Goldenhour 2024 29 min

Operationalizing AI for the Marketing Funnel


Learn how GTM teams can use AI right now in their workflows to drive more efficient outcomes.



0:00

Welcome to Future of AI.

0:02

We're going to talk about Operation Lacking AI.

0:05

And you know, if you're in this room and AI trust,

0:10

there's some of us that probably are AI skeptics,

0:12

some that are AI survivors.

0:14

What we typically hear is that most people are waiting for a day in the future

0:18

where AI can actually do things for that.

0:21

And the reality is that most people don't actually realize

0:24

what AI is capable of.

0:26

And you know, unfortunately, there's a big aspect of related education

0:31

in the lead curve that's been eroded, I think,

0:34

by people's first experience in the AI.

0:36

Your sales reps went to chat to be taken and said,

0:38

"Right, me, I see, I'll be now."

0:40

They probably didn't get a great result, right?

0:42

And that puts us in a place, right,

0:46

where it's really difficult to be able to win what that journey should look

0:51

like.

0:52

And so, you know, ultimately, the purpose of today's presentation

0:55

is not a sales page.

0:56

It's to help you think of how go-to-market teams can actually leverage AI

1:00

to drive all of the outcomes that they want,

1:03

and how this is really possible today.

1:06

So, before I jump too far ahead of it,

1:08

probably should introduce myself and my co-presenter, Nathan, over here.

1:12

Shakar saying I lead solutions at Copy AI.

1:14

Now, I know that you might think we have a lot in common

1:17

in that we support the sales team,

1:19

but solutions is actually a really unique role at Copy AI.

1:22

It actually encompasses customer success, too.

1:24

We think about org design very differently due to the investments that we're

1:27

making,

1:28

but, you know, in a sense, half my team is responsible for making empty

1:31

promises

1:32

to customers, and the other half of my team has to cash the check.

1:35

So, it's really important for us to stay in lockstep,

1:37

and, of course, I spend a good day of my, you know, my time

1:40

thinking about how we can leverage it, how our customers can leverage it,

1:42

how to make their dreams come true.

1:44

Hi, I am head of content strategy here at Copy AI.

1:50

I'll tell a little bit more about my background with AI in a minute.

1:53

The biggest thing, though, is I've walked around telling people I work in AI,

1:56

and I feel like public enemy number one at this conference.

1:58

So, just bear with me.

2:00

I'm happy to talk about any objections afterwards,

2:02

but we do have some really cool stuff that will walk you through

2:04

that can make marketing, I believe, cheaper, faster,

2:07

but also higher quality, and that last part is the most important, so.

2:10

Love it.

2:13

Uphill battles is what I fight best.

2:15

So, let's talk about Copy AI for a second.

2:18

Again, promise is not a sales pitch.

2:20

There's three numbers that I want to use to tell the story.

2:23

15 million, 37 and 40.

2:25

Now, the best part about the story is I actually don't even want to complete it

2:28

right now.

2:29

Just keep these numbers in the back of your head.

2:31

We'll talk about what they actually mean and represent later.

2:33

What's more relevant to this audience is the problem that we solve at Copy AI,

2:37

which is really all around what I think, you know, the mandate is in 2024.

2:42

Time and time again, we talk to dozens of go-to-market leaders,

2:46

and this comes up almost every single time.

2:48

The go-to-market team is simultaneously the most expensive

2:51

and the least efficient part of every single organization.

2:55

I see a lot of people nodding their heads because you've probably experienced

2:58

this

2:59

in your own day-to-day.

3:01

And, you know, whereas it's easy to point fingers at someone that, you know,

3:04

maybe isn't working as hard as they could, the reality is that's not really

3:07

what the problem is.

3:08

When you zoom out, there's hundreds and thousands of to-dos in typical

3:13

organizations,

3:14

and the team needs to get these things done every single day.

3:18

As things continue to evolve, we continue to layer on new pieces of the tech

3:21

stack.

3:22

They continue to add up and tasks begin to, you know, constantly pile up.

3:26

And the unfortunate reality for marketing teams is that

3:29

whereas sales is a little bit more discreet, and it's easy to say,

3:31

"Look at how many accounts I'm working."

3:33

It's really hard to say, "Hey, remember these initiatives that I've been

3:36

working on for three years that have been continuing on an ongoing basis?"

3:39

Yeah, I'm still responsible for doing all of that,

3:41

and now we're adding more layers to it.

3:43

And so, you know, at the end of the day, what we've really done is invest in a

3:48

lot of tools, amass these tools and say they're solving a problem for us.

3:51

When really we're creating more button-clicking work, more pain for the people

3:55

that have to use it day in and day out,

3:56

and we've built huge tech stacks and, you know, ultimately not really seeing

4:00

the results that would allow us to invest more.

4:03

This is the problem that the industry is calling Go to Market Bloat,

4:08

and it is the sum total of all of the investments that you've made in the Go to

4:11

Market engine.

4:13

And as I said, the big problem, right, is that they're not driving the results

4:16

that we had all hoped.

4:18

And, of course, you know, to add insult to injury, we're in a very different

4:23

place than we were a few years ago.

4:25

I thought Anthony actually put it best at Screen Capt. his post from a week ago

4:30

, right?

4:31

Ultimately, we're exhausted.

4:33

There's a lot that -- there's a lot of pain that's happened over the last few

4:37

years,

4:38

and a lot of change of the whiplash of the market correction that has really

4:43

pushed us to think differently about how we can actually drive the impact that we need.

4:48

And, again, this is to say that, you know, it's okay to be exhausted.

4:52

It's okay to feel like there has to be another way, because as we all discussed

4:55

this morning,

4:56

that is really what innovation is founded on.

5:00

And in hindsight, the zero interest rate phenomena is actually something that

5:05

shouldn't have ever been

5:07

a reality in the first place.

5:08

And I'll give you a simple example, right?

5:10

Who thinks that their head of engineering or engineering department would have

5:14

signed up for continuous

5:15

deployment six times a day in a world where they didn't have the ability to do

5:18

automated testing?

5:20

No, right?

5:21

Like, who would sign up for opening -- deploying the app and then having to pay

5:24

someone to go click on every

5:26

text box, every button, every dialogue modal, and determine that your latest

5:29

release didn't break it.

5:31

And so, in that world, technology kind of led us to decide to adopt a new way

5:36

of working.

5:37

The zero interest rate phenomena that we've all experienced has really just

5:39

said,

5:40

hey, data is important, it's an asset, you're a person, you can analyze it,

5:44

why don't you take on this task as well?

5:47

And so, what's really happened over the last several years is we're tired

5:50

because we've become the tools of our tools.

5:54

We are effectively data rich but very information poor.

5:58

And in a lot of instances, this actually drives us to not really be able to act

6:03

And when we do act, as we shared in the last session, right, it's really hard

6:06

to attribute your actions to

6:08

actually driving the impact that we wish is in the world.

6:10

And that is a glim-grim reality.

6:14

And, you know, I hope that I'm not belaboring the pain too much.

6:18

I think I took Anthony's message to heart this morning that it's important to

6:21

dwell on it so that we can actually

6:22

think about, you know, what good really looks like.

6:26

And then enter 2024, you know, again, not a surprise, do more with less is

6:29

literally the mandate.

6:30

And that line should actually be even more crooked, that's not a UI error, we

6:34

're getting less resources

6:35

and we need to actually drive more results.

6:39

But if we're spending all of our time keeping the lights on, how can we

6:42

possibly innovate in the long run?

6:45

How do we change the way that we go to market and how do we actually work

6:49

better together as a team?

6:51

So here's a hot take for you, right?

6:53

AI is supposed to solve this problem. I think we've all heard that before.

6:56

We've also probably heard this one.

6:58

AI is a calculator for words.

7:01

Anyone here this, believe it, maybe, like I'm hoping, you know, I'm not the

7:05

only person that's so deep in AI news that

7:08

this is, you know, an analogy that only makes sense to me.

7:11

But, you know, I hate to say this is actually one of my least favorite analog

7:15

ies for AI

7:16

because a hot take on is actually chat GPT is a calculator for words.

7:21

And that, don't get me wrong, right? There's a ton of value in having a

7:24

calculator.

7:25

If I didn't have a TA89, I probably couldn't graph anything other than basic

7:28

trig functions.

7:30

However, calculators are a tool designed to do a specific job, which, you know,

7:35

in this case is helping me solve

7:36

an ad hoc math problem. In the case of chat GPT is helping you to solve an ad

7:40

hoc work problem.

7:41

Calculators are great, they just don't run businesses.

7:44

And so when you hear AI is a solution in all of your problems and then contrast

7:47

that to a world in which you have to

7:49

talk to a chat bot for, you know, a couple hours in order to get something

7:51

basic done,

7:53

it's pretty challenging to be like, yes, it's going to drastically transform

7:56

the way that we go to work.

7:57

It'll make you 10, 20, 50% more efficient, but that's not really the outcome

8:01

that is going to help you do more with less.

8:04

And so, you know, a better take on this, we think, is that AI is much more than

8:08

a calculator.

8:09

It's actually a second, third, fourth brain that you can actually use in order

8:13

to be able to scale your own impact

8:16

across the organization. And ultimately, we have to get over this idea that the

8:21

only way to use AI is by

8:23

interacting with a chat bot back and forth. And, you know, unfortunately, this

8:26

is pretty common, right?

8:28

We just heard three, you know, amazing marketing leaders all say that chat GPT

8:32

is like the end-all-be-all of leveraging

8:34

AI. I'm sure we've all also used other AI tools out there. But the reality is

8:39

AI tools also don't really hit the

8:40

mark, right? In many ways, they have a, you know, buried logic. There's no way

8:44

to actually change the way

8:45

that you're using it. I'm sure a lot of the folks in here probably use GONG and

8:48

review GONG transcripts all the time.

8:50

GONG meeting summaries convince me that meeting summaries are valuable. That's

8:53

not actually what I'm looking for out of

8:55

a meeting. And so, there's this nexus of recognizing that AI is a second, third

8:59

, and fourth brain, but really it's

9:01

driven by your brain. And that is the foundation of actually how you can get AI

9:05

to work for you. And, you know,

9:07

hopefully that should be something that's exciting to everyone in the room. And

9:10

so the question, right, how can we use

9:13

AI systematically to actually scale our own impact? You know, what does it mean

9:17

to actually even do that?

9:19

How can we actually do that in a way that drives impact and results in

9:22

production and not just, you know,

9:24

do things a little bit faster? And so, I want to come back to the three numbers

9:28

I told you, told the copy AI story.

9:30

15 million customers in 37 months powered by 40 employees. Our marketing team

9:36

is four people. Four people, total.

9:39

The entire company is, you know, 40 employees. We're proud to call customers

9:45

like Siemens, Lenovo, Thermo Fisher,

9:47

Crow, Rubric, Urban, our largest enterprise customers. Obviously, we're a

9:50

premium platform, so this is counting

9:53

literally all of our customer base. What's interesting to know is that we don't

9:58

compete with Jasper on content.

9:59

We compete with HubSpot on content. And Jasper is a company that's, you know, I

10:03

would argue they're not our competitor,

10:05

and hopefully agree after we show you what our product can do. But, you know,

10:08

ultimately, that's very different

10:10

because we think differently in scarcity at Copy AI is really what drives

10:14

innovation. When I joined Copy AI a year and a half ago,

10:17

it was me and one other person that was responsible for every single customer

10:20

interaction from every part of the sales cycle

10:23

to actually deploying the customer, getting them live, training them, teaching

10:26

them how to use the platform.

10:27

And in that year, we closed 37 logos, which I'm pretty proud of. It was a lot

10:32

of work, but what we realized was all of the things

10:36

that we didn't have time for were an invitation to actually think differently

10:39

about how we could use our own platform.

10:41

And in many ways, that's really driven the entire story of Copy AI. If we can

10:45

unlock what it takes to get 100 times the value

10:48

out of each person and allow those people to continue being happy and thinking

10:52

about their next problem,

10:54

and ultimately leaving some of the load behind, we can really drive some really

10:58

incredible outcomes.

11:00

But what it takes to achieve this is shedding a lie that we've all been told.

11:05

There's one thing that you take from today's session. It's that you've all been

11:09

likely sold to lie if you believe that chat

11:12

GPT or chat modalities are the only way of thinking about using AI. And this is

11:17

controversial. I know a lot of people are afraid of, you know,

11:20

automation. But the reality is autopilots are greater than copilots. Autopilots

11:26

actually allow us to completely shed a

11:28

responsibility and leverage our own brain to come up with an idea that we can

11:32

encode into AI to follow our best practices

11:36

in order to be able to drive the outcomes that we want, repeatably, reliably at

11:41

scale.

11:42

And if you think about most marketing tasks, right, it's not like you need to

11:44

just write one blog post. It's not like you just

11:46

need to write one nurture email. It's not like you need to just write one sale

11:49

as the email that's personalized.

11:51

Practically, everything needs to happen at scale. But the critical part of this

11:54

is actually thinking about your role in it

11:57

and thinking about your ability to actually instruct AI to drive the outcomes

12:02

that you want and scale your own impact in an

12:05

organization. So I know that's a lot of waxing, right, we haven't even told you

12:09

how to pragmatically use AI.

12:11

I wanted to set this up so you have a good understanding of really where we're

12:14

coming from here. But really, what we'll

12:17

jump to is actually a live discussion and demo of exactly what this means and

12:21

how we at Copy AI are leveraging our own platform.

12:24

And so Nathan, I'd love to, you know, just have you answer a few questions for

12:29

the audience. Can you just tell us a bit about your

12:32

AI journey, so far? >> Yeah. And can you guys hear me without the microphone?

12:35

Is that okay? >> Can't record without.

12:37

Then I'll use the microphone. That's just fine. That's good. My AI journey has

12:41

not been a super technical one. I talked to some

12:44

of the engineers and the founders. They tell me about how they started coding

12:48

at eight years old after school for 12 hours

12:50

a day in the library. I was watching The Simpsons, if I'm being totally honest.

12:53

I was watching The Simpsons. I was eating junk

12:55

food. And that was kind of my childhood. That was my background. When I got

13:01

into marketing as an adult, all of that pop culture

13:04

really played off. It was great. But I don't have the technical stuff. I don't

13:07

code. That's just not who I am. So I've always been

13:10

kind of confined to buy how well I can use the tools, how technical those tools

13:14

are. When I came, I was at an agency

13:17

when I first started writing for Copy AI. I was writing the landing pages. And

13:20

at that point, it was really just the chat feature.

13:23

And the chat feature is very cool. We'll talk about that and how we use it. But

13:26

I was using it differently back then. How many people

13:27

are familiar with Copy AI? How many people have tried it? Okay. How many people

13:31

have tried chat GPT? Yeah.

13:33

You'd use chat about the same way. Our chat function. It uses the same language

13:36

model. So you'll get about the same quality results

13:38

out of that. Soon you'll be able to use different language models, which is

13:41

neat. But I was using it about the same way

13:43

you're probably using chat GPT now. You'd want to write a LinkedIn post. You'd

13:47

write a LinkedIn post. You'd want to write a blog post.

13:49

You'd want to write a blog post. I don't use chat so much for that anymore

13:52

since we got workflows. So at first, as I was coming into

13:57

AI, it was very technical. Chat, Copy AI broke that down. Made it accessible to

14:02

me. As Shakar said, Nathan, you are the perfect

14:05

simpleton to be presenting on why this is so easy to use. And it's true. Not a

14:10

big technical background. What I loved

14:15

the big shift in the company was once we took chat and transitioned into work

14:19

flows. When we went to B2B

14:21

and we became a GTMAI platform, not just a chat tool, but a GTMAI platform,

14:27

that was when the big shift was for me. Because what we did was have

14:30

workflows, which was different than anything I'd used before. And as a content

14:33

marketer, it really helped me out, scale, and keep quality

14:38

very high. That was important for me. So I'll tell you quickly about how I use

14:40

chat, and then I'll tell you about workflows and how those are

14:43

different. Now I use chat as like a little brainstorm buddy. It's really good

14:48

at bouncing ideas back. I don't expect them all to be

14:51

great ideas. But me as the human can go through with my creativity, look at the

14:54

ideas it's giving me as a launch pad and start

14:57

coming up with something better. So we got a CMO last November. His name's Kyle

15:01

Coleman. You might know him on LinkedIn. And we sat down, I went

15:04

in person to Denver. And he said, Nathan, I want to see how you make a content

15:08

calendar. I want to see what you do. So no pressure, new CMO, I'm

15:12

sitting down. And I told him, we'll do this together in real time. We'll make

15:15

six weeks worth of planned out content. We won't make the

15:18

content, but we'll make the plan. And we did it in about an hour. And the

15:21

process was simple with chat. I knew we wanted 20 SEO plays. I knew we

15:26

wanted 10 thought leadership pieces. So I asked for 70 ideas. I asked for way

15:30

more than I needed so that we could go through with our human insight

15:34

and pick out the ones that were most interesting. That probably took 20, 30, 40

15:38

minutes. Then I took what we agreed on. All the titles, the

15:41

topics, just general ideas that excited us. I'm not looking at keywords yet.

15:45

Looking at ideas that excite us. Then I could go through

15:49

and ask it for, hey, what are some likely keywords that would be good for this?

15:53

And I would get, I'd say, give me five or ten per topic. Then I could take

15:56

that, head over to AHRF's, check for any volume opportunities, and do all of

16:00

that grunt work, but really efficiently and really

16:03

quickly. This is very important because I'm going to share about what I do with

16:07

the time savings later. That part's really important. Sure,

16:10

you're saving time. But what are you doing with it when you get it back? It's

16:13

really important. So that's how I started using chat. At the

16:15

end of it, like I said, in about an hour, hour and a half, we had six weeks'

16:18

worth of content planned out, mapped out. He was excited

16:22

about that. All we had to do now was find the thought leaders for our thought

16:25

leadership articles and get these drafts over to the

16:28

content writers. So that's how we use chat. It was great for one-off tasks, ad-

16:32

hask tasks. It can give you some insight

16:34

via brainstorming, but awesome. Then we have workflows. And that's when we

16:37

became a platform. So workflows are different

16:40

than chat. The best way I can think to describe it is workflows create an end-

16:44

to-end process, and it completes the little

16:48

tasks along the way. So I'll give you an example. I promised myself I wouldn't

16:51

teach you how to use AI to write a blog post. I'm not

16:54

going to. You will have access to a workflow that's pretty cool that we built.

16:58

So we'll talk in a simple way in a second. But the

17:01

workflow breaks things down into separate actions is what they're called. And

17:07

like I said, it builds a process, a

17:09

process that you can run at scale. So for example, we're going to start with

17:13

the most simple form of workflow. It's not on the

17:16

screen just yet. This is about as simple as it gets. There we go. All right. So

17:22

we're going to start with something very simple. One of the things

17:23

that we were doing is coordinate LinkedIn posts across our leadership team. And

17:27

one thing that I was doing as I was

17:29

looking at these LinkedIn posts was we would try to draw attention back to an

17:33

article that that thought leader had written or as you'll see

17:36

collaborated on together. What I wanted to do was make those better because a

17:41

lot of the stuff -- a lot of the LinkedIn posts

17:43

that I said, hey, go check out this blog post. It was trying to put in way too

17:46

much information. Because one blog post might

17:48

have 10, 12 points in it. Each one a little nugget, at least they should. And I

17:52

think it's a little bit more

17:53

important. And I think it's a little bit more important. And I think it's a

17:59

little bit more important. And I think it's a little bit more important. And I think it's a little bit more important. And

18:03

I think it's a little bit more important. And I

18:04

think it's a little bit more important. And I think it's a little bit more

18:07

important. And I think it's a little bit more

18:10

important. And I think it's a little bit more important. And I think it's a

18:14

little bit more important. And I think it's a

18:17

little bit more important. And I think it's a little bit more important. And I

18:20

think it's a little bit more important. And I

18:23

think it's a little bit more important. And I think it's a little bit more

18:26

important. And I think it's a little bit more

18:28

important. And I think it's a little bit more important. And I think it's a

18:32

little bit more important. And I think it's

18:35

a little bit more important. And I think it's a little bit more important. And

18:39

I think it's a little bit more important. And I

18:41

think it's a little bit more important. And I think it's a little bit more

18:44

important. And I think it's a little bit more important. And I

18:46

think it's a little bit more important. And I think it's a little bit more

18:47

important. And I think it's a little bit more

18:49

important. And I think it's a little bit more important. And I think it's a

18:51

little bit more important. And I think it's a

18:52

little bit more important. And I think it's a little bit more important. And I

18:52

think it's a little bit more important. And I

18:53

think it's a little bit more important. And I think it's a little bit more important. And I think it's a little bit more important. And I think it's a little bit more important. And I think it's a

18:54

little bit more important. And I think it's a little bit

18:57

important. And I think it's a little bit more important. And I think it's a

18:58

little bit more important. And I think it's a little bit more

19:01

important. And I think it's a little bit more important. And I think it's a

19:05

little bit more important. And I think it's a little bit

19:08

more important. And I think it's a little bit more important. And I think it's

19:13

a little bit more important. And I think it's a

19:16

little bit more important. And I think it's a little bit more important. And I

19:20

think it's a little bit more important. And I think

19:23

it's a little bit more important. And I think it's a little bit more important.

19:27

And I think it's a little bit more important. And I think it's a little bit more important. And I think it's a little bit

19:29

more important. And I think it's a little bit

19:32

more important. And I think it's a little bit more important. And I think it's

19:35

a little bit more important. And I think it's a

19:37

little bit more important. And I think it's a little bit more important. And I

19:41

think it's a little bit more important. And I think

19:43

it's a little bit more important. And I think it's a little bit more important.

19:47

And I think it's a little bit more important. And I

19:48

think it's a little bit more important. And I think it's a little bit more

19:51

important. And I think it's a little bit more important. And I

19:54

think it's a little bit more important. And I think it's a little bit more

19:57

important. And I think it's a little bit more important. And

20:00

if we refine this and we get to the point where that's the best possible social

20:04

media content, I would imagine you go

20:06

from 10% edits to 5% edits to 3% edits to maybe 1% or QA review. And so if we

20:13

think about how we can make that even smarter, we

20:15

can actually connect it to another process that does this for us. It's going to

20:20

take the blog post, identify interesting

20:21

interactions, and then pipe it through the exact same workflow that we already

20:24

know delivers good final product. And this

20:27

ultimately allows us to get a whole lot more posts at once. And you can simply

20:31

QA and go through and post the ones

20:33

that are most interesting or potentially have the ones that are really going to

20:36

drive your audience at the right time. And

20:39

so what I'm trying to communicate here is that this is a very different way

20:42

about thinking, about utilizing AI. You

20:45

are the process architect. Your process is actually what has been driven

20:49

results. And as we all agreed today, that is

20:52

literally what we need to do best in the next, I don't know, call it five years

20:55

, to change the things that aren't working and

20:59

figure out the things that are working. And if you go back to the marketing

21:02

problem, right, unfortunately content has

21:04

always been the bottleneck or the production of that content has always been

21:07

the bottleneck. So what happens in a world where

21:10

you can actually scale that content production and actually think about it more

21:14

from a hypothesis testing paradigm, as opposed to

21:16

say, okay, well, I've got $3,000, I can get 10 LinkedIn posts written for that.

21:20

I'll post one of them. That's not

21:23

efficient. It takes God way too long. And it's never going to drive the

21:25

outcomes that you actually want.

21:29

All right. Well, we're going to take it now a little more complex. You've seen

21:32

this is really another task. We can't

21:35

expand it. That's awesome. But where to come up with the original posts? Now, I

21:39

view the kind of posts that we're creating

21:41

intentionally right now as we switch from B to C to a more B to B platform in

21:45

two ways. We have to have the standard

21:48

SEO plays. We got a lot of ground to cover. I know SEO is dying. It's going to

21:51

be dead, whatever you want to say. But it's

21:54

still important. And it is one of those keep the lights on. However, it is one

21:57

of the most expensive and time consuming

22:00

effects in content marketing, if you want to do it well. Takes a lot to hire

22:03

writers and editors and then to get

22:06

everything in the right places. So we did something a little bit different. I

22:10

built a workflow that started with

22:12

just -- actually, yeah, it's not this. Don't worry about that. We'll show the

22:15

output of that. Basically, it starts

22:17

with the keyword. And this is another workflow that I built. So it starts with

22:19

the keyword. And that's the only

22:21

thing that we enter at first. From that keyword, what it does -- and this is

22:25

when I said that workflows create a

22:27

process, but accomplish the tasks along the way, an end-to-end process. The

22:31

workflow will take a keyword and it

22:33

will actually run that keyword through Google. And it will take the top five

22:36

results. Then it will scan those

22:39

results and look at all the H2 and H3 headings. Trying to figure out what's

22:43

important information relative to

22:45

that keyword and what does the reader want to get back. Then it will come up

22:48

with some ideas, well, how can we make

22:50

that better? Then it will come up with some related keywords. It will

22:53

ultimately create a content outline, a

22:56

content brief, because we do work with human writers and we want to send them

22:59

that context as well. Once that's done

23:01

generating, what you get is the result, it automatically kicks off this

23:05

workflow to create a long-form blog post. And

23:09

the tools that we use -- we use Surfer SEO. It's not a magic bullet solution,

23:13

but we use it as kind of a

23:14

gauge. Out of the box, these were scoring 60 to 70 on a -- most scales were

23:18

between 75 and 85 at the top scores. So we

23:22

can hand these off to our human writers and say, hey, make sure this is not

23:25

garbage. Make sure there are no

23:27

hallucinations. Go in, add some insights. But they have a draft that they can

23:31

work through way more quickly. And my

23:34

goal was with our writers with the boring SEO stuff. It's my least favorite

23:37

part of content marketing, our blog. It's

23:40

my least favorite part, but it's very important. And I wanted to find a way,

23:42

how can we get more content back without

23:45

increasing the budget and get our writers earning more per hour based on the

23:49

campaigns that they're working? I

23:50

didn't want to rip them off either. And I'm happy to say with this -- with that

23:54

version, what we did is we give our

23:56

writers something like this, and they're able to do it in maybe 25% of the time

24:00

, so they're earning the same. But we

24:03

get two or three articles back in the time it used to take them to make one, if

24:06

that makes sense. So in that

24:08

way, I do believe we're scaling very responsibly. >> So I'm going to, you know,

24:13

keep with the same punch line. What does

24:14

this actually look like at scale? And one of the things -- for those of you

24:18

that were in the last session, you

24:20

probably heard me ask about transcript data. That's because that has been a

24:23

goldmine and a very fascinating topic

24:26

for me. We use it a lot of different ways. But imagine a slightly more complex

24:29

workflow. What if we could take a

24:31

meeting transcript, automatically identify all of the -- and extract all the

24:34

topics that are discussed, look at

24:37

all the topics and identify is it a how-to article, is it a best practices

24:40

article, is it a thought leadership

24:41

article, like what could we write with this, go back and scan to extract all

24:44

the supporting details from the

24:46

original article, then generate an outline, then use that outline to actually

24:51

write a full form article, then

24:52

identify of all the people that work at your company, who is the most prolific

24:56

person and knowledgeable person

24:58

about this, adapt it to their own brand voice using writing samples that they

25:03

've provided us, generate social

25:05

promo content, and then ultimately publish the blog article directly to Notion

25:08

and ultimately to Webflow, and

25:10

then just send the person a slack saying, you know, you wrote something, right?

25:16

That's actually what they say. And here's some social media posts, LinkedIn posts to promote it, written in your

25:19

voice. That sounds like a pipe

25:22

dream. I'm sure I lost a lot of people that are like, no way, AI can do that.

25:24

>> Are you just telling us a dream? >> No.

25:27

I'm about to show you that this is reality. Thank you. That was a great tea out

25:32

. >> I did say it sounds like a

25:34

flower. >> Your work look credits will be in your account. Don't worry. What

25:40

you're looking at here is an

25:42

experiment we ran for six months. >> Right. The stupid. >> Here. >> Don't curse

25:49

the computer. >> Yeah, they're

25:54

listening. So what you're looking at here is a Notion database that contains

25:58

content that has been published for

26:01

six months. The reason we're having all this content is because what my team

26:06

does on a day-to-day basis, it

26:07

talks to customers about their most prolific problems in the market space.

26:10

Every customer is different. Some of

26:11

these articles are super long tail. We had a customer that is using AI to try

26:14

to write radio ad copy. And

26:16

that's the most long tail thing in the world. They only have one customer that

26:19

has ever asked me about that. We

26:21

teased that out as a topic. It extracted all of the things we discussed on the

26:24

call, the considerations that we

26:25

mentioned as to what model they use, why it would be hard, what kinds of

26:28

challenges you might need to overcome, and

26:30

ultimately wrote some really high quality content from it. I don't want to look

26:34

for a specific article here, but let

26:35

me just open this guy up. You can get a sense of the quality of content. It's

26:40

well formatted. It actually

26:42

follows the overall logical order of how we're thinking about things. And of

26:46

course has a general outline in

26:47

mind. We want to talk about impact. We want to talk about a new way for AI-

26:52

powered content repurposing. I

26:53

didn't even look at the title of the article. The content repurposing super

26:56

power, creating tailor variations for every persona. That might as well be something

26:59

that is based on the conversation we're

27:01

having right now. And so here's the thing, right? The old process is something

27:04

that was broken in the first

27:06

place. You would have an interview with someone in the field or a founder or a

27:11

CEO. You would ultimately use

27:13

that interview to translate that with your own brain into your understanding of

27:16

the topic. You would then

27:18

have them probably QA the topic. Maybe then it would go to a writing agency and

27:21

you'd pay them, you know, countless

27:23

amount of dollars to give you back a piece of content from someone that is

27:26

further removed from their writing

27:28

style than you are. Ultimately, great content in many cases, right? People are

27:31

really good at what they do. But

27:33

it's a really inefficient process. Whereas when we're actually using the

27:35

details of what happened on the call,

27:37

letting AI process that and then telling it to go right just like you do, we

27:41

can actually drive some

27:43

incredible outcomes. And you know, the impact of this is actually we don't

27:48

publish all of it. We've actually come up

27:50

with a couple of things. The author has been predicted. Nathan, you're prolific

27:53

and clearly an expert in a lot of

27:55

different topics. It's tagged it. The only thing we haven't automated is this

28:00

publish flag. And that is because

28:03

Google guidelines and not wanting to get dinged. But what this means is that

28:07

Nathan gets to sit on a treasure

28:09

trope of more content that you could publish in two, three years. >> I'm

28:12

exhausted thinking about it. I'm going to be

28:16

talking about it a little bit. When we were given this, I was told Nathan, go.

28:20

Publish. It's absolutely not. I don't know if you can see this number.

28:24

How long has this been up? >> We started this in October of last year. >> 6,000

28:29

full-length articles written

28:31

from customer calls with real pain points. Can I see this for a second? >> The

28:34

reason that this is really important to me is as I'm

28:37

writing a blog post or an article or anything, if I find a use case that I want

28:42

to explain, I can pop in here and I can

28:45

search it. It's already written step by step. I have to verify. I don't have to

28:51

go through it. Again, the emphasis on verify. Stay the human in the loop. Make sure it's accurate. Make sure it's working.

28:55

But there's a second benefit that I have found

28:58

with this. Sales bugs me a lot less throughout the week for things that they

29:03

need to send customers. They can take this. They

29:05

have access. They pop in. And if a customer has a similar question to a call

29:10

they had last week, they can say, hey, let me

29:12

send you a quick one pager for that. They have it ready. They have 6,000

29:16

options if they want. Which is incredible.

X

This is a test comment.

X

This is a longer test comment to see how this looks if the person decides to ramble a bit. So they're rambling and rambling and then they even lorem ipsum.


author