and history,
different states, different periods
of being work- working with you,
interacting with you and your team.
So that all consolidates into a
harness so that you can actually
execute work at a higher quality with
lower inputs and without being super
sophisticated with your prompts.
My favorite analogy is for you to
pretend that you own a restaurant or that
think about your favorite restaurant.
If you think about it, the chef at
that restaurant isn't the restaurant.
And the analogy for AI, the chef
is the orchestrator, the one
that's orchestrating what needs
to happen, the ch- the head chef.
Then we have station chefs, some cutting
the vegetables, some making the soups,
some making the beef, some making the
chicken, others focusing on the desserts.
Those are the agents that are
working at the restaurant.
Then they have, they follow certain
recipes of how to cook what they're
doing, how to prepare the meals.
Those are the skills, specific
ways to prepare a meal curated
to that specific restaurant and
that specific audience demographic
and the brand of the restaurant.
Then there's finally health codes.
Like, it doesn't matter what the
chef believes, it can't illegally
p- operate if it wants to stay
in business for long term.
So there are specific codes or laws.
Those are your rules.
Those are your hooks.
And then finally, there's the chef's
notebook, history around what to do, which
agent to put where, which cook to put
where everything is gonna be organized.
That's the memory and the context layer.
And the cool thing is, and the, the,
the moral of the story is, you own the
restaurant And you set the health code,
and you tell exactly what you want to
cook, who you want to cook it for, how
you want to cook it, who's executing the
kitchen, how everything is organized.
You own it.
So the key ingredients that you need
are the agent, which is the worker;
a skill, which is the recipe of how
to execute the work; hooks, which is
just reflexes, like if this happens,
that happens, like if X then Y, like
different reflexes depending on what
you're doing, what your workflows are.
Memory, so that we have
everything stored in memory.
Not, not everything.
You do want things stored with
the agent or with the tool, but
memory is a good contextual layer.
Now, the, the interesting anecdote
here is that with memory, like humans,
our memory sometimes is fickle.
Sometimes AI does the same thing.
So you don't want to save everything in
memory, but you do want to leverage memory
so that you have a longer context window.
And then finally, the rules or the law.
What are the laws for your AI land?
Now, there's a few things
that we can talk about.
What I really want to
dive into are also values.
You can instill your own values and set of
beliefs within the system that you built.
Every product that we build, and
this is actually from Microsoft,
uh, Taylor Black was on my podcast,
and he mentioned that every product
that you build has an anthropology.
What that means is that the creator
of every product that we use
has a belief of what humans are.
And I know we're getting philosophical
here, but a lot of the conversations
with AI do tend to lean philosophical.
This stems from the, the principles
of philosophy, metaphysics.
What is reality?
Epistemology.
How do we talk about reality?
And then based on those
things, what is a human?
Which is anthropology.
Then how do we treat humans?
Which is ethics.
So those are the simple basic questions,
and I know there's much deeper, uh,
ways to talk about philosophy, but those
are very simple practical approaches
for you to think about, "Okay, cool.
What is your definition of a human?"
When you're working with your team,
what is your definition of a human?
It's important for you to answer that
because then when you describe the
ethics of the AI system, it'll operate
based on your definition of a human.
So for me, the system or the values that
I use are based around human flourishing,
where I believe, I hope you believe
too, that humans are very great and that
we can empower them to do even greater
work, and that they should be in control
and not be subservient to a technology.
So I have nine components.
I'll give you an idea of the set of
values that I've instilled in the system.
So these aren't just prompted in.
These are hard rules.
Like these are laws that regardless of
what the agent thinks or the orchestrator
thinks, it has to do these things.
For example, the dignity of the person.
I believe that humans have dignity.
I also believe in honesty and
truth, so the AI has to tell me
the truth and cite its sources.
I also believe because I'm in the
relationship game and I'm in a business
where I do services and relationships
is the name of the game, I believe
that we have to serve over exploit.
Sometimes you can get an answer or you
can cut corners and exploit the other
person- And still get, still get ahead.
But that's not the name of the game,
at least the way that I play the game.
So I focus on service over exploitation.
I also have no manipulation.
We don't focus on manipulating the
user, AKA me or your team, to working a
certain way or thinking a certain way.
We also have the idea of the
common good, and we have to
define what is that common good?
So this is, again, more
strategic thinking.
We also have to have care and
respect, which is pretty obvious.
Give the human control so the
human stays in control, and we also
have to have the dignity of work.
We're not taking away work.
We're amplifying good work.
We're not just replacing
humans with technology.
We're amplifying our team to do
better, to do b- do more, to help
people go further, better, faster, and
or cheaper and more cost-effective.
And then we also have truth
over agreement, meaning I
want it to argue with me.
If I'm wrong, it should tell me it's
wrong, and it does so in a polite way.
But we do have arguments, and this is
good so that you can actually force
your own thinking or question your
thinking and see, am I missing something?
Not that you have the, the AI does
not direct you, say, "I'm always
right," but it'll argue with you.
It's not forced to argue, but I do
want to focus over truth versus,
"Yeah, Raul, that's just a great idea.
You're brilliant.
Oh my gosh, you're on a pedestal.
Why can't I be as smart as you?"
Which is what a lot of LLMs tend to do, to
just agree with you and say, "Hey, you're
the most brilliant human I ever spoke to."
And you know, we definitely
know that's not true.
All right, so- Those are key things that I
want you to think about, the value system.
Then we also want to think
about the memory system.
Why does it matter to have memory?
The reason we have memory is because
the more that you use your own system,
the better it can get over time.
It's like training.
Uh, it's not like training a human,
but it is like working with a really
good intern that evolves over time.
Another key thing that I want you to
keep in mind, and these are again,
part of the ingredients that I built
for myself so that you can ins- get
inspired or you can disregard what
you don't need, is having security.
Security for whatever is written into your
vaults, like your own internal database.
Also security running across the board,
like if you're installing something
from the outside or you're looking
at a repository or some, a set of
code from the internet or a plugin
where you don't just install things.
It has to go through a quality assurance
and an audit check to make sure that
this is legit, this is not malware.
And instead of actually downloading
or just plugging in a plugin, you're
auditing the code and only adding
the code manually with your agents
of what you actually need to evolve.
Most of these things, uh, at least for
some of these open source repositories,
uh, over a third are malware, and I
can think the number is even more now.
So just think about that for a second.
Um, and anything that you write or
publish needs to be approved by a
human because you're putting your
reputation out there when you're actually
putting your work out into the public.
So we have to abide by these things.
We can't just have AI just
do all the work for you.
So you have to have
security plus compliance.
Another key area is thinking about your
agents working under a one orchestrator.
Uh, now an orchestrator, you might
need a higher intelligent LLM.
It could be a frontier model, or it
could be a very huge open source model.
But the whole idea is to have a team
of specialists centered around an
orchestrator, and that orchestrator
works with you, AKA the human or
the user, so that you collaborate
with that thinking partner.
That's why I call it my AI COO.
And then from there you're able
to execute and take action.
So I have a series, I think a
12 to 14 or, um, uh, specialists
and one master orchestrator.
And these specialists focus on all
parts of the business from file
organizations to operations to sales.
It doesn't do the sales for me, it
just helps me with sales research
and CRM and logging and making
sure that I follow up with the
right person, et cetera, et cetera.
So like the content, I'm filming this, I'm
thinking and actually doing this for you.
It didn't write it for me so that
I can just read off a script.
A lot of YouTube videos or a lot
of content that we see nowadays are
people just reading off a script
that AI wrote, and I don't want that.
I actually want the human, I want the
raw organic insight, uh, because theâ¦
Anyways, we can dive into the
philosophy that AI has, has underst-
has a understanding, but it doesn't
have insight or it, it can't know
for certain what can be true.
It just can consolidate all the
information to present this might be
most probable truth because of its, uh,
assuming or inferring the next token.
Anyways, long story short, I have
a series of specialists around
one ma- master orchestrator.
And when you think about that,
when you look at your business,
okay, you have an orchestrator.
What are the key skill sets or
agents that you need to have
associated for your business?
If it's sales, marketing, content
design, client value, product security,
file organization, operations.
Those are some that I have.
What are the ones that you
need to have in your area?
So those are the core ingredients,
like the-- that's what makes the
AI soup work and taste delicious.
But what's possible?
What happens when you actually do this?
In my opinion, I think that AI
is not gonna replace your value,
and I'm gonna say that again.
I'm gonna say it carefully.
It's not gonna replace your value.
It's going to amplify it.
However, you can't amplify
crap I'll say that again.
You can't amplify crap.
That means that you need to be
focusing on where do you have the
most unfair advantage in the work
that you do, and you have the most
experience or nuanced information
that can help your clients go further?
Where should you be playing it?
You already know that answer.
You want AI to amplify that so that
it can delegate most of the work that
you shouldn't be doing, or that a
team can, shouldn't be doing, or that
your team can add more value, and
we'll talk about that in a second.
Now, in my opinion, there are
three layers really when it comes
to leveraging this technology.
The first one is automation.
That's pretty simple and obvious.
We've had automation even
since before I was born.
In the past, it was
deterministic automation.
Now it's agentic automation, which is
pretty sweet, but that's nothing new.
That's cool.
That's like level one.
Great, you can breathe.
What's level two?
Augmentation.
This is where things start really
exciting, where you're able to augment
your work, package your IP, and
literally take hours to deliver twenty
or thirty hours and deliver that same
value in maybe three or four hours.
That is not an exaggeration.
That's actually what I've been able to do.
And you might be thinking, some
might say, "Well, that's crap.
You might be producing crap."
No.
I'll tell you a story.
When I started using AI-- I
mean, I used it in '22, '23
like everyone else in the world.
I started using it with my work,
just seeing how it can support me.
In the past, if it's something, if a
document or a strategy or an idea took me
twenty to thirty hours to create, I still
took those twenty to thirty hours with AI.
I just was able to do sixty hours of
work in those twenty or thirty hours
because I could do more deep research,
more insight, more reviews, and it would
still be my strategy, my IP, my process.
Now, with augmentation, you're able to
leverage agents, specific agents with
skill sets, give them access to tools,
and you can now think, "Okay, cool.
That thing that we need to do, I want
you guys to-- or you, you set of agents,
like a swarm of agents to do that.
I want these ones to research here.
I want these to get leads over there.
I want these to look at the market
over there, and then I'll focus on this
strategy with the client and see what
we can do to orchestrate that together."
Augmentation, literally twenty,
thirty hours, maybe forty, fifty hours
condensed into three to five hours so
that you can focus on highest value.
That is how you leverage your IP,
agents, and skills to augment.
And this is where it becomes interesting
because a lot of people think
about, "Oh, I'm just gonna augment,
and all that time saving I'm just
gonna take, take it to the bank."
That's great.
You can increase margin.
Fantastic.
Others have no vision and they're
thinking, "Okay, if we can-- we don't need
these team members, let's just fire them."
And that's the wrong thinking because
when you buy back that time and your
team has more availability to think
You wanna create net new value.
You wanna focus what is the value
that I could be creating for my
clients that I'm not creating that
I wish I could create, or that I
wish I could help them go further?
Only if you want to, though.
This is a choice.
This is a significant choice.
I've had a friend who runs an agency,
and he was mentioning that he could solve
all these problems with the clients.
He just doesn't want to.
He wants to focus on his
line, and great for him.
That is a choice.
That is a strategic choice,
but a choice nonetheless.
So looking at where are you creating
value for your clients, how can you help
them go to done, finished faster at a
higher quality or better, overall better?
Or could you help them go further,
or could you help them go wider?
This is theâ¦
This is not, um, uh, like
a cookie cutter solution.
Every single business model is unique.
Most businesses I know are 95% the
same, just like the human skeleton is
95% to other, the same of other humans.
It's that 5% of your business
model that is unique.
Who do you serve?
What are your unique advantages?
Who are your key people?
What is your go-to-market?
What are their actual
needs and pain points?
H- what is your positioning, your pricing?
That all determines how you can
add more value, increase or add new
products or services, or just help
your users with a greater experience.
Like I know an, uh, an app, a consumer
app that has about a $100 million
run rate that their customer service,
they made a the strategic decision
to not fire anyone because of AI.
That's a pretty bold decision.
You can't just say that and not
walk away from doing that, okay?
'Cause they have stakeholders and
shareholders that want a return.
But anyways, this app, a consumer app, I
think it's like 60 bucks a year, I don't
know if they raised prices, but it's
60 bucks a year, $100 million run rate,
did not fire the customer support team.
Leveraged AI.
Now their customer support team can work
with team members, but what they do now
is they do outbound calls to their users.
And I know it's like, "Raul,
that's a simple exâ¦"
It's a true example.
It's a simple example, but they're
adding net new value for seeing
if they can reduce churn, increase
LTV, increase product satisfaction,
increase net promoter score.
Whatever they need to-- Whatever
metric they wanna increase, they're
looking at what do our users
need, and how can we support them?
That is a very simple step.
It's not super complicated, but it
is a strategic step, not firing,
leveraging that new time for outbound.
For other clients, like for clients
that I've personally worked with,
it's creating new offerings to
help their clients go further.
It's condensing time for some of
their clients that are focusing
on education and/or coaching.
They're looking at how can I give
them three weeks within one session?
And not because they wanna cram
information, but increase the quality of
the experience so that the learning can
actually stick versus having to leverage
repetition for learnings to stick.
So those are just some examples,
and I know that's notâ¦
A lot of people always ask
me for deeper examples.
It's hard to give a deeper example simply
because I don't know your business model.
But I want you to think about your
business model, your positioning, your
pricing, your customer segments, who
do you serve, and how do you add value?
How do you measure that value?
And how can you go deeper?
How can you go further,
or how can you go wider?
And that would help you with your
mental models to figure out where
am I le- leaving net new value?
I got the margins.
I got more time.
Great.
I have augmentation.
Do I just go surf all day?
Sure, you can.
But where else can you add
value for your clients?
That is a q- the question that
I want you to leave with here.
Um, a few examples that I have from,
from my own AI CO, what it helps me do.
It has everything prepared
for me on, in a day.
So let's just go through a few examples
of what I've built for myself just
to show you, again, simple tasters.
These aren't super sophisticated.
I'll give you maybe one
example that is pretty cool.
But these f- five examples are just
ideas of what I'm doing for myself to
help me operate better and literally
work at the speed of my thoughts.
Um, everything that I have is local
to my computer, but then it's also
cloud related, so I can be working
mobily as if I was working on my
computer, create presentations on the
fly, create ideas on the fly, execute
things, have my day pre-planned for
me, uh, without having to useâ¦
And I don't use, um, what is it?
The daily planner.
I don't use the harnesses like
Codex or what is it, Cowork, to
be able to have my daily planner.
This is all custom brewed for
myself, and I've tested this also
with open source runtimes and open
source LLMs, which is fantastic.
Uh, but I have everything briefed for me.
I know the agendas that I'm gonna have
each day, the goal, who I'm speaking
to, their CRM card, insights, and
just nuanced information that would be
helpful for me to be more present, to
add more value to the other person, if
it's a client or a prospect or a new
relationship that I'm working with.
The other thing that I have
is, um, first for some of the
content, so this is a podcast.
You'll see this.
This is a video.
This is also a Substack, and
this is also an audio podcast.
But there's other areas where we're
able to cut clips or syndicate
those clips around YouTube Shorts.
That takes a lot of time.
I have an agent that literally takes the
YouTube video, cuts it up into clips,
writes amazing headlines already, like
I would write the headlines, and all
I have to do is approve, and then it
just distributes and posts it for me.
And I really like that because that
used to take a whole lot of time.
Now it really doesn't have
to take that much time.
Now I have to be a curator, and
I know this is cliché, but I'm a
curator of taste, and hopefully,
uh, my taste improves over time.
The other thing too that I, that I have
for myself is I have almost like an intra-
like an intranet of things for all of my
content around clients, my work, what I'm
doing, and it's all indexed using an open
source from Toby from Shopify, his QMD.
So everything is indexed locally, so I
can have a database of searching, "Hey,
what did, what was that conversation I
had with X so and so three weeks ago?"
Around like halfway through
the call, we said something
specific about this presentation.
We actually had this
happen live with a client.
Uh, we were on a call.
She was looking at a presentation she
was gonna make in the UK, and she was
figuring out, "Hey, we talked about
the second half of that presentation.
What was that thing?"
I just went through the indexing.
It asked the agent to go find it, and
then when three minutes we had-- within
three minutes, we had exactly, "Oh, here
is the last bit that we missed out."
And then we were able to close that
off, and she was able to do a, a
phenomenal presentation in the UK.
So it's just a simple example,
but I can index everything in my
business and retrieve it, so it
knows context about me, business, and
the goals that I have specifically.
This is why it's important, and I
referenced this earlier, for security
and making sure that you have the
right settings and the right security
set up because you are leveraging your
information, your business information.
Another thing that I built, this
is more on the net new value
side, is like a lead engine.
Most of my clients will be
looking at, "Okay, cool.
Here's who we're going to target.
Here's our go-to-market.
Who do we talk to at stakeholders?"
I'll give you a simple example.
I had a, a global manufacturing client.
They were selling about five to $20
million contracts, and they were
looking at specifically, who do
we target based on companies that
want to move their manufacturing,
uh, nearshore or here in the US?
And from there, because I've done this
by hand manually, I was able to have a
12-step process that I leverage and have
agents now following, I think it's either
eight or seven skills to identify not only
the accounts to break into, but also why
we need to actually talk to these people.
Who at these accounts we need to talk to?
What is the relevance?
Who should we talk to?
What is the research of each person?
And by the way, here's their name,
email, and LinkedIn, and in some
cases, phone number if you need to.
And sales teams were able to take this
and leverage, build real relationships,
so you don't just mass send outbound
'cause these are larger deals.
Uh, build real relationships and
go all the way to the board and
have conversations about how they
can move their manufacturing either
nearshore or here in the States.
And that's just an example, but I was
able to leverage a swarm of agents that
I built from actually doing this by hand.
So- That's just an example here for you.
Other examples that I wanna give you
is quality assurance, is whatever you
ship, I would recommend creating a loop.
It could be three or five loops for
it to go through quality assurance to
make sure that it matches what you see
and qualify as that is a good quality.
That way, you don't ship crap.
That way, when you create something,
you're able to have AI infer and
continue to loop itself until it reaches
the quality minimum that you have.
And then from there, you still
have to review, and you can send
it back to quality assurance.
The other thing that I like to have, and
I would recommend everyone have this,
is a, a chain of thought process where
you have a chief agent manage a worker
agent manage an actual, like, grunt
agent for whatever task you need to do.
I call it the chain.
So I tell it, if we're gonna do an
execution or a job or a big project or
planning, I need it to go through the
chain where there's the chief, like
the chief orchestrator leverages the
specialists, and those are domain experts,
and they collaborate together with the,
the end, like the grunt agents, and
they all have to grade their homework.
We don't trust their word.
We review their output, grade their
output, move it up the chain, and the
chief orchestrator tells you specifically,
"Here is what they came up with.
Here's where they didn't agree.
Here's where they did agree.
Here's the conditional approval.
Here's my idea.
Where do you wanna go?"
So it gives me the power to decide we're
not gonna do this, we are gonna do that.
And the chain, and I call it the
chain, the chain is a process
to help you do-- make better
decisions, but it also has like aâ¦
Think of it as a committee or a good,
uh, set of counselors that are reviewing
based on your standards, your goals,
your orchestration, your business
values, and how you like to treat humans.
And I think that's important that
you stay at the center of control.
So how does this impact your business?
That's cool.
I built this, whatever.
How does this impact your business?
I think value is moving up the stack.
So I wrote about this in the services
stack, that we have five different levels
of execution in service businesses.
We have the act- execution itself,
where you're actually doing the work.
We have template-based strategies,
where a lot of, uh, people who
are certified are just following
someone else's strategy or following
another person's idea or a framework.
That's what I call template
strategy, so following someone else's
framework to produce an outcome.
Then we have judgment-driven strategy.
This is where you take your insights
and your experience, and you're
able to make a true judgment.
You don't just leverage AI with,
quote-unquote, "thinking and judgment" on.
You still have to make a decision
because your judgment, your decision
has real-world implications.
Then there's finally transformation
and accountability, and that stems with
the top of the stack, which is belief.
The key thing, so you can know what
to do, then you can actually transform
and hold someone else accountable
to doing what they need to do.
But that only happens because at the
top of the stack is a shared belief of
what the future can be like, what the
future can look like, and probably most
likely shared values within that belief.
That is human.
That is uniquely human.
I think all the value that we can
do as a service business is hold
the top of belief as much as we can.
This isn't some
psychological manipulation.
This is more about here's the potentiality
of what the future can look like.
Let's make it real.
Let's actualize it.
But that belief has to be in the minds
of two humans, of imagining, visualizing,
seeing that future, and holding each
other accountable to that future,
making strategic decisions together and
leveraging technology, AI on execution and
some like template strategies or whatever
you wanna execute to help you move forward
to realize that potential new future.
That's where value is moving.
So how does this work
with, like, your pricing?
So I'll tell you a quick story, and
if you've been a follower of the
pod, you might remember this story.
Had a, a gentleman from an agency tell
me that the team that he was working
with, the agency, was doing discounts
because AI was doing some of the work.
It's because, you know, they didn't
think of a better pricing strategy,
said, "I'm gonna give you an AI discount.
So usually my price is
$X thousand a month.
Because AI is helping us do it,
I'm gonna give you a discount."
Now, that is somewhat ridiculous because
clients, the end clients, don't care
what it takes you to get the job done.
They care that the job is done correctly.
And if they're looking at costs and
overhead and labor, they just say,
"Oh, if you can get it done cheaper,
just give us that savings too."
But instead, you're not focused--
You're just literally giving away
your learnings, your intellectual
property, and you're discounting your,
like, the new stuff that you built.
The key thing is to focus on what
are the problems that you're solving.
If you're solving very simple
problems with AI, then yeah,
you're gonna get discounted.
But you need to point those tokens and
your energy to a higher level problem
which is harder to solve because
clients don't care how you solve it.
They care is the job done.
The best clients care about, the
best clients care about two things.
Number one, what's next?
Number two, is it done correctly?
Now what's next?
What's next?
What's next?
And if you can focus on increasing
the value of your work and focus on
anchoring towards the outcome, not just
anchoring, here's the cost of goods sold
or here's my labor and overhead, that
you don't wanna give an AI discount.
You don't wanna discount your innovation.
This is happening for real now.
A lot of firms are getting, uh, at least
services firms are getting asked to
lower their fees because of that, because
pricing in the past has been ridiculous.
Pricing has just been, "Here are my
hours," or, "Here's how many staff
members I'm gonna put on," or, "Here's
how many," like, "the hours you get
from my retainer every single month."
Like, that way of pricing is going
to get shot down because that
pricing just counts on costs.
It doesn't count on value.
It doesn't look at, "Here's
what we're creating for you.
Here's what you're gonna be able to do."
Because most of that, they're not
pointing their energy and their
tokens to high-value problems.
So because of that, I want you to think
about your pricing with what I call
the net new value quadrant, which are
four ways to price in the agentic era.
The first one we already discussed.
The first one is the margin play.
You can either cash in that margin and
just take it to the bank, and that's okay.
That's a margin play.
A lot of companies are doing the margin
play, and we'll talk about it in just
a second, because of the time to market
and the speed of how things are going.
It's not a wrong play.
I'm just telling you
that is a margin play.
The next play is a m- is a volume
play, where you charge based on
how much your clients consume the
service that you're providing.
So it's a volume play, so the
more that they consume, the
more that they can charge.
And they can also get price breaks
eventually if they exceed a certain
number, but it's based on consumption.
The third is the IP play, which we talked
about in augmentation, where you're
able to take your methodology, and that
methodology becomes your operating system.
Your agents execute it, and the client
either access, accesses your operating
system, or you execute the work
that you do on behalf of the client,
leveraging your operating system,
creating, like, a play with, with IP.
So you're just focusing, you're
getting access to the IP, or
we're executing on this IP, or
we're pricing based off the value.
And again, that's different from
just saying, "We're gonna, uh, give
you 20 workers, and we're gonna
give you X amount of hours per week,
and this is how much we're gonna
bill you," 'cause our hourly rate.
That way it opens the door for
negotiation, saying, "Well,
that's your hourly rate.
I know you're using AI.
Discount your hourly rate."
It's just the wrong way
of thinking about it.
You have to definitely leverage
an operating system and
deliver that at value at scale.
I've shared this in a past podcast, but I
think it's important to revisit it here.
I had a call yesterday, literally, uh,
with a new web design agency, and what
they're doing is they're pricing their
services like a software, which is great.
It's a cool competitive amo- uh, way.
It's a simple, uh, deposit, then a
simple monthly flat retainer, and
they do specific things for you,
and it feels like I'm buying a
software, but it's actually a service.
That's cool, but their retainers,
like the top-tier retainer is anywhere
from like $1,000 to $3,000 a month.
Okay, that's great.
Compare that to a friend of mine who
is delivering value, like a service,
like an agency, but because the way
that they deliver that value is through
concierge support, they're able to
charge $10,000 to $40,000 a month.
Leveraging AI, leveraging technology,
different positioning, different way
of creating value, and different way
of leveraging their IP to deliver
that value to their customer.
That's the IP play.
It isn't to charge more
because they're abusive.
It's because they're solving a much
more important problem, and they're
l- they're leveraging a human layer
of concierge support to help that
problem get resolved and help their
end clients get that peace of mind.
Completely different business model,
but a simple, almost similar underlying
leveraging of the technology.
Just a different approach, different
positioning, different pricing.
Now, the final quadrant for
pricing is the value capture.
So in the past, we could do
performance-based pricing,
which is n- not new.
That's been around forever.
However, now with AI and the, what we
just talked about, your agents, you're
now able to influence performance
more than you could in the past.
In the past, you were just hopeful
that the client could get that,
that milestone or do, or close that
sale or do whatever they need to
do with the work that you're doing.
Now, if you focus on that new value,
you can create an ecosystem where the
outcome increases the probability of
happening, so you can influence it.
And if you're able to influence it
in such a way where it favors both
parties, you can charge a base fee
plus co-risking, meaning that you
actually capture the value that you
create with the client in agreement
with them in an easy way to measure.
So those are the four ways that I see
pricing changing in the agentic era.
The margin play, the volume
consumption play, the IP play, and
the value capture play, where you're
co-risking with clients for the net
new future that you want to create.
Now, how you measure success
in your business might change.
I think a new KPI that we're gonna
be focusing on is specifically for,
uh, product teams or for fulfillment
teams, is the speed to value.
How quickly can we influence and
create real value for our clients
because you're leveraging AI?
In the past, sometimes even for services,
it could be they might not see value
until month, month one or month three.
Now the key is how can we get to real
value either faster, more effective,
or get them quick wins in a way that is
actually meaningful, not just saying,
"Hey, we did this thing for you.
That's a quick win," where it's not
really meaningful to the client.
Like, they don't really care about that.
It's not calling it a quick win.
It's actually measuring that quick win.
And the window, in my opinion,
for new technology usually is
about six, maybe seven years.
At least that's what I remember
and experienced in the past.
Uh, but here I think the
window's accelerating.
No one knows the timeframe.
I certainly don't know.
If I were to predict it, it could
probably be 18 months to three years.
Uh, this is whyâ¦
That, the 18 months, by the way, started,
I think, last summer So the, uh, last
summer this is in August of 2026, so 2025.
So just a heads up.
Uh, the reason I say that is that
the advantage that you have now
to reach augmentation and net
new value won't be open forever.
Markets do catch up.
People get savvy.
So just thinking that, hey, you just
need to hire your, your AI person and
dev team and just have them roll and
see what they can do is not the play.
You need to get your hands dirty.
You need to go in there, roll up your
sleeves, and actually leverage the
technology, have your AI moment, and
identify where are we missing value?
Where can we create more value
for our clients, our shareholders,
our stakeholders, our team, so
that we can go further together?
Now, I'm gonna leave you with
two reflection questions.
This is something that you
can work on with your team.
This is, again, a podcast, but I'm
not there with you, so this is gonna
be a workshop style where I'm gonna
leave you with these two questions.
The first question is: What do
you wish you could do, but aren't?
What do you wish you could do
but are not currently doing?
Put aside every desire and
every vision that you have.
Just go specifically.
You probably, if you're in the
business and operating, you
know where the bottlenecks are.
You know what you would
like to solve first.
Based on what you learned today,
what would you leverage AI to solve
immediately based on what you wish
you could do but you aren't doing?
Is there a certain client that
you wanna help go further?
Is there a certain workflow that you
know is broken but you never got to?
Are there new ways or new products
or services that you wanna integrate,
or new markets you wanna expand into?
What do you wish you could do but aren't?
And the ingredients that I want you
to use here, just think about it.
If you're leveraging AI for
this, 'cause not all problems
are gonna be solved with AI.
So again, you gotta make that decision.
Is this AI solvable, yes or no?
But the key ingredients go back
to your data, the LLM that you're
gonna use, and the harness.
And the harness are the agents, the
memory context, and the orchestrator
You gotta think about that together
to figure out what do you wish
you could do but you aren't doing.
To go further, when you think about
data, you have to think about compliance.
You have to think about privacy.
Same thing when you think about the LLM.
Am I uploading all of my customers'
data to the internet, to the cloud?
You gotta think about that.
You shouldn't be doing that
The second question, and this is
more on the business side, is where
will your business model evolve
or will it be forced to change?
Artificial intelligence is
a deflationary technology.
It's deflationary on inputs
On value creation, it's the opposite
Where do you see your pricing, your
value creation, your go-to market, your
customer support, your segments, your
channels, your key customer relationships,
your revenue structure, which is not
just pricing, but it's also other ways
that you create revenue from upsells,
cross-sells, back-end offers, whatever
you have, your cost structure, key
partners, how is that gonna change?
Those are the components
of a business model.
How are those areas going to change?
And how are you gonna focus
on, again, value proposition?
Maybe not changing the value prop,
but going deeper on the value prop.
Those are the master two questions that
I wanna leave you with if you work with
your team or you just consider this
with yourself, with your AI agents.
I also have more resources that I'll
put in the show notes where I'll
give you the actual build manual.
I open source how I build my own harness.
So I'll give you the actual link to that.
It'll be dogoodwork.io/aicobuildmanual.
I'll give you the link to that so you
can actually work with your agent to
build literally what I described earlier.
Separately, I have more podcasts and
Substacks around how your business
model will change with AI on creating
that new value, on the new value
quadrant, and on the services stack.
So those are more strategic podcasts and
Substacks for you to dive into so that
you can answer those master two questions.
And as always, feel free to
reach out to me, DM, email, and
until next time, do good work.