I think cowork will be the

place where most business

people will park it and say.

"Wow." Well,

hello and welcome to another edition

of the eCommerce Evolution Podcast.

I'm your host, Brett

Curry, CEO of OMG Commerce.

And today I have a returning

guest, a multi-time guest,

maybe like fourth or fifth time.

My good friend for a long time,

fellow Missouri resident, Russ Hinnaberry.

And for those who don't know, Russ,

Russ led some teams at Digital Marketer,

used to help run the Traffic

and Conversion or TNC Summit

back in its glory days, which that

was just a rich, fun time- RIP. ...

in this industry. RIP

to TNC, it's so true.

But you and I met in 2010 or something

at a marketing conference in St. Louis,

Missouri, The Lou, And

really connected then.

But more recently, you are

the founder of the Click.ai.

You're also the co-author of

Digital Marketing for Dummies.

And when I have AI questions,

when I want to know what are people doing

with AI inside of agencies and inside

of marketing orgs, I talked to Russ

Henneberry. And so with that, Russ,

welcome back to the show. And how's

it going? Real good, man. How are you?

Dude, I'm doing good. Doing good.

Just feel like every

day is going to unlock

something new on the

AI front and exciting,

disorienting, scary, but

mostly exciting. And so-.

I mean, I didn't know how bored I was

with marketing until we got like this.

Yeah. Yeah. This technology

has reinvigorated me for sure.

Yeah.

And it's also one of those interesting

things where obviously AI is progressing

very rapidly. I heard someone say on a

podcast just yesterday, they were like,

"No,

2026 is going to be the year of the most

rapid disruption ever on AI." And I'm

like, okay. Yeah.

And then there'll be more disruption

this year than all the AI years past or

whatever. So buckle up,

which is fun for sure.

So I want to dive into a few

things. I want to talk into

agentic AI and kind of how things are

flowing and then just some of the latest

news. But also you're plugged into

agencies, marketing orgs, brands,

and you're seeing how they're using

AI and what you're doing with AI.

But I think maybe the

first place to start,

because I know this is something you

were just absolutely bullish on as MI,

and that's Claude cowork.

And so it's just absolutely

ripping right now.

All my business friends are using it.

I just started testing it

actually just this past Saturday,

started tinkering with it and holy

cow. So for those that don't know,

what is Claude Cowork? And then

let's dive into some applications.

Well, so there's a few

things about Cloud Cowork.

We've probably all used

regular ChatGPT, regular Cloud,

regular Gemini or whatever. The

new thing about Cloud Cowork,

so it's a desktop app right

now only available on Mac.

And the big three differences

I think to think about with

Cowork is first that it

is far more agentic. So

it makes plans and

unfolds those plans right

in front of your eyes.

So you can ask it far more ambitious

for far more ambitious tasks

than you could with a regular chatbot.

And the reason it's able to pull this off

from watching it work is that it works

with your local files.

So it can create files,

it can delete files, it

can move files around.

And one of the most important types

of files that it produces is called a

skill.md file or a skill file.

And that skill file,

you can think of it like you

would think of a custom GPT

or a Gemini Gem,

but it's just a set of instructions

as to how you want something executed.

And here's the crazy part about this is

that when you're working inside a cloud

cowork and you do something,

or you can even describe

something that you want done,

it can build that skill

file and then just save it

into your directory.

So it starts to organize

this entire set of folders

and files.

And one quick tip on this is that when

you do go to start to play with this,

go slow and just kind of think

through how you want that

folder set up, that structure set up.

Because if you've worked with something

like ChatGPT before and you create a

custom GPT that does X, like let's say

it writes hooks for ads or something.

So it's a custom GPT that's

really good at writing ad hooks.

Maybe it's personalized for your business

and your persona and all these things.

So it's an excellent GPT.

But the problem is with it is

that it's saved in the cloud and

if it's not performing properly, you

have to go back over there, edit it,

and you got to go in there, figure out

where it's messing up and change it.

Well, you don't do that in Claude cowork.

So if it produces an output

and you say, "You know what,

you're being way too hypey with those

hooks. Those hooks are way too hype.

I like this one because it's a little

more down to earth and that's a little

more on brand." It'll go back and edit

the skill. So it'll ask you for it.

Say,

"You want me to go and adjust the skill

so that I just nail this for you every

time?" And so these skills are

almost like self-healing, right?

They build themselves pretty much. They

ask you, "Do you want me to build a.

Skill?" So self-improvement

or recursing at most.

Yeah. Right. And it all is happening on

your local machine. In other words,

there's a version control

part of this as well.

So one of the things I've always

recommended with AI is that the very first

piece of information that you need to

feed any AI when you're doing anything

about marketing or business

growth is a persona document.

You need to know who you're talking to.

So it always boggles my mind

when people are like, "Oh,

my AI doesn't give me very good

output." And it's like, well,

does it know anything about who you're

trying to reach or who you're talking to,

who you're writing to,

who you're planning for?

The second document that

I always recommend is a

document that clearly outlines

your offer. So what do you sell? What's

the cost? Do you have a guarantee?

What are the deliverables, et cetera.

Those two pieces of information,

and most of us have kind

of caught onto that,

that we could feed the AI this and

we feed it that and it gets much,

much better. The problem has been that

there's a version control problem.

So you build a GPT and you attach

these persona and offer over here,

and then you build another GPT,

you got to attach the second one and

the third one. But in Claude Cowork,

since you're working from a file system

and it's basically just plugging a brain

onto Claude cowork,

there's a single persona.md

file, a markdown file,

or it could be any, I guess it would

be a text file if you wanted it to be,

but Claude Bill's markdown files that

you can adjust that one place.

And anytime it needs that persona

or it needs that offer or it needs

this skill or it needs this

spreadsheet or it needs whatever,

it just goes and finds

it. It's very, very,

very good at understanding when it

needs a particular piece of context

that's located somewhere in your brain.

It's amazing. Amazing.

I think it's still a little bit trippy

for people and they're still maybe not

fully wrapping their heads

around it. But I mean,

Claudco work can really become your

personal assistant in a lot of ways,

your research assistant, your

marketing assistant, your copywriter,

all those things. But more than just a

chatbot, it's just like doing things,

right? It's just running things.

And so maybe you could talk through some

specific examples, like either where,

how have you used Claude Cowork?

I'll explain what I was experimenting

with this weekend as well,

but how have you used it?

What are some of the best use cases

you've seen from agencies and brands?

What can this do for us?

Right. So one of the other

things about Claude is that,

and I think this is the

Cloud ecosystem in general,

is how good they've gotten at

connecting to external sources.

So for example, I've

connected my notion to it,

and so Claude Cohort can just

go fetch something out of ...

So for example, when I

start a Zoom meeting now,

my Notion starts to transcribe

those notes using Notions AI,

which is cool. But at the same

time, do I do anything with that?

Do I do much with it?

And so my workflow now though that

I've got Cloud connected to Notion is,

I jump on a Zoomie. I did a webinar today,

a training today with

some people and 90 minutes

on ChatGPT projects and I was

going through that and then

Notion's transcribing,

and then at the end I can run a skill

through my Claude cohort that just says,

"Go grab that transcript

and do X to it. " So

I think that's really the unlock here,

is to understand that we have to have

sources of material.

Where is the source of some idea,

some data, some source,

and how can we plug that in and

then what do we want to happen?

And if we know those two things,

what do we have to put into the AI and

then what do we want to happen from

there? Even if we can just describe it,

the AI will figure it out from

there and then it'll ask you,

"Do you want me to just build a skill

that just does this every time?" So you

talked about, before we jumped on,

like doing things with your

financials and stuff like that.

So dropping some source

material into your claude

brain, if you will, spreadsheets,

et cetera, and then pointing Claude

cowork at it and saying like,

"I want you to transform this into

whatever." It could be charge graphs,

insights, whatever. And

then at the end saying,

"Write that up as a skill

or even a full routine."

So I mentioned about trying to get a

little more ambitious with what we're

asking.

You. What's the reason

why a skill and a routine?

Can you talk about that real quickly?

Well,

a skill would be the technical

term for it inside of Claude,

but I do think about these things

as routines. So for example,

when I arrive at my desktop,

I just open Claude co-work and I say good

morning. And that triggers a routine,

but it's really a skill. So

it's written into a skill file,

but I think of it as a routine

because what it does is it greets me,

it puts my manifesto out, which is like

this thing that I like to read each day.

And so it's got steps and then the next

step pulls my weather because I'm always

wanting to know what the weather is when

I'm sitting inside at my desk all day.

Isn't that funny, but Will, but

we both live in Missouri, right?

So typical Midwest,

it's like it's going to be 70

degrees a few days from now.

It was like negative five a

couple weeks ago. It's just crazy.

But like it matters. We're sitting

in the AC or the heat, so we're fine,

but we still want to know. But yeah,

you want to kind of run this skill

or this routine, right? So yeah,

what else does it do

for you in the morning?

It goes and grabs everything out of

my calendar and displays that for me.

And then it pulls top, I don't know,

five headlines off of several

sources about things that I'm

interested in and pulls them

in and gets my news and stuff.

But not that this is

tremendously groundbreaking.

What it is is it's a routine. It's

something that I told, do this, then this,

then this, then this and this, right?

And it just writes it into a skill file,

sits it into my clog brain there.

And then anytime I say good morning.

So the same deal with my newsletter.

So I write a newsletter each week.

It's quite involved. It

has several parts to it.

And I used to have to have like 15

GPTs going and like different deep

research prompts that I had to keep

copying and paste. Instead now,

once I ran through the process

one time, the way I wanted it,

I just said, "Create skills for that.

Create skills that format this into this,

create skills that do this deep research."

And so now it's just more or less

like, "Hey, I'm building the

newsletter." It's like, cool,

where do you want to start? And

it's like, let's start with this.

And it goes out and does the research.

And the thing is you can spin up

multiple tasks at the same time.

So it's like it goes off, does the

deep research, open a new task,

start something else up.

And the way Claude is built,

the way Cloud cowork is built

is that it can work in parallel.

So it does things a heck of a lot faster

than you would think it would be able

to do something because

it'll spin up four, five,

six agents at the same time. Each one,

this might get a little technical,

but each one has its own context

window. So in other words,

this one's out there doing this.

It's literally like running

five clouds at the same time. Right.

Which this is actually important.

So let's talk about Context

Window a little bit,

because I first heard about this on

the Andrew Ferris podcast recently,

but what I think a lot of people don't

know is if you give Claude a really big

file or maybe like a transcript

from a really long call,

it's not necessarily

crawling all of that, right?

It's maybe looking at the end and the

beginning and maybe summarizing some

things.

And if you give Claude a whole bunch

of stuff like all in one prompt or

something,

it's going to take shortcuts potentially

instead, But having multiple agents,

you can have a lot more context

that you're feeding the AI.

Hey, thanks again for tuning in. This

episode's brought to you by OMG Commerce.

That's my agency. Hey,

we're specialists at creating

omnichannel growth for brands

profitably. Now, the greatest brands

we know are no longer just D2C.

Yes, they're masters of D2C,

but they're also growing and scaling

on marketplaces and in retail stores.

And we understand the complexities of

how to grow in all of those channels from

a campaign strategy, a creative strategy,

and a measurement strategy. In fact,

we recently won a Google Agency Excellence

Award for helping Arctic coolers

grow their retail sales

in Walmart using YouTube.

We've helped add almost eight

figures in growth on Amazon for

brands, and we've even helped a

brand go from nine to 10 figures.

And so we want to help you grow.

So if you're not satisfied with your

growth in any of those channels or you're

looking to unlock new growth,

we should probably chat.

Visit us at omgcommerce.com.

Click that Let's Talk button.

We love to schedule a strategy session

with you. With that, back to the show.

Yeah. And the thing is, right

now we have Opus 4.6. We've got

Gemini three or whatever. We've

got ... These models are all very,

very intelligent,

but CloudCowork really isn't a

breakthrough in intelligence as much as a

breakthrough in sort of

architecture of how the tool works.

It's not that the tool is

that much smarter, although

it's a little bit smarter.

These tools get incrementally

smarter every couple of months.

They release something that's smarter,

but it's the UI that's different,

right?

And it's sort of the what's under the

hood that's different about Claude

Cowork. And it does take a little bit of

getting the hang of how it

works, but once you get going,

it's actually super good at kind

of walking you through like,

"Do you want me to do this?

Do you want me to do that?

" And so if you're going to

play around with CloudCowork,

I would just say start with a simple

use case and just start to type

and watch it start to build something

out for you. It's pretty amazing.

The other thing that's really interesting

about these skills is that they're

extremely shareable. So for example,

I've built out an entire workflow

around building my newsletter and in my

membership, I'm just going to give

it to my people. So it's like,

here's a zip file.

It's got all the skills in it and

all the context in it that it needs.

And so just upload it, zip it, upload it,

and now you have that

skill. So Pretty cool too,

and there's little marketplaces

springing up that are

thousands of skills, like anything you

could think of that are already there,

you just download the zip, Zip

uploaded inside of Claude as a new

skill. And I think I'd be a little.

Careful about that. It'd be good to

go with the song there. Yeah. Be a.

Little careful about downloading other

people's skills because people do stupid

stuff with the instructions

and stuff like that. But

it's a different way of

working with AI that I think OpenAI

sort of dropped the ball with not

updating how custom GPTs work all this

time.

And now skills have come along and

cloud coworks come along and the two

together, it's a

combination that's hard to.

Beat. It's a winning combo. Yeah.

So a couple of things I want to unpack

and I'll kind of talk through a little

bit of what I did this weekend and where

I think this is going to unlock some

pretty cool stuff for my agency.

You talked about connections.

And so basically what I wanted to do,

we got this very detailed financial

dashboard that's got everything in there,

client revenue, cost of employees,

cost of various costs of

different service items.

We kind of group our expenses

into delivery or all the team,

all the tools that deliver

services into growth.

So sales and marketing expenses and

tools and payroll and then ops, right?

So all the tools and overhead and

employees and stuff that fit there.

But I wanted to analyze some

things. And so this is kind of hard,

like what do we dump into

what spreadsheet or whatever?

And so basically I gave

Claude, I was like, "Hey,

this is a framework that I want to

work within. Here's some of our goals.

Net revenue retention is a number we're

going to start tracking regularly." We

did this in the past,

but the calculations are actually kind

of difficult and building that on a

spreadsheet is also kind

of a pain of the butt.

But basically that's where you're looking

at starting revenue for a beginning

period of time. And so we just

take the beginning of the year,

what's our starting revenue?

Over time then minus any churn,

so logo churn or clients

a churn minus contraction.

So maybe a client didn't

churn, but they reduced scope,

so now they're spending less.

So it's beginning revenue minus those

two things plus expansion, meaning, yeah,

but some clients actually add the scope

and actually do more work with us.

And so then what is that

that's net revenue retention?

Basically I started like, "Hey,

Claude, this is what I want to do.

" Cloud cowork. It's

like, "Oh, cool. Well,

do you want me to connect with

QuickBooks so I can connect directly to

QuickBooks?" I'm like, "Well,

why don't you just look at this Google

Sheet first?" And then it's like, "Oh,

this is a gold mine of information." And

so then it starts spinning stuff out.

And then I started talking about some of

the sales goals and stuff and looked at

our sales pipeline and the sales goal

sheet that I put together and I started to

say, "Hey, this is good.

Here's where you maybe have

some weaknesses." And so

spit out these different

analyses and I'm like, "Holy crap,

this is awesome." And so right now I've

just got to looking at our Google Sheets

because you can do the browser plugin

where it looks at the Google sheets and

can read it,

or you can upload a Min Excel file

or you can plug it into QuickBooks.

So there's different

ways you can run this.

And so yeah,

it's going to be a real

unlock for financial insights

because we don't have a huge finance

team. And so it's going to be very,

very powerful. Yeah.

Yeah. I mean,

I think it's a good bet

that every day that goes

by,

the people doing work at computers are

going to spend a little bit less time

outside of an AI tool

than the day before. So

as AI tools like Claude

Cowork keep developing

connections to other outside tools and

they've already got that protocol in

place called MCP and everybody's pretty

much on board with it and starting to

connect everything together,

it'll just become sort of a push

and pull type of a situation

where I'm able to just

pull this from here.

And the nice thing is that you got

this sort of central terminal with

intelligence inside of it,

and you can pull from disparate

sources and pull things together and

either create new things or get

new analysis out of those connected

things. And what it does is

it's opening up things that

you never would have had time

to do without these tools. So

a small company now can do things

that large companies have been able to

do for a long time because they

have a whole team of financial,

you know what I'm saying?

Yeah. Yeah. You have a

whole finance department.

To go pull this out.

A department there that's running all

this analysis and doing all this stuff,

but

because they got to pull from all this

stuff and it takes like human effort to

get through all that. But what

I would encourage you to do,

and you probably did do,

is as you're working

through something manually,

you can always pause

and stop and say, "Okay,

I got this kind of how I

wanted it sort of manually,

but even using the AI, so

it's not completely manual,

but you understand you're just kind of

prompting through and you get to this

sort of end state and you're like,

that's what I wanted." That's when

it's a good time to pause and reverse

engineer into a skill or

even if you're using ChatGPT,

you reverse into a custom GPT, say,

"I want you to take a look at everything

we did here in the end output and I

want you to codify this and create

the instructions to get here

like that. " And the nice

thing about cowork is it knows

if you said something like,

"I want to run the financial analysis

again," or give it some other kind of

name, it'll know to go grab that skill-.

Shows what to do.

... and then it has a lot more autonomy.

It'll go and run several steps up ahead

and it might hit a point where it's

like, I need some input back from Brett,

and so it's going to come back and say,

"Do you want me to do this this way?"

And you just select that and sometimes

it'll write that into the skill

so it doesn't need to ask again.

And so it's really quite

intuitive in that way.

And it's just a massive step

change from what we're used to with

regular just chatbots.

Totally. And then I want to get your

take on OpenClaw in a second as well,

and then a few other things.

But you also said something that

because tools like Cowork and

there'll be others, I'm sure,

they'll connect to almost anything.

And over time,

they're going to connect to basically

every piece of SaaS you use,

every other piece of tool, whatever.

There is a realistic future where

most of the interaction we have

on our desktops or on our phones is

with an AI and it's the one connecting

to all the tools and pulling things

together, doing what we want it to do.

And so yeah, just really,

really fascinating.

What's your take on OpenClaude for

those that have been following the news?

There's a tool that's had what, like

three or five names in the last week.

It was Claude, C-L-A-W-D,

bot. Then Claude,

who we've been talking about

was like, "Don't do that.

That sounds just like our name." And

then it was Molt bot and then now I think

they landed on OpenClaw.

Who knows by someone else to

this maybe something different,

but what's your take on that?

Because that has been wild.

It's been such a big deal.

Even my 23-year-old son

who's in the roofing business

in sales, he bought

an Apple Mac Pro or whatever,

a Mac Pro and he's running it on a local

machine and it's doing some business

development stuff for him. But

what's your take on OpenClaw?

Where is this taking us?

Well, for the people listening to this,

I would probably be listening to this,

right? I mean, the person who's

hardcore and you want pure control

and you want all that stuff,

you're probably not

listening to this show.

But if you're a normal business person,

you're running a company or

you're in marketing or whatever,

I think that there are essentially

sort of four steps here

that we can think about. So

you've got your sort of chatbot.

So regular ChatGPT 5.2 or whatever,

regular Claude still available,

that's level one, right?

Level two would be to kick up the cowork.

And I think by the end of this year,

everybody will be in the

working knowledge workspace,

working on a computer, you'll be

working in something like cowork.

It may not be cloud cowork, but

it'd be something like it. Yeah.

Google's version of it or

OpenAIs, version or whatever.

Then the third level would be something

like ClaudeCode or over at OpenAI,

it's called Codex, where you're

going straight to the tap,

you're kicking past any

sort of UI user interface

and you're just going

straight to the source. And

you're already starting to get into

pretty hardcore when you're doing that

because you're working in terminal and

you're having to use

some coding languages and stuff

like that, but it's doable.

If you really committed to it-.

It's vibe coding, right? So I mean,

you don't have to have a ton of

programming knowledge, but maybe some,

or you're understanding

prompts in a different way.

Could the average person just jump into

Claude code or something similar or

that's going to take a little bit of.

Work? Yeah, I think you could.

It would be a longer learning

curve than Claude Cowork.

Because what Cloud cowork is,

is it's Claude code with a UI

sitting on top of it. Yeah, I got it.

And what happens when you do that

is it puts some restrictions on you.

You're not going straight to the source

where it's just like you can do anything

in here

because you've got the restrictions of

that sort of harness that's sitting over

the top of Claude cohort.

So going straight to Claude

code is like level three.

And then you get into the open

source stuff where your son

wants to buy an extra computer

because he doesn't want that

open source

sort of no guardrails AI

to be on his own machine,

you want to create a.

Combined space. We're

pointing into his email.

He's creating a separate email and a

separate browser and a separate machine,

but- Yeah, people are.

Daisy chaining 10, 200

Macs together and creating armies

of employees that don't exist

with Slack accounts and

email accounts and all these things.

I would say the average person,

level four is, forget about it.

It's so powerful and it really is.

It's not that it's not powerful,

but it's precisely because

it's so powerful that you

shouldn't rush to install

it.

Just pump the brakes a minute because

what's going to happen is some

more secure company is

going to release something.

And Claude Code is already just insanely

powerful if you just go straight to

Claude code and you still have

a lot of guardrails there.

The thing that I would.

Point out though- There's all kinds of

unlocks, all kinds of stuff you can do,

a whole world open up to you in those

levels one through three that you don't

need, or one through four, you

don't need to go open claw just yet.

And I think

cowork will be The place where

most business people will park

it and say, "Wow, this is quite a.

Lot of power." And just so you

know, we're not sponsored by Claude.

We get no kickback from co-work.

We're just like, this is awesome.

We're geeking out about it using

it. And so it's phenomenal. Yeah.

The one thing I wanted

to point out though,

the other comment about

OpenClaw that it raised is that,

so in the summer I started to

realize I was doing so much

hefty work with AI tools

and running Zoom and

other things that my laptop was a beast,

but it wasn't able to keep up. And

I do think that we as business people

need to be thinking about our own

tech the way that a lab thinks

about how many GPUs they have.

Starting to think about what is the

processing power of your company,

of any individual in your company.

So I went ahead and bought

a pretty hefty Mac Mini

because I was like,

"I don't want to be constricted by

my computer." So that's kind of

interesting that you start to think about

your own ... I mean,

you always think about you don't want

your computer to be slowing you down,

but at this point,

it's sort of like you're going

to see Apple just pick up a

giant windfall from like- Yeah. Well.

Their stock is already up just because

so many people are buying Mac Minis.

Yeah. All the hardware that's needed.

That Apple silicon is what

everybody's after. Yeah,

it's wild to think about

somebody like your son who's young

and starting out in a business,

having essentially desks

with nobody sitting at them,

but there's people working.

And he just ended up, he's in

roofing sales and so he's like, "Dad,

I'm going to reach out to all these

insurance agents and I want to get them an

email here and then I'm going to drop

by their office and get them donuts.

I'm going to do these things.

And then I want this to be able to respond

via email over here and pull together

my calendar and I've got these Notion

apps." I'm like, "Love it. " I'm like,

"This is awesome.

This is great." And also I'm glad you're

doing it on that machine because I'm

not ready for Open Claw, not for

OMG. Heck no. But I want to see it.

And yeah, I'm all in on

coworking and figuring that out.

And so one quick thing though,

using AI,

still the ROI can be amazing

and it's low cost compared to

what it should be based

on the power of the tool.

But we start using CoWork

or OpenCall or whatever,

you can start to, you're

spending more on these tools.

This is not necessarily the $20

a month subscription, right?

What are you seeing since

you're leaning hard into cowork?

What has that done to your monthly fees

and your usage and what kind of plans do

you have to be on to make this work?

Well, I use the $100 a

month max plan, but the fact

they do have cowork now

available at that $20 a month,

you probably will find

that heavy users of it.

They're going to need that

$100 a month plan. But I think

if it's being used right, it's just

a no-brainer for 100 bucks a month.

A month for an assistant.

Okay. Yeah, plus.

Yeah. I think that with the proper

setup and structure ... See,

I've always said that these

AI tools are extremely

useful,

but even going back to my last time

on this show where I was talking so

much about context,

And I mentioned that a little bit today

about how if you're not giving it any

context about your offer, your persona,

don't be surprised if you're just

not getting anything out of it.

And that sort of blows up when you

start to think about how something like

Cowork can access really any context when

it thinks it needs it and it can find

it. It can find it in your machine.

So when you go back to your

example around the finances,

when you talked about your sales goals

and all these different documents,

I know how organized you guys

are at OMG. It's impressive.

And you guys document a lot of things

and you have goals and you have rocks and

you have quarterly and yearly

plans and all that stuff.

And that kind of stuff all can be

brought into your Claude cowork space.

And the way to do this, honestly,

is when you're playing with it, is to ask.

Take something like a

quarterly plan or something,

drop it in and say,

"Here's the quarterly plan,

where do you think this should be

most properly structured inside.

Of.

Claude?" And

before you go and decide where

it goes and where to put it,

ask it to think it through. So to go

analyze that file and start to connect it,

well, and I see this skill over here

where you run a financial analysis,

I see this skill here,

and maybe the best thing would be to

organize it this way and it'll ask you,

"Does that sound good?".

Yeah.

And it starts to build out ...

There's a file at the very

central core of Claude called the

Claude MD file, claude.mdfile,

which is basically,

you could put anything in there, but

what you do is you just tell Claude,

keep updating that Claude MD file

because what it's doing is it's telling

Claude, how does this structure

work? It's sort of like

a central plan as to how

Claude is supposed to

interact with your ...

So anytime you use Claude and it can,

it will access that claud.md file. So

that's the thing is that

the more you give it, and

there's definitely an unreasonable

amount of stuff you could give it,

but things like that, it's like, "Well,

that's really something that I consult

my own brain when I work on my finances."

You might consider,

should I drop this into the Claude Cowork

space so that it has access to that?

And you should be shocked at like, wow,

it realized it needed to go look at

that and then go grab the skill and then

connect over here to HubSpot and then

... You know what I mean? Right, right.

And.

That's what I mean when I say

agentic, right? So agentic.

It's amazing. Let's do this.

I know you talked about a couple workflows

that you built for content creation,

and I think this maybe is

around your newsletter,

but do you want to actually share that?

We had to talk through it for those

people that are just listening and not

watching, but would that be

worthwhile to kind of dig into? Yeah.

I'm curious while you're doing

that, are you using Gemini at all?

And are you leaning into

Gemini gems or Gemini as a.

Tool? Yeah. Yeah. So I use Gemini

almost purely for image generation. Oh.

Nice.

Yeah.

A banana.

That nano banana model is still-

It's pretty insane. ... the thing.

But yeah, let me run

through this workflow.

Maybe it'd be helpful to a

lot of people to see how ...

Is that going full screen for you? Yeah.

Totally see it. Yep. So it's full

screen. So for those watching on YouTube,

they'll see it. For those listening,

we'll do our best to describe it and

make it come to life in your mind's eye.

Yeah.

Well.

So this is really about

creating any content or copy

of any kind. It could be

text, images, video, audio.

Any kind of content is ...

I started to realize, as you know,

if I had to be pegged to any one

digital marketing discipline,

it would probably be content marketing.

And it's the biggest reason why I hopped

on AI so fast because I saw it creating

content and I was like, wow, this

thing can actually create things. So

as the years have gone

by, I mean using this,

I'm getting better and better and

better at creating content and copy

with AI tools.

And I started to realize there's

this underlying process that

constantly gets used over and over and

over again, and it's really this source,

structure, format,

polished, sort of steps.

And the way it works is that

the first step is the source.

So if I'm going to create a piece of

copy, like a sales page, an email, an ad,

or a content like a video or

anything, it doesn't matter.

There's got to be some source.

And most people that are failing with

content creation, copy creation with AI,

they start with the AI and

they start to say, "Well,

come up with the idea and then write

the idea or come up with the idea." And

so it's like, well, where

are you in this scenario?

Where is your voice? Where

is your brand? Where are you?

And so

when we think about setting up any

sort of workflow to create something,

the first step is to figure out, well,

where am I going to get

this source material from?

And that source material can either

come from you, in other words,

your brain or you're from

internally in your organization,

or it could come from outside.

It's the same as before, before AI.

Yep.

So from a standpoint of you,

you might rant into

your phone or something,

some idea that you have, or you might,

maybe you do podcast interviews

like this one, right?

Or maybe you shoot videos

or maybe you do webinars

or you write or something.

There needs to be ...

So if it's going to come

from you, at some point,

some ideas got to come out of your

brain and essentially your mouth.

If it's going to come from

elsewhere, that's great too.

You can go and grab source

material from all over the ...

I've said many times that the web

doesn't necessarily need more content.

There is a great service to be

done in curation of content, right?

Like this is good. I read 10 articles.

Here's the one you should read.

And so you can go and

curate from elsewhere,

but you've got to have something

that the AI is starting with. Now,

step two is to structure

that in some way. So

the fact is, if you go and

rant into your phone an idea,

that's a great way to create a source,

but it's not in any

structure that you can do.

Something- Not usable, not valuable,

not shareable really. It's just brain.

Dog. Yeah. So typically a

workflow might look like ...

First I get this raw input, let's

say it's a rant into the phone,

and then I take that and I put

some kind of structure around it.

The two best ones, in my opinion,

are a set of bullets or into a table.

And so you just tell the

AI, "Hey, take this input,

whether it's a rant, a transcript,

somebody else's YouTube video,

somebody else's article, it

doesn't matter if it came from you,

but it's got to have a source and

then take that and organize it in this

structure." And that's a

second pass with an AI.

So the first pass is

ingest this or go get this

source. The second pass with the AI is,

now organize it in some structured format.

Then the third is to

literally format thing into

whatever structure you

want, that piece of content,

whatever's going to

come out the other side.

So if I have a rant into a phone

that's just me rambling and rambling,

rambling as I'm driving up to the gym

or something and I save that and then I

drop it into the A.

The first pass is to ingest

it and then take it and

organize it into a table

or into a set of bullets.

And then the third pass might be,

"And here's how to take that and

organize it into a LinkedIn post

for me. " That's a third pass

with an AI. So it's not one pass.

It needs separate instructions

at each step along the way.

And then that last step is to polish.

So bring it up to publish quality,

and that could be a

combination of you and the AI,

or it could be just you,

or it could be just AI for those of us

that are wanting to automate our lives

completely away and just are publishing

things to LinkedIn and elsewhere that

...

But I have found that this structure

reoccurs no matter what I'm

building, is that I need to figure

out where's the content coming from,

where's the idea? How do I

organize it in a way that's useful?

How do I then transform

it into whatever ...

Do I want a script to come out

of that? Well, that's fine.

Do I want a written post?

Do I want a cartoon?

Do I want a slide for a

presentation or a set of slides?

Anything is possible. I take the source,

I give it some structure to

organize it so I can look at it.

Oftentimes there's some curation

that happens here, by the way.

You get it into a table and you're

like, just take ideas three, 10, and 12,

and then move it on to step three,

which is, how do I take this? I mean,

you almost think about this just a set

of rules that you keep moving things

through.

And then that last step could be

add a brand voice to it using AI

or do like we have always done. Type.

Actually.

Edit the thing. Wait a minute.

We're in there typing and editing.

What's up with that? What is this?

So what does this look like then?

Is this maybe a four step process just

inside of our favorite chat interface?

Are we pulling something together

that's a little more agentic?

What are you recommending here?

Well, and that's going to depend

on the tool and the actual process,

but take a look at this. This is just

how I put my newsletter together here. So

the step one is deep research.

That's a great place to go get a

whole bunch of source material.

So you tell an AI, "Hey, I want you to

go out there." This is for a newsletter,

this example.

So I want you to go out there and

find the 10 best stories about X.

They're most popular. I

want you to check Twitter.

I want you to check here and it comes

back and I want you to structure that.

As- Do you have a favorite tool

there? So for deep research,

I know they're all getting

better. They're all good,

but what are you using there, John?

I use ChatGPT for that,

but now with cowork,

it doesn't ever call it deep research.

It just goes into research mode.

You know how in the past it's been like,

I am now entering deep

research. You know what I mean?

When you select it, I want you

to do deep research. Go. Yeah.

Yeah. No, I mean, they sort

of evaporated that because

it's just over there doing things

and it'll just enter into more

of a deep research mode where it might

be gone for 10 minutes. Like I said,

I spin up another task and work on

something else. But in this case,

I use deep research. You can use any

tool pretty much has deep research now.

You create a prompt and you could

create a prompt and save that prompt if

something you use over and

over again, have it come back.

And my big thing about deep research

is that it usually brings you back this

report that's like, "Okay,

I guess there's my afternoon

to read this freaking report.".

Take four hours to read that.

So what I'll do is I'll tell the AI,

"Don't bring me back this big long report,

structure it into a table."

I love tables, right?

I love organizing- Scannable.

... especially things that come.

Back. Easy to digest. Yeah.

Yeah. So imagine in this

scenario that I have 10 stories

from various news outlets, not my ideas,

this is other people's ideas that I went

and had the AI go and grab together.

And then this next step represents

a set of rules as to how

this particular story needs to be

transformed. So I just pull up,

I use Beehive for my newsletter and

you can see this is a story that I

ran last week, put this

Jerry Seinfeld gift,

OpenAI is putting ads in ChatGPT

because GPUs don't pay for themselves.

And there's a structure

to how this story works.

It's the facts and then why I

think this matters to my audience.

And so all I do is create

another pass of the AI that says,

"Go in here,

grab whichever story that Russ

wants." So Russ curated story three,

let's say, and apply

this set of rules to it.

And that set of rules then

spits out on the other side a

formatted piece of content.

And then that last step is to polish it,

edit it, get it right, check the links,

fact check the story in this case,

right? It's always different,

but the same deep

research is used to create

another block of my newsletter called,

it's actually called By the Numbers,

but I used to call it Stat of the Week.

And all I do is I take a story in here,

let's say Story five in the list

has a good stat in and I'm like,

"That's good to share." And I say, "Okay,

run Stat of the Week

rules on Story five and it

outputs a different output."

Does that make sense?

Totally, totally. Yeah. And then it

ultimately ends. We're telling it then,

take these ideas, run this process.

It's the source structure.

What was the third one?

Format. And.

Polish it. And then the third

idea here is the tool of the week.

So if I find something that's like,

okay, that's a bad, nice tool there,

I've played with it, I want to bring

it up. So I've got a structure on that.

So it's the same set of research,

three different outputs.

And the difference is

that this set of rules is

different for each of those structures.

And if you think about

this from a standpoint of

building any content or copy,

what is the story? It doesn't have

to be deep research. Of course,

it could be a rant, it could be a

transcript, it could be a YouTube video,

it could be any place that we generate

ideas, then take that,

get it into some format.

And again, I almost always recommend

bullets or tables because from there,

once you get that stuff into a

table, you can then say, "Okay,

apply these rules and

apply these rules." Now,

if I take this one step further and

go into cowork and how cowork works,

how cowork works, the way ChatGPT,

let's first start with

how ChatGPT still works,

would be if I want to format it

this way and this way and this way,

I would need three different GPTs,

I would have to stop and I would have

to call that GPT or copy and paste a

prompt. Instead, what I do is I say,

take story one, turn it into a

news story format, take story four,

turn it inot of the week, take story

six and turn it into tool a week,

and it's got all those skills saved in

there and it just goes and runs and it's

just like ... And

if I get to the end and I'm like,

"I really don't like how you're

writing those headlines for the news

story,

I want you to change it so that it works

this way." And then it'll just say,

"Cool, I changed that and do you want

me to update the skill?" And it's like,

yeah, so that next time

I run it, it's done.

So it's sort of that self-healing

idea that we talked about.

Earlier. I love it. I love

it. This is fantastic.

I could spend another hour going

through this stuff. Actually,

I would like to because I've got

like a million questions for you now.

I guess we are running out of time.

So I guess the next thing is I got

to have you back for the 10th time or

whatever it is. So we'll for sure do that.

But where can people find more? So you've

got a community called the Click.ai.

Talk to us a little bit about that.

And then also your great follow

on LinkedIn and other places.

So talk about that as well.

Yeah. You can always find me on

LinkedIn and even message me over there.

I'm almost on LinkedIn all day long.

The Click.ai has two components to it.

It has a membership for individuals.

And then I do team

training and implementation

for teams. So both of those things

you can find on the Click.ai,

the membership is really about using

AI to do business work. These tools,

they're general purpose technologies.

They're like electricity or something.

We have the electricities everywhere

and it's used for dang near everything.

So AI is the same, right?

It wasn't built for business and it wasn't

built for science and it wasn't built

for education. It was

built for all those things.

And there's no instruction manuals out

there for each one of these massive

things that we can do.

And this is a big time

tech transformation.

And so the membership is there for

people that are trying to figure out,

how do I use these tools

to do real business work?

And I do think that we are entering in,

you mentioned that the 2026

is going to be a big year.

I do think this Cloud

cowork jump is a big one.

And we are going to start

seeing a gap between both

individuals and companies

that adopt this stuff and

start to get more out of the same team.

Totally, totally.

It's just crazy.

And that's what I've

heard recently. It's like,

we shouldn't be fearful that AI is

coming for our jobs. Most people,

there are going to be some exceptions,

so don't want to downplay that.

But hearing people talk on

podcasts, the team members you have,

the team members that

really understand AI,

and maybe they can both vibe code

and use cloud coworker or whatever,

they become worth two or three

or four employees to you.

And so that improves someone's

marketability and someone's earning

potential. It doesn't

diminish it. And so yes,

the gap is going to be widening between

those that embrace this and Excel at it

and the companies that embrace it

and Excel and those that don't,

the gap is going to widen for sure.

Yeah. An interesting number right now

to look at is revenue per employee,

right?

Yeah. I love that number.

How big can your ...

You mentioned I was with digital marketer

and I was texting with Richard Linder

the other day,

and we were talking about how many people

would have taken for us to get DM to

wherever we got it was like

25 million or something.

And by the time we did that,

we had like 80 people and we're

sitting there thinking, man,

we could have done that with

four or five people maybe

with these.

Tools. It's crazy what's

possible now. Crazy.

Yeah.

Yeah, for sure. It's awesome,

man. Really appreciate it.

I do want to even jam with you one on

one because I've got some stuff we got to

talk about, maybe get you in for some

training. And so for those listening,

do that as well. Check out

the click.ai, hire Russ,

put these tools to work for your

business. With that, Russ, thanks,

man. Ton of fun. Looking forward to

the next time. Good to see you, buddy.

Absolutely. And as always, thank you for

tuning in. We'd love to hear from you,

leave us a review on iTunes or

wherever if you haven't done it. Also,

if you found this episode helpful,

share with somebody else you think

will benefit from it. And with that,

until next time, thank

you for listening. Hey,

as we wrap up this week's episode, I

want to mention, if you're a great brand,

if you're scaling high seven, eight,

nine figures in D2C or Omnichannel,

we should potentially talk.

We've worked with some of your favorite

brands and we'd love to consider working

with you as well. We are masters at

unlocking new channels like YouTube,

unlocking new scale on platforms like

Amazon where we can add up to eight

figures in new growth, and we've got

multiple ways we can work with you.

So we can do the full service thing and

work like a partner with your team and

really run everything, or

we can offer consulting.

So maybe you've got an internal

team that really knows their stuff,

but there's an area they don't know

really well and they'd like to get some

consulting. We can do that. We also have

tons of free guides, free resources,

free materials you can check out.

All of that gets started

at omgcommerce.com,

and we can't wait to help

you scale profitably.