In this episode we discuss: AI employees. We are joined by Matt Lhoumeau, Cofounder & CEO at Concord

Love The Operations Room? Please support us by rating and reviewing it here.

We chat about the following with Matt Lhoumeau: 

1. How do you balance the need for structure with the chaos of fast-growing operations?
2. What happens when marketing becomes a true conversation rather than a one-way message?
3. Why are standards and expectations so critical to building trust in a growing team?
4. How can leaders create clarity without stifling creativity?
5. What’s the real difference between being busy and being effective?

References 

• https://www.linkedin.com/in/mattlhoumeau/
• https://www.concord.app/
• https://eu1.hubs.ly/H0lMWln0

Biography 

Matt Lhoumeau is the CEO and co-founder of Concord, the leading provider of AI-powered Agreement Intelligence solutions. With over a decade of experience transforming how businesses manage contracts, Matt helps operations leaders unlock strategic value from their agreements and turn contracts from cost centers into profit drivers.

To learn more about Beth and Brandon or to find out about sponsorship opportunities click here. 

Summary

04:05 – AI tools and operational leadership.

08:05 – The power of systems thinking in AI bots

12:01 – Using AI to challenge you

15:16 – AI-first mantra

19:40 – Legal department AI substitutes

22:36 – Being a systems owner

24:30 – ChatGPT as a therapist

27:37 – Using AI to evolve and be durable

31:34 – How tools and frameworks are used over time.

34:58 – The balance between AI and human intervention

35:19 – Evolving with AI

37:05 – The right inputs

This podcast uses the following third-party services for analysis:

Podcorn - https://podcorn.com/privacy

Speaker: 00:00:05

Hello everyone and welcome to

Speaker: 00:00:07

another episode of the operations

Speaker: 00:00:08

room a podcast for COOs.

Speaker: 00:00:10

I'm Brandon Mensinga joined by my

Speaker: 00:00:12

amazing co-host Bethany Ayers.

Speaker: 00:00:14

How are things going Bethany this

Speaker: 00:00:15

morning?

Speaker: 00:00:16

I'm well.

Speaker: 00:00:18

I had our first

Speaker: 00:00:20

company off site this week,

Speaker: 00:00:22

Tuesday and Wednesday.

Speaker: 00:00:23

It was really good.

Speaker: 00:00:25

We randomly had a voucher

Speaker: 00:00:27

from a canceled or

Speaker: 00:00:29

a postponed off site from last year,

Speaker: 00:00:32

where the team were going to go to

Speaker: 00:00:34

Serbia for

Speaker: 00:00:36

four days, three nights.

Speaker: 00:00:39

Ended up having to postpone it

Speaker: 00:00:41

because a large customer win came

Speaker: 00:00:43

in and like too much of the

Speaker: 00:00:44

team needed to be involved in that

Speaker: 00:00:46

to go away for four days.

Speaker: 00:00:48

And so we had this voucher to spend

Speaker: 00:00:49

and they really wanted us to go back

Speaker: 00:00:51

to Serbia. Like I think they

Speaker: 00:00:52

might've owed the Serbian

Speaker: 00:00:54

place some money.

Speaker: 00:00:55

And there was just no way it was

Speaker: 00:00:57

going to take four days out of the

Speaker: 00:00:58

business. And also it was

Speaker: 00:01:00

getting to Belgrade and then a two

Speaker: 00:01:02

hour transfer to the middle of the

Speaker: 00:01:03

mountains.

Speaker: 00:01:04

What? Who decided to do this?

Speaker: 00:01:07

So A, why Serbia?

Speaker: 00:01:08

B, why two hours out of the airport?

Speaker: 00:01:12

And you have a voucher as a result.

Speaker: 00:01:14

Yeah, so it was like

Speaker: 00:01:16

seven hours of transport.

Speaker: 00:01:18

So you're basically spending two

Speaker: 00:01:19

days traveling in order to spend two

Speaker: 00:01:21

days together.

Speaker: 00:01:22

So we switched

Speaker: 00:01:24

it to Brighton and

Speaker: 00:01:25

one night in Brighton cost the

Speaker: 00:01:27

same as three nights in Serbia.

Speaker: 00:01:31

I can imagine.

Speaker: 00:01:33

But also like transport

Speaker: 00:01:34

was way cheaper.

Speaker: 00:01:36

So overall NetNet we were definitely

Speaker: 00:01:38

winners going to Brighton.

Speaker: 00:01:40

Nice hotel, really easy

Speaker: 00:01:42

for everyone to get to.

Speaker: 00:01:43

We had quite the night out

Speaker: 00:01:45

or the team had quite a night out on

Speaker: 00:01:48

Wednesday night.

Speaker: 00:01:49

Like I purposely went to bed at half

Speaker: 00:01:50

10. One, I'm just not much of a

Speaker: 00:01:52

drinker anymore.

Speaker: 00:01:53

And two, everybody can have more

Speaker: 00:01:55

fun when the boss isn't around

Speaker: 00:01:57

and like.

Speaker: 00:01:59

Three CEOs in six months

Speaker: 00:02:01

is a lot for the team.

Speaker: 00:02:03

There's a lot of change.

Speaker: 00:02:04

Just let everybody blow off a bit of

Speaker: 00:02:05

steam. But the blowing off a big of

Speaker: 00:02:07

steam meant that some members

Speaker: 00:02:09

of the team did not go

Speaker: 00:02:11

to bed till 4.30 in the morning.

Speaker: 00:02:13

What?

Speaker: 00:02:14

And you have the off-site the next

Speaker: 00:02:15

day, and they're going to bed at

Speaker: 00:02:16

4.30.

Speaker: 00:02:17

Wow. Like the second day of the

Speaker: 00:02:19

offsite, yeah.

Speaker: 00:02:20

The second day. Okay, fine.

Speaker: 00:02:21

And the hotel was right

Speaker: 00:02:23

on the beach.

Speaker: 00:02:25

If you've been to Brighton, there's

Speaker: 00:02:26

like a big road and then we were

Speaker: 00:02:28

the very first hotel there and

Speaker: 00:02:30

members of the team just had to go

Speaker: 00:02:31

and get fresh air on a fairly

Speaker: 00:02:32

regular basis. They were looking

Speaker: 00:02:33

very peaky.

Speaker: 00:02:35

They'd go get some fresh sea air for

Speaker: 00:02:37

20 minutes, come back looking

Speaker: 00:02:39

fine. And about half an hour later,

Speaker: 00:02:41

they were just collapsing again in

Speaker: 00:02:42

their peakiness.

Speaker: 00:02:43

I was just entertained.

Speaker: 00:02:45

So the bulk of us were able to

Speaker: 00:02:46

build our AI employees reason.

Speaker: 00:02:49

Test a bunch of technology and

Speaker: 00:02:51

have a bit of a, not a hackathon

Speaker: 00:02:53

because everything that was built

Speaker: 00:02:54

needs to be able to be used.

Speaker: 00:02:56

But then there was like a handful of

Speaker: 00:02:58

people who used the second

Speaker: 00:03:00

day to breathe in sea air.

Speaker: 00:03:02

Okay, so pull us back for a second.

Speaker: 00:03:04

So two day off site, you're building

Speaker: 00:03:06

AI agents.

Speaker: 00:03:07

What is going on here?

Speaker: 00:03:09

So, one of our OKRs

Speaker: 00:03:11

for the quarter is to become an AI

Speaker: 00:03:12

first company, and our

Speaker: 00:03:15

key results for that is

Speaker: 00:03:17

to have 50 AI

Speaker: 00:03:19

employees in the business

Speaker: 00:03:21

and then have

Speaker: 00:03:23

end-to-end automation

Speaker: 00:03:25

in the go-to market team and an end-

Speaker: 00:03:27

to-end animation in the engineering

Speaker: 00:03:29

team. Go-to market have already done

Speaker: 00:03:31

it, so we might have to make that

Speaker: 00:03:33

key result a bit harder.

Speaker: 00:03:34

And then on the engineering

Speaker: 00:03:36

team, they have an idea.

Speaker: 00:03:38

Which is basically linear

Speaker: 00:03:40

will talk to Claude

Speaker: 00:03:42

code and

Speaker: 00:03:44

fix small bugs automated.

Speaker: 00:03:47

So you log a bug

Speaker: 00:03:49

and then Claude Code goes off and

Speaker: 00:03:51

fixes it.

Speaker: 00:03:52

So they're gonna test that and see

Speaker: 00:03:54

for a lot of like just the small,

Speaker: 00:03:56

like simple ones.

Speaker: 00:03:57

And then on the go-to-market team,

Speaker: 00:03:59

we have built, I don't

Speaker: 00:04:01

know like the timings of

Speaker: 00:04:03

these episodes that what Donna

Speaker: 00:04:05

McCurley talked about in terms of.

Speaker: 00:04:07

Figuring out, we have one agent

Speaker: 00:04:09

that looks at what our best

Speaker: 00:04:10

customers look like and goes off and

Speaker: 00:04:12

finds look-alikes with the buying

Speaker: 00:04:14

intent that then

Speaker: 00:04:17

writes personalized emails

Speaker: 00:04:19

based on what good looks like

Speaker: 00:04:21

and then enters them into

Speaker: 00:04:23

a HubSpot sequence.

Speaker: 00:04:25

And then from responses to the

Speaker: 00:04:27

sequence, it feeds the SDR.

Speaker: 00:04:29

That is super awesome.

Speaker: 00:04:31

And then what are you using for the

Speaker: 00:04:32

tooling to make that happen?

Speaker: 00:04:34

So we now have business

Speaker: 00:04:36

chat GPT for everyone in the team,

Speaker: 00:04:38

and we are

Speaker: 00:04:40

using in go-to-market a tool

Speaker: 00:04:42

called Cargo, which is kind

Speaker: 00:04:44

of like an N8N or a

Speaker: 00:04:46

Clay, but specific, like it's

Speaker: 00:04:48

like Clay in that it's specific to

Speaker: 00:04:50

go-go-to market.

Speaker: 00:04:51

But from what I understand,

Speaker: 00:04:53

it can also do N8n or

Speaker: 00:04:55

Zapier type stuff.

Speaker: 00:04:56

So even though it happens to have

Speaker: 00:04:58

some nice plugins for

Speaker: 00:05:00

like dealing with LinkedIn.

Speaker: 00:05:02

Finding lookalikes, different areas

Speaker: 00:05:04

that it can search over.

Speaker: 00:05:07

It also then just does automated

Speaker: 00:05:09

workflows like any of these tools.

Speaker: 00:05:11

And then we also bought

Speaker: 00:05:13

Claude code subscriptions

Speaker: 00:05:15

for everybody in the

Speaker: 00:05:18

engineering team.

Speaker: 00:05:19

We had cursor and the

Speaker: 00:05:21

cursor you can put different things

Speaker: 00:05:22

in, but apparently Claude Code is

Speaker: 00:05:24

just like, it's a whole new thing

Speaker: 00:05:25

and it's way better.

Speaker: 00:05:27

And it's not just the model that you

Speaker: 00:05:29

plug into Cursor.

Speaker: 00:05:30

What Cloud Code can do is

Speaker: 00:05:32

astounding.

Speaker: 00:05:33

And we had training on

Speaker: 00:05:35

day one, because I'm

Speaker: 00:05:37

bouncing over a place, so I can tell

Speaker: 00:05:39

I'm a bit tired.

Speaker: 00:05:40

So we are working on

Speaker: 00:05:42

creating these AI employees and

Speaker: 00:05:44

becoming the AI first company.

Speaker: 00:05:46

And so we use the offsite to ring

Speaker: 00:05:48

fence time for everybody to

Speaker: 00:05:50

learn and explore.

Speaker: 00:05:52

And so day one before

Speaker: 00:05:54

the fun.

Speaker: 00:05:55

Letting off steam till 430

Speaker: 00:05:57

in the morning evening.

Speaker: 00:05:59

Everybody was sober and not

Speaker: 00:06:01

hung over and capable of being in

Speaker: 00:06:03

training for the day.

Speaker: 00:06:04

We did two

Speaker: 00:06:06

training sessions.

Speaker: 00:06:07

So the morning session was for the

Speaker: 00:06:09

entire company.

Speaker: 00:06:10

What do you call it?

Speaker: 00:06:10

Not when you're the number one

Speaker: 00:06:12

fan of somebody like My idol?

Speaker: 00:06:15

You're a fanboy, a fangirl in this

Speaker: 00:06:17

case.

Speaker: 00:06:18

I'm a fangirl, yeah, of

Speaker: 00:06:20

Charlie Cowan, who's been on the

Speaker: 00:06:21

podcast a couple times.

Speaker: 00:06:22

Charlie ran his

Speaker: 00:06:24

101 and 102 classes

Speaker: 00:06:27

on ChatGPT for the whole

Speaker: 00:06:29

team.

Speaker: 00:06:30

The most amazing feedback.

Speaker: 00:06:33

Everybody in the team loved it,

Speaker: 00:06:35

including the engineers, because I

Speaker: 00:06:36

thought it might just be a bit basic

Speaker: 00:06:37

for the engineers.

Speaker: 00:06:38

But everybody learned, everybody

Speaker: 00:06:40

loved Charlie's energy.

Speaker: 00:06:41

And at the wrap up

Speaker: 00:06:43

on the second day.

Speaker: 00:06:45

Multiple people asked if we could do

Speaker: 00:06:47

quarterly training with him.

Speaker: 00:06:48

Oh, wow.

Speaker: 00:06:49

OK. Kudos, Charlie.

Speaker: 00:06:50

That's amazing.

Speaker: 00:06:51

I have never had training

Speaker: 00:06:53

in a company where people

Speaker: 00:06:55

were begging for more training.

Speaker: 00:06:57

Yeah, that's super awesome.

Speaker: 00:06:59

And we all just learned so much

Speaker: 00:07:01

around what chat GPT can do,

Speaker: 00:07:03

all the different functionality,

Speaker: 00:07:05

ways to approach it.

Speaker: 00:07:06

He had a very safe environment.

Speaker: 00:07:09

Lots of people would own up to

Speaker: 00:07:10

mistakes they made or problems they

Speaker: 00:07:12

were having.

Speaker: 00:07:13

He gave different techniques.

Speaker: 00:07:15

And then in the afternoon,

Speaker: 00:07:17

the go-to-market team went and

Speaker: 00:07:20

started building their

Speaker: 00:07:21

AI employees had a brainstorm on

Speaker: 00:07:23

like. What kinds of things

Speaker: 00:07:25

could be built, what would be

Speaker: 00:07:26

helpful in their jobs, how to build

Speaker: 00:07:28

them, and started doing that.

Speaker: 00:07:30

And the afternoon was a combination

Speaker: 00:07:32

of Charlie and somebody else who's

Speaker: 00:07:34

going to be a guest on our show

Speaker: 00:07:35

soon, Ryan Fuller.

Speaker: 00:07:37

And we did a more like workshoppy

Speaker: 00:07:39

thing and more

Speaker: 00:07:41

of a deep dive into Cloud Code.

Speaker: 00:07:43

But it wasn't as structured

Speaker: 00:07:45

a format as Charlie did

Speaker: 00:07:47

for Chat GPT.

Speaker: 00:07:48

So I'm thinking for the next

Speaker: 00:07:49

quarterly training, we do a deep

Speaker: 00:07:51

on cloud code, because apparently...

Speaker: 00:07:53

It can do stuff even for non-coders.

Speaker: 00:07:56

Like Charlie's using it to run his

Speaker: 00:07:57

strategy.

Speaker: 00:07:58

Charlie used it to help him

Speaker: 00:08:00

build his new website.

Speaker: 00:08:02

Like the way that the structure is,

Speaker: 00:08:04

I don't entirely understand, means

Speaker: 00:08:05

that it's super powerful, not

Speaker: 00:08:07

just for coding.

Speaker: 00:08:08

That's amazing.

Speaker: 00:08:09

So then, Charlie Cowan, 101,

Speaker: 00:08:11

102, the sessions up front for

Speaker: 00:08:12

ChatGBT, this fellow doing more

Speaker: 00:08:14

of a workshop on cloud code in

Speaker: 00:08:16

day one, and then you're doing a bit

Speaker: 00:08:18

of like brainstorming with the

Speaker: 00:08:19

functional teams around AI employees

Speaker: 00:08:21

and having the teams think through

Speaker: 00:08:23

how to build those things as part of

Speaker: 00:08:25

the day one.

Speaker: 00:08:26

As part of day one, and then day two

Speaker: 00:08:28

was building things.

Speaker: 00:08:30

And then we all got

Speaker: 00:08:31

together, and there was a bit of a

Speaker: 00:08:33

show and tell for anybody who wanted

Speaker: 00:08:34

to show what they'd built and

Speaker: 00:08:36

how they'd build it.

Speaker: 00:08:38

So I built a

Speaker: 00:08:40

demo planner, GPT,

Speaker: 00:08:43

where you can drop in transcripts

Speaker: 00:08:45

from your first meet,

Speaker: 00:08:47

and it will help you build a

Speaker: 00:08:49

tailored demo.

Speaker: 00:08:50

I put in the Sandler methodology,

Speaker: 00:08:53

MedPic.

Speaker: 00:08:54

Another methodology on like what

Speaker: 00:08:55

good demos look like,

Speaker: 00:08:58

some of our transcripts anonymized

Speaker: 00:09:00

around first meets and

Speaker: 00:09:02

good demos so it knows what good

Speaker: 00:09:04

looks like, our knowledge

Speaker: 00:09:06

base so it understands our platform

Speaker: 00:09:08

and how it works, and I can't

Speaker: 00:09:09

remember a couple other things, plus

Speaker: 00:09:11

the instructions.

Speaker: 00:09:12

ChatGPT helped me write the

Speaker: 00:09:14

instructions for the GPT

Speaker: 00:09:16

on the structure of what it should

Speaker: 00:09:17

look like the questions you should

Speaker: 00:09:18

ask, what the outputs are, and

Speaker: 00:09:21

then When you drop in

Speaker: 00:09:23

a transcript, it tells you

Speaker: 00:09:25

a summary of the calls of who the

Speaker: 00:09:26

key decision makers are, where the

Speaker: 00:09:28

power is, all of the stuff that like

Speaker: 00:09:30

between Sandler and MedPic you get.

Speaker: 00:09:32

It gives you the flow for the

Speaker: 00:09:34

demo, questions to drop

Speaker: 00:09:36

in and pepper, talking points

Speaker: 00:09:38

and which person to address because

Speaker: 00:09:40

that was their particular need in

Speaker: 00:09:42

the demo. A follow-up

Speaker: 00:09:44

email, and a

Speaker: 00:09:47

mini- mutual action

Speaker: 00:09:49

plan to get to the proof of value

Speaker: 00:09:51

because it also understands our

Speaker: 00:09:52

sales process.

Speaker: 00:09:54

So it's like, so we need to do a

Speaker: 00:09:55

proof of value in three weeks.

Speaker: 00:09:56

Therefore, this step needs to

Speaker: 00:09:58

happen. These are the owners and

Speaker: 00:09:59

it's ready to go that you can just

Speaker: 00:10:01

share with a customer.

Speaker: 00:10:02

That's amazing.

Speaker: 00:10:03

And then is this a folder in

Speaker: 00:10:05

ChachiBT that you're using to do

Speaker: 00:10:06

this or?

Speaker: 00:10:07

So this is a GPT

Speaker: 00:10:09

that I've used on this one.

Speaker: 00:10:12

So you, a custom GPT,

Speaker: 00:10:14

that's just shared within Matomic.

Speaker: 00:10:16

Cause now that we have the

Speaker: 00:10:17

GPT workspace, you can

Speaker: 00:10:19

publish things and everything can be

Speaker: 00:10:22

available to everybody else.

Speaker: 00:10:23

So can talk into the GP

Speaker: 00:10:25

T and it has all of this structure

Speaker: 00:10:27

and then you get your own chat.

Speaker: 00:10:29

And then that chat will have your

Speaker: 00:10:31

memory in it of just your

Speaker: 00:10:33

chat. Whereas the GPt doesn't get

Speaker: 00:10:35

affected by other people's memory.

Speaker: 00:10:37

If that makes sense.

Speaker: 00:10:38

That makes sense.

Speaker: 00:10:40

Shareable custom GBT.

Speaker: 00:10:41

You can do your own work in there

Speaker: 00:10:42

and maintains a memory persistence

Speaker: 00:10:44

thing whereby it understands your

Speaker: 00:10:45

context specifically, but doesn't

Speaker: 00:10:46

pollute others with yours.

Speaker: 00:10:48

Exactly, because it's in like a

Speaker: 00:10:50

separate chat with yours, but the

Speaker: 00:10:52

GPT itself has not been polluted by

Speaker: 00:10:53

your content.

Speaker: 00:10:55

And it can also, if you're doing a

Speaker: 00:10:56

first meet, you can give a little

Speaker: 00:10:58

bit of background and ask for

Speaker: 00:11:00

questions to make sure to

Speaker: 00:11:01

ask so that it'll give you a good

Speaker: 00:11:03

tailored demo for the second meet.

Speaker: 00:11:05

So that's what I built.

Speaker: 00:11:06

And then I'm in the process

Speaker: 00:11:08

of building a CMO

Speaker: 00:11:10

for us using some

Speaker: 00:11:12

content of somebody online that

Speaker: 00:11:14

Charlie recommended.

Speaker: 00:11:15

Our CSM team.

Speaker: 00:11:18

Connected all of their different

Speaker: 00:11:21

like HubSpot plus

Speaker: 00:11:23

other spreadsheets they have of all

Speaker: 00:11:24

kinds of data.

Speaker: 00:11:26

And they can now query when

Speaker: 00:11:28

a renewal is coming up, what's our

Speaker: 00:11:29

forecast, what are my

Speaker: 00:11:31

actions, which QBRs do I

Speaker: 00:11:33

need to do?

Speaker: 00:11:34

And also it goes and looks

Speaker: 00:11:36

at the news for

Speaker: 00:11:38

your customers.

Speaker: 00:11:39

So it could give you actions plus

Speaker: 00:11:40

news that you need to know about

Speaker: 00:11:42

like in the outside world and little

Speaker: 00:11:44

snippets to share with them.

Speaker: 00:11:45

In retrospect, what's your sense

Speaker: 00:11:47

of the value of the two-day

Speaker: 00:11:49

session itself?

Speaker: 00:11:50

What do you have done anything differently?

Speaker: 00:11:52

Do you think this is really the

Speaker: 00:11:53

springboard to become, to achieve

Speaker: 00:11:54

your OKRs and be AI first?

Speaker: 00:11:56

Oh, totally. We needed those

Speaker: 00:11:59

days of time outside

Speaker: 00:12:01

of your day-to-day work to just

Speaker: 00:12:03

open people's minds and the

Speaker: 00:12:04

training.

Speaker: 00:12:05

Because actually, one of the things

Speaker: 00:12:06

that I learned that I wasn't

Speaker: 00:12:07

expecting is I tried the chat

Speaker: 00:12:09

GPT voice

Speaker: 00:12:11

mode before and it was quite

Speaker: 00:12:13

slow and didn't feel like a human.

Speaker: 00:12:15

It's massively improved.

Speaker: 00:12:18

So Charlie demoed it and then

Speaker: 00:12:20

yesterday in one-to-one

Speaker: 00:12:22

with an SDR.

Speaker: 00:12:24

So we've created another

Speaker: 00:12:27

custom GPT that is

Speaker: 00:12:29

a cynical CISO.

Speaker: 00:12:30

So we'd done like a persona of a

Speaker: 00:12:32

chief information officer and we run

Speaker: 00:12:34

all of our content through that to

Speaker: 00:12:36

make sure that we are not too

Speaker: 00:12:37

salesy, that we seem credible.

Speaker: 00:12:40

And we just flipped that into chat

Speaker: 00:12:42

mode, said,

Speaker: 00:12:44

we're calling you, here's the

Speaker: 00:12:46

first pitch.

Speaker: 00:12:47

How can we make it better?

Speaker: 00:12:49

And because ChatGPT is always very

Speaker: 00:12:51

sensible, it's like, oh, well that

Speaker: 00:12:52

was a really good pitch.

Speaker: 00:12:53

These are, you know, some notes I'd

Speaker: 00:12:55

have on how to improve it.

Speaker: 00:12:56

And it came up with like, it was an

Speaker: 00:12:57

excellent pitch and it feels

Speaker: 00:12:59

like an actual person.

Speaker: 00:13:01

So you can really practice your

Speaker: 00:13:03

pitches. You can practice going out

Speaker: 00:13:05

to investors.

Speaker: 00:13:06

You can practice everything.

Speaker: 00:13:08

Charlie and I were chatting this

Speaker: 00:13:09

morning and he'd had another

Speaker: 00:13:11

training session with another team

Speaker: 00:13:13

of execs yesterday and

Speaker: 00:13:15

they're going out for fundraising

Speaker: 00:13:16

and the CEO, they built

Speaker: 00:13:18

a grumpy investor.

Speaker: 00:13:20

And we're just practicing pitching

Speaker: 00:13:22

to a grumpy investor like what

Speaker: 00:13:23

landed what didn't land what would

Speaker: 00:13:24

be better ideas amazing

Speaker: 00:13:26

for anybody who's doing human

Speaker: 00:13:28

stuff like how to practice without

Speaker: 00:13:30

having to be on other humans.

Speaker: 00:13:32

I think one of the things that I'm

Speaker: 00:13:34

thinking about right now is this

Speaker: 00:13:35

question of time and space to like

Speaker: 00:13:37

think about the stuff.

Speaker: 00:13:38

Because I feel like in my current

Speaker: 00:13:40

company and previous companies,

Speaker: 00:13:42

the day-to-day grind of getting

Speaker: 00:13:43

stuff done and what has to get done

Speaker: 00:13:45

for the business doesn't allow you

Speaker: 00:13:47

to have a lot of extra space and

Speaker: 00:13:48

capacity to think about different

Speaker: 00:13:50

stuff, you know, and in particular

Speaker: 00:13:51

for AI to your point, it takes

Speaker: 00:13:53

a bit of time in effort to think

Speaker: 00:13:55

through like if we wanted to do

Speaker: 00:13:56

something here properly with AI

Speaker: 00:13:58

employees or what have you or for

Speaker: 00:13:59

myself or for the team, it requires

Speaker: 00:14:01

space to do that.

Speaker: 00:14:02

So this idea of doing a

Speaker: 00:14:04

proper off-site in the format that

Speaker: 00:14:05

you just talked about Feels like

Speaker: 00:14:06

that's almost like an instrumental

Speaker: 00:14:08

linchpin to like get

Speaker: 00:14:10

everyone's mind in the same place,

Speaker: 00:14:12

get the training there to enable

Speaker: 00:14:13

them to be successful creating these

Speaker: 00:14:15

employees and set yourself on

Speaker: 00:14:17

an actual path.

Speaker: 00:14:18

100% because also like

Speaker: 00:14:21

so many times companies waste days

Speaker: 00:14:23

on hackathons and like it's

Speaker: 00:14:24

quite nice to have a hackathon

Speaker: 00:14:26

because you can experiment but

Speaker: 00:14:28

it all ends up just being

Speaker: 00:14:30

shelf wear and so

Speaker: 00:14:31

it was like this was not

Speaker: 00:14:33

around experimenting I mean

Speaker: 00:14:35

it was because you're always going

Speaker: 00:14:36

to get better but it needed to be

Speaker: 00:14:38

things that would actually make

Speaker: 00:14:40

your lives better tomorrow and

Speaker: 00:14:43

figure stuff out and also like

Speaker: 00:14:45

playing around with AI employees,

Speaker: 00:14:47

but also playing around with or

Speaker: 00:14:49

taking the time to get

Speaker: 00:14:50

our HubSpot data better, get

Speaker: 00:14:52

our dev environment faster to move.

Speaker: 00:14:55

Like there's almost ring-fencing

Speaker: 00:14:57

time to slow down with

Speaker: 00:14:58

all of those things that you know if

Speaker: 00:15:00

you do, you'll speed up, but aren't

Speaker: 00:15:02

quite bad enough.

Speaker: 00:15:04

So I think we're going to do this at

Speaker: 00:15:06

least quarterly.

Speaker: 00:15:07

It might not be an offsite every

Speaker: 00:15:08

time, but ring-fence

Speaker: 00:15:10

time to slowly down to

Speaker: 00:15:13

speed up.

Speaker: 00:15:14

Specifically around AI and

Speaker: 00:15:15

automation.

Speaker: 00:15:16

And then the overall mindset

Speaker: 00:15:18

and motivation, morale,

Speaker: 00:15:20

enthusiasm of the company now

Speaker: 00:15:22

with this AI first mantra that

Speaker: 00:15:24

you've brought to the company, can

Speaker: 00:15:25

you sense a real momentum now off

Speaker: 00:15:27

the back of this in the sense that

Speaker: 00:15:28

folks are really enthused and

Speaker: 00:15:30

excited to move forward with it?

Speaker: 00:15:31

Yes.

Speaker: 00:15:32

My hesitation is just that there's

Speaker: 00:15:34

always a spectrum.

Speaker: 00:15:35

You have some people who have a

Speaker: 00:15:37

hundred percent embraced it.

Speaker: 00:15:39

We have one engineer

Speaker: 00:15:41

in particular who has just opened

Speaker: 00:15:43

his mind and he is doing

Speaker: 00:15:45

so much.

Speaker: 00:15:47

And you have others who are still a

Speaker: 00:15:48

little bit skeptical.

Speaker: 00:15:49

They're into it.

Speaker: 00:15:50

They're not defensive,

Speaker: 00:15:52

but they're not as

Speaker: 00:15:54

just totally excited.

Speaker: 00:15:56

And then on the go-to-market side, I

Speaker: 00:15:59

would say it's similar.

Speaker: 00:16:01

For some people, there's just a

Speaker: 00:16:02

click. And they're just like,

Speaker: 00:16:04

this is amazing. And for others,

Speaker: 00:16:06

that click hasn't 100% happened yet,

Speaker: 00:16:08

but they're curious,

Speaker: 00:16:10

want to see how it goes,

Speaker: 00:16:12

and just need a little bit more

Speaker: 00:16:13

support to have that aha moment.

Speaker: 00:16:15

Here's a thought that I was having

Speaker: 00:16:17

the other day.

Speaker: 00:16:18

So there's another CEO, a

Speaker: 00:16:20

friend of mine, that works in a

Speaker: 00:16:21

organization, they're currently

Speaker: 00:16:22

doing fundraising, and they've

Speaker: 00:16:24

put together their financial

Speaker: 00:16:25

forecast of their shopping route

Speaker: 00:16:26

with investors in terms of here's

Speaker: 00:16:28

our revenue, the assumptions, here's

Speaker: 00:16:30

your budget and so on.

Speaker: 00:16:31

So that individual has effectively

Speaker: 00:16:33

shaved.

Speaker: 00:16:34

Outside of the classic B2B SaaS

Speaker: 00:16:36

scaling for headcount that

Speaker: 00:16:38

we had previously, they've shaved

Speaker: 00:16:39

off, let's say, 30%, 40%

Speaker: 00:16:41

of the headcount.

Speaker: 00:16:42

They've taken that cash and

Speaker: 00:16:44

mostly put it into a line item

Speaker: 00:16:46

called IT software.

Speaker: 00:16:48

So per person, per GTM

Speaker: 00:16:50

person, for developer allocating

Speaker: 00:16:52

a spend amount annually for

Speaker: 00:16:54

individuals in the company, and

Speaker: 00:16:56

the amounts that are being put into

Speaker: 00:16:57

the line item are quite significant.

Speaker: 00:16:59

We don't know exactly how much is

Speaker: 00:17:00

the right amount right now, but just

Speaker: 00:17:02

from a forecasting standpoint, let's

Speaker: 00:17:03

just take some of that headcount

Speaker: 00:17:05

spend reallocated to this line item

Speaker: 00:17:07

for the time being with really not

Speaker: 00:17:08

knowing how the future is going to

Speaker: 00:17:10

pan out, but let's not minimize the

Speaker: 00:17:11

potential cost that might be

Speaker: 00:17:13

associated with significant

Speaker: 00:17:15

token spend as it were, in

Speaker: 00:17:17

particular for developers as they

Speaker: 00:17:19

start using tooling quite

Speaker: 00:17:20

aggressively going forward.

Speaker: 00:17:21

What do you think of that?

Speaker: 00:17:23

I think it's wise.

Speaker: 00:17:24

I think, it's also generous.

Speaker: 00:17:26

Although it would future proof

Speaker: 00:17:28

you because everything right now is

Speaker: 00:17:29

a loss leader.

Speaker: 00:17:31

Nobody's making money and prices are

Speaker: 00:17:32

going to have to go up.

Speaker: 00:17:34

And as people move into the

Speaker: 00:17:35

enterprise, prices are gonna go up,

Speaker: 00:17:37

I kind of hoping that the enterprise

Speaker: 00:17:38

will subsidize the

Speaker: 00:17:40

startups. I would say it's probably

Speaker: 00:17:42

wise to forecast a bit more

Speaker: 00:17:44

spend so that when the prices do

Speaker: 00:17:46

ratchet up, you have the money?

Speaker: 00:17:48

So as we have

Speaker: 00:17:50

just bought Claude code for

Speaker: 00:17:51

everybody, right now they have three

Speaker: 00:17:53

subscriptions. They have a 20 pound

Speaker: 00:17:55

one that year, a $20 one that you

Speaker: 00:17:56

apparently run out of within

Speaker: 00:17:58

seconds, a

Speaker: 00:17:59

hundred dollar one that

Speaker: 00:18:01

seems to work for most people and

Speaker: 00:18:03

then a 200.

Speaker: 00:18:04

And we've gone for the hundred and

Speaker: 00:18:07

see if that works.

Speaker: 00:18:08

Otherwise go to the 200.

Speaker: 00:18:09

With the API calls, you don't know

Speaker: 00:18:11

what's going to happen in the

Speaker: 00:18:12

future. At least you know there'll

Speaker: 00:18:13

be enough money.

Speaker: 00:18:14

I think on the go-to-market side,

Speaker: 00:18:16

it'll clear out a bit.

Speaker: 00:18:18

Yeah, like right now, there's so

Speaker: 00:18:19

many tools coming in, and I think

Speaker: 00:18:21

quite a few of those tools will come

Speaker: 00:18:22

back out again.

Speaker: 00:18:23

And then I also think that

Speaker: 00:18:25

chat GPT or OpenAI will

Speaker: 00:18:28

end up building a lot of things that

Speaker: 00:18:29

let you consolidate.

Speaker: 00:18:31

Yes, I suspect so.

Speaker: 00:18:32

If I were a note taker, I'd be very

Speaker: 00:18:34

worried.

Speaker: 00:18:39

So we have a topic for today, which

Speaker: 00:18:41

we are talking about right now,

Speaker: 00:18:42

which is AI employees.

Speaker: 00:18:43

We have an amazing guest for this,

Speaker: 00:18:44

which Matt Lumo.

Speaker: 00:18:46

He is the CEO and co-founder of

Speaker: 00:18:47

Concord.

Speaker: 00:18:48

And we had a long conversation with

Speaker: 00:18:50

Matt around contracting

Speaker: 00:18:52

for companies and a bit of

Speaker: 00:18:54

a chat around CS.

Speaker: 00:18:55

But the interesting bit that we

Speaker: 00:18:57

spoke about was about AI employees

Speaker: 00:18:59

and what Matt is doing in that

Speaker: 00:19:00

respect. So before we get into the

Speaker: 00:19:02

AI employees piece of it,

Speaker: 00:19:04

is there any thoughts just on his

Speaker: 00:19:06

other bits that you talked about

Speaker: 00:19:07

related to CS and

Speaker: 00:19:09

contract management.

Speaker: 00:19:11

I guess on the contract management

Speaker: 00:19:13

part, I think I

Speaker: 00:19:15

will, at some point,

Speaker: 00:19:17

have a look at Concord, their

Speaker: 00:19:19

technology, because how

Speaker: 00:19:21

much do we actually need lawyers for

Speaker: 00:19:23

for contracts versus how much can

Speaker: 00:19:24

we use AI?

Speaker: 00:19:25

And they've basically created a

Speaker: 00:19:27

system where you can do a lot of

Speaker: 00:19:29

your contract negotiation via

Speaker: 00:19:31

their platform and then also not

Speaker: 00:19:33

lose all of your contracts.

Speaker: 00:19:34

I haven't tried it yet, but we

Speaker: 00:19:36

don't have a lawyer in-house as we

Speaker: 00:19:38

go up market and that are dealing

Speaker: 00:19:40

with larger.

Speaker: 00:19:42

Institutions, having somebody

Speaker: 00:19:43

help us review contracts for

Speaker: 00:19:46

standard ones seems like a good

Speaker: 00:19:47

idea.

Speaker: 00:19:48

Right now, for

Speaker: 00:19:50

commercial contracts in my current

Speaker: 00:19:51

company, we're just about to kick

Speaker: 00:19:53

off a process to make this

Speaker: 00:19:55

as friction-free and as fast as

Speaker: 00:19:57

humanly possible, leveraging tools

Speaker: 00:19:59

and a revised process.

Speaker: 00:20:02

Right now we've been using a lot of

Speaker: 00:20:04

human legal powers and work to

Speaker: 00:20:06

make modifications to contracts, but

Speaker: 00:20:08

it's all been very sloppy, I think,

Speaker: 00:20:10

in the sense of just relying upon

Speaker: 00:20:12

a person to inject themselves

Speaker: 00:20:14

to make these changes and not really

Speaker: 00:20:16

thinking through to make this

Speaker: 00:20:18

a friction-free process that is

Speaker: 00:20:19

much, much faster and doesn't lean

Speaker: 00:20:20

on that person so much.

Speaker: 00:20:22

The interesting bit is here

Speaker: 00:20:24

in the UK, with some of these

Speaker: 00:20:25

providers, there's some good

Speaker: 00:20:26

providers out there. We've mentioned

Speaker: 00:20:27

before like Harbor James and so on,

Speaker: 00:20:29

where they're very SaaS friendly and

Speaker: 00:20:31

they have subscription models

Speaker: 00:20:33

whereby they're actually reasonably

Speaker: 00:20:34

priced, I think, to be honest, like

Speaker: 00:20:36

really good value.

Speaker: 00:20:37

Talking to this woman in the US, I

Speaker: 00:20:39

don't know if this is just like a US

Speaker: 00:20:40

thing overall or just her

Speaker: 00:20:41

specifically, but immediately after

Speaker: 00:20:43

all this scale up talk, we went

Speaker: 00:20:45

into what is your business

Speaker: 00:20:47

model, how do you charge

Speaker: 00:20:48

effectively?

Speaker: 00:20:49

Oh, we charge by the hour.

Speaker: 00:20:51

So it's very transactional, hourly

Speaker: 00:20:52

based. And by the way, it's 750 US

Speaker: 00:20:55

per hour.

Speaker: 00:20:55

That is a lot of money.

Speaker: 00:20:58

So let me get this straight.

Speaker: 00:21:00

You're scale up focused.

Speaker: 00:21:01

SaaS focused and you have this

Speaker: 00:21:04

legacy old school business

Speaker: 00:21:06

model charging hourly defeats

Speaker: 00:21:08

the entire purpose because we want

Speaker: 00:21:10

fractional embedded time

Speaker: 00:21:12

consumption where you are

Speaker: 00:21:14

representative as part of our

Speaker: 00:21:15

company and we would like to pay you

Speaker: 00:21:17

in a way that is not transactional.

Speaker: 00:21:19

Sounds like you really need to talk

Speaker: 00:21:20

to Matt, given that you're like

Speaker: 00:21:22

streamlining everything and you're

Speaker: 00:21:24

looking at using, you know, and

Speaker: 00:21:26

like, how much of it can you use for

Speaker: 00:21:27

AI?

Speaker: 00:21:28

Well, so after this, I think you and

Speaker: 00:21:29

Matt need to have a conversation,

Speaker: 00:21:30

basically.

Speaker: 00:21:31

So our conversation with Matt was

Speaker: 00:21:34

for me wildly valuable, because

Speaker: 00:21:36

I had never heard of the concept of

Speaker: 00:21:37

AI employees until we chatted with

Speaker: 00:21:39

him. And obviously, it has changed

Speaker: 00:21:41

my worldview.

Speaker: 00:21:42

And it has meant that these are

Speaker: 00:21:44

the OKRs that we're having.

Speaker: 00:21:46

And Concord are ahead of

Speaker: 00:21:48

us on doing it, and the

Speaker: 00:21:50

productivity that he's seeing out of

Speaker: 00:21:51

his team is astounding.

Speaker: 00:21:53

He's also a bit further ahead

Speaker: 00:21:56

than us in that he doesn't have a

Speaker: 00:21:57

CMO and a CRO anymore.

Speaker: 00:21:59

He has go-to-market engineers.

Speaker: 00:22:01

There's definitely a lot of hype

Speaker: 00:22:02

around go- to-market-engineers.

Speaker: 00:22:04

I've seen it on LinkedIn.

Speaker: 00:22:05

It's the job of the future, and

Speaker: 00:22:07

then for some people it's like it's

Speaker: 00:22:09

already the job of the past and

Speaker: 00:22:10

proven wrong. So I think the verdict

Speaker: 00:22:11

is definitely still out for it.

Speaker: 00:22:13

But what is the role

Speaker: 00:22:15

of management?

Speaker: 00:22:17

In the future when everybody

Speaker: 00:22:19

has AI employees and

Speaker: 00:22:21

live coaching is interesting.

Speaker: 00:22:23

The SDR was getting way

Speaker: 00:22:26

better pitch coaching out

Speaker: 00:22:28

of the voice mode of our

Speaker: 00:22:30

CISO GPT than from

Speaker: 00:22:32

me or from a sales

Speaker: 00:22:34

leader.

Speaker: 00:22:35

I think one of the ideas that

Speaker: 00:22:37

Matt had introduced was that

Speaker: 00:22:39

the role of a manager, an actual

Speaker: 00:22:41

human being in this case, being a

Speaker: 00:22:42

manager becomes much more about

Speaker: 00:22:44

being a systems owner and not

Speaker: 00:22:46

being a people coach.

Speaker: 00:22:47

That the people coach is actually

Speaker: 00:22:48

left to the AI part of it, which is

Speaker: 00:22:49

ironic in some ways.

Speaker: 00:22:50

But as a systems' owner, your job is

Speaker: 00:22:53

to create AI

Speaker: 00:22:55

employees to onboard them,

Speaker: 00:22:57

contextualize them, train them.

Speaker: 00:22:59

Iterate them to make them better,

Speaker: 00:23:01

give them performance reviews, give

Speaker: 00:23:02

them kind of OKRs and results

Speaker: 00:23:04

that need to be achieved and so on.

Speaker: 00:23:06

Taking the time and effort to not

Speaker: 00:23:08

just do a one-off kind of GPT

Speaker: 00:23:11

prompt, but in fact care-take

Speaker: 00:23:13

a given GPT

Speaker: 00:23:15

employee to get the most out of

Speaker: 00:23:17

them. That whole mindset shift

Speaker: 00:23:18

is kind of what he'd spoken

Speaker: 00:23:20

about, in particular for creating

Speaker: 00:23:22

coaches. I can see that, yeah,

Speaker: 00:23:24

being quite effective.

Speaker: 00:23:25

Yeah, there's two points

Speaker: 00:23:27

to make. One is,

Speaker: 00:23:29

I think we need to be careful on

Speaker: 00:23:31

what we talk about

Speaker: 00:23:33

for like coaching versus

Speaker: 00:23:35

in a school context, they call it

Speaker: 00:23:36

pastoral care.

Speaker: 00:23:38

So like the check-ins that you're

Speaker: 00:23:39

doing okay, and maybe

Speaker: 00:23:41

not coaching on figuring out a

Speaker: 00:23:43

problem at work, but like the

Speaker: 00:23:45

human contact.

Speaker: 00:23:46

And I think that's still going to be

Speaker: 00:23:47

important.

Speaker: 00:23:48

And it shouldn't be

Speaker: 00:23:50

outsourced to an

Speaker: 00:23:52

AI employee.

Speaker: 00:23:54

You know, as a CEO, I'm worried

Speaker: 00:23:55

about what's the morale

Speaker: 00:23:57

like? How do we bring the team

Speaker: 00:23:59

together?

Speaker: 00:24:00

When is it okay to let off steam?

Speaker: 00:24:03

It's not just tactically,

Speaker: 00:24:05

how do I make a phone call and

Speaker: 00:24:07

do a pitch or this deal is stuck?

Speaker: 00:24:09

How do I unstick it?

Speaker: 00:24:09

I think those things we should

Speaker: 00:24:11

outsource, but we shouldn't

Speaker: 00:24:12

outsource the emotional care

Speaker: 00:24:14

of our employees.

Speaker: 00:24:16

So the coaching that we're referring

Speaker: 00:24:18

to is really technical skills,

Speaker: 00:24:20

competence coaching, as opposed to

Speaker: 00:24:22

more of the employee broadly

Speaker: 00:24:23

speaking around their

Speaker: 00:24:25

life and their concerns and their

Speaker: 00:24:28

kind of challenges that they're

Speaker: 00:24:29

having, is your point.

Speaker: 00:24:30

And I think that there will be

Speaker: 00:24:31

people who want to do that because

Speaker: 00:24:33

you have people that are using

Speaker: 00:24:35

ChatGPT as a therapist.

Speaker: 00:24:37

You have people who are bonding and

Speaker: 00:24:39

having it be their best

Speaker: 00:24:42

friend.

Speaker: 00:24:43

But in a work environment,

Speaker: 00:24:44

I think it's dangerous to lose

Speaker: 00:24:46

the human connection.

Speaker: 00:24:48

And even if difficult conversations

Speaker: 00:24:50

and forming bonds with

Speaker: 00:24:52

humans is something that not all

Speaker: 00:24:53

managers want to, I think is

Speaker: 00:24:55

important for a business

Speaker: 00:24:57

to work well for those human bonds

Speaker: 00:24:58

to still be there.

Speaker: 00:24:59

And then the second point is

Speaker: 00:25:02

around managing

Speaker: 00:25:04

AI employees.

Speaker: 00:25:05

And it's one of the things I've been

Speaker: 00:25:07

thinking about, and I'm going to do

Speaker: 00:25:08

some training and some process

Speaker: 00:25:10

at work because basically

Speaker: 00:25:13

everybody's going to have to become

Speaker: 00:25:14

a manager. And not everybody has

Speaker: 00:25:16

managerial experience.

Speaker: 00:25:18

I was listening to a podcast

Speaker: 00:25:20

where, I think,

Speaker: 00:25:22

I can't remember, Deloitte

Speaker: 00:25:23

Accenture, somebody had

Speaker: 00:25:25

done a thing about AI

Speaker: 00:25:27

adoption within the enterprise.

Speaker: 00:25:30

It was like execs

Speaker: 00:25:32

had the highest adoption, then

Speaker: 00:25:34

middle management the next and then

Speaker: 00:25:36

junior people the least.

Speaker: 00:25:38

And like part of the methodology is

Speaker: 00:25:39

you don't know if junior people

Speaker: 00:25:41

are genuinely not using it or

Speaker: 00:25:43

lying about not using because they

Speaker: 00:25:45

don't what they should use and

Speaker: 00:25:46

shouldn't use and so it's just

Speaker: 00:25:48

safer to say they're not.

Speaker: 00:25:50

So there's probably an element of

Speaker: 00:25:51

that But then there's also an

Speaker: 00:25:52

element of senior

Speaker: 00:25:54

people and managers.

Speaker: 00:25:57

Understand the art of delegation,

Speaker: 00:25:59

understand what can be delegated,

Speaker: 00:26:01

are very good at giving precise

Speaker: 00:26:02

communication.

Speaker: 00:26:04

So part of it is people not

Speaker: 00:26:06

owning up, but the other part of it

Speaker: 00:26:07

is I think that it's actually easier

Speaker: 00:26:10

for senior execs and

Speaker: 00:26:12

managers to adopt AI

Speaker: 00:26:15

because it's a lot of the same

Speaker: 00:26:17

managerial skills.

Speaker: 00:26:18

You understand what you

Speaker: 00:26:20

can delegate, what you cannot

Speaker: 00:26:21

delegate, very good at giving

Speaker: 00:26:23

precise.

Speaker: 00:26:24

Instruction, very good at giving

Speaker: 00:26:26

feedback when the instructions

Speaker: 00:26:27

aren't precise, good at asking

Speaker: 00:26:29

clarifying questions, and

Speaker: 00:26:32

those are all the things that you do

Speaker: 00:26:34

in order to get good results out of

Speaker: 00:26:35

ChatGPT.

Speaker: 00:26:37

And so one of the next

Speaker: 00:26:39

things that we're going to do at

Speaker: 00:26:41

Matomic is a

Speaker: 00:26:43

training session for everyone on

Speaker: 00:26:45

how to think about delegation,

Speaker: 00:26:47

how to give precise questions.

Speaker: 00:26:49

One of the great things about AI

Speaker: 00:26:51

is it doesn't have feelings, so you

Speaker: 00:26:53

don't have to do like.

Speaker: 00:26:54

Was really good, but why

Speaker: 00:26:56

don't you think about this instead

Speaker: 00:26:59

or in addition to what you're doing?

Speaker: 00:27:01

You can just be like, no, I said

Speaker: 00:27:02

this.

Speaker: 00:27:03

Why do you keep doing this?

Speaker: 00:27:05

Walk me through your reasoning and

Speaker: 00:27:07

it'll explain why it's doing the

Speaker: 00:27:08

wrong thing. You're like, ah, no.

Speaker: 00:27:11

This is what I asked and this is

Speaker: 00:27:12

what i mean and do this.

Speaker: 00:27:14

So I think this is a fabulous point

Speaker: 00:27:16

and a point I haven't really heard

Speaker: 00:27:17

before, which is, you're right,

Speaker: 00:27:19

every single employee in the

Speaker: 00:27:20

company, regardless of their level,

Speaker: 00:27:22

has to be a manager to be able to

Speaker: 00:27:24

manage AI employees.

Speaker: 00:27:26

And not everybody's had that

Speaker: 00:27:27

experience yet, so that's part of

Speaker: 00:27:30

training that I'll be doing rather

Speaker: 00:27:31

than bringing somebody in.

Speaker: 00:27:33

And then the next stage after that

Speaker: 00:27:35

is to think through the way that

Speaker: 00:27:37

Matt has performance

Speaker: 00:27:39

reviews, OKRs,

Speaker: 00:27:42

job descriptions for

Speaker: 00:27:44

your AI employees, so that you

Speaker: 00:27:45

actually have a high-performance

Speaker: 00:27:47

team rather than a

Speaker: 00:27:49

team that goes a bit wonky.

Speaker: 00:27:50

I'm thinking back to our previous

Speaker: 00:27:52

conversations before you entered

Speaker: 00:27:53

Atomic and kind of this AI

Speaker: 00:27:56

chats that we had had.

Speaker: 00:27:57

And I feel like between Matt,

Speaker: 00:27:59

the inspiration of Matt and now what

Speaker: 00:28:00

you've actually picked up here, if

Speaker: 00:28:02

you're thinking about an entry

Speaker: 00:28:04

point in your company of

Speaker: 00:28:06

where to start with this stuff,

Speaker: 00:28:07

where you're moving past just using

Speaker: 00:28:08

individually, just using chat GBT

Speaker: 00:28:10

as a bit of a support mechanism for

Speaker: 00:28:12

one-off prompting, which we're all

Speaker: 00:28:13

doing right now, but really the

Speaker: 00:28:15

question as an operator of how you

Speaker: 00:28:17

take this and do something that is

Speaker: 00:28:18

much more.

Speaker: 00:28:20

Durable and evolved within

Speaker: 00:28:22

an organization to really transform

Speaker: 00:28:24

it as opposed to this one-off stuff

Speaker: 00:28:25

that we're currently doing.

Speaker: 00:28:26

This entry point of the mindset

Speaker: 00:28:28

and the idea that these are AI

Speaker: 00:28:30

employees that need to be onboarded,

Speaker: 00:28:33

contextualized, measured,

Speaker: 00:28:35

and so on is maybe the key

Speaker: 00:28:37

mindset distinction of the actual

Speaker: 00:28:39

way in which you can take this AI

Speaker: 00:28:41

stuff in a company right now and

Speaker: 00:28:44

set yourself on a journey.

Speaker: 00:28:46

Yeah, 100%. And I

Speaker: 00:28:48

feel like it's all over LinkedIn,

Speaker: 00:28:49

and we might be a bit late.

Speaker: 00:28:50

But then in reality, when I speak to

Speaker: 00:28:52

people, nobody's even like

Speaker: 00:28:54

my husband has to keep reminding me

Speaker: 00:28:55

that I'm an early adopter.

Speaker: 00:28:57

Yeah, I think there's so I mean I

Speaker: 00:28:59

might be deluding myself here, but I

Speaker: 00:29:00

feel like you are ahead of the game

Speaker: 00:29:02

I think, there's very few mats that

Speaker: 00:29:03

are out there that are doing this

Speaker: 00:29:05

stuff in a very earnest way

Speaker: 00:29:06

I hope so because this is going to

Speaker: 00:29:08

be our competitive advantage and now

Speaker: 00:29:09

sharing it with everybody else, but

Speaker: 00:29:11

you know, it's around how you

Speaker: 00:29:12

actually do it rather than just the

Speaker: 00:29:13

idea.

Speaker: 00:29:14

So let's move on to our conversation

Speaker: 00:29:16

with Matt Lummo,

Speaker: 00:29:18

AI employees.

Speaker: 00:29:23

Have you seen an increase

Speaker: 00:29:25

in productivity per AE?

Speaker: 00:29:27

And by productivity, I mean how much

Speaker: 00:29:30

they are selling, like how much ARR

Speaker: 00:29:32

they're selling per head.

Speaker: 00:29:33

Yes, just to give you some numbers,

Speaker: 00:29:36

but basically I gave the goal to

Speaker: 00:29:38

all our departments in the company

Speaker: 00:29:40

to be able to do at least 3x

Speaker: 00:29:42

productivity over Q1 and

Speaker: 00:29:44

Q2 of 2025.

Speaker: 00:29:45

We reached the 3x productivity

Speaker: 00:29:47

by March basically.

Speaker: 00:29:50

So right now, whether it's on the

Speaker: 00:29:52

sales side or on the

Speaker: 00:29:54

developers, for instance, we're more

Speaker: 00:29:55

in the 5 to 6x productivity.

Speaker: 00:29:58

And currently what we're doing is

Speaker: 00:30:00

we're actually starting to hire a

Speaker: 00:30:01

bunch of AI employees.

Speaker: 00:30:04

And we're really calling them this

Speaker: 00:30:05

way, we actually have a job

Speaker: 00:30:07

description and we even give them a

Speaker: 00:30:08

name. And so this is important

Speaker: 00:30:10

because we're trying right now to

Speaker: 00:30:12

really let our company not

Speaker: 00:30:14

start using AI as a tool,

Speaker: 00:30:16

but more to partner with AI.

Speaker: 00:30:18

And I think this is an important

Speaker: 00:30:20

distinction that people need to

Speaker: 00:30:21

start making because AI is not

Speaker: 00:30:23

a tool. If you're just using a tool

Speaker: 00:30:24

and you expect to have the result

Speaker: 00:30:26

right away, it's not going

Speaker: 00:30:28

to work.

Speaker: 00:30:29

If you want to use AI well,

Speaker: 00:30:31

you have to consider AI as

Speaker: 00:30:33

an employee, meaning that you

Speaker: 00:30:34

can hire the best person on the

Speaker: 00:30:36

market if you need to onboard that

Speaker: 00:30:38

person.

Speaker: 00:30:39

You need to make sure that person

Speaker: 00:30:40

has all the context of your company,

Speaker: 00:30:42

of your deal, of your offering.

Speaker: 00:30:44

And then you need do performance

Speaker: 00:30:45

reviews with that person, making

Speaker: 00:30:47

sure that they're actually

Speaker: 00:30:48

delivering on the job.

Speaker: 00:30:49

And if not, making changes in

Speaker: 00:30:51

the instructions, et cetera.

Speaker: 00:30:53

And so it's interesting because

Speaker: 00:30:54

since we started this initiative,

Speaker: 00:30:56

we've seen that the use of AI has

Speaker: 00:30:58

changed, the

Speaker: 00:31:00

understanding of it has changed, and

Speaker: 00:31:02

we're starting to get better and

Speaker: 00:31:03

better and the results of this.

Speaker: 00:31:04

And we're just at the beginning of

Speaker: 00:31:05

it, right? So my personal goal

Speaker: 00:31:07

right now is by the end of this

Speaker: 00:31:09

year, I was studying you were on the

Speaker: 00:31:10

5x to 6x right now, depending on the

Speaker: 00:31:12

teams. I think we'll be at 10x

Speaker: 00:31:14

compared to the beginning of the

Speaker: 00:31:15

year as we start wrapping up more

Speaker: 00:31:17

and more with new AI colleagues for

Speaker: 00:31:19

everyone.

Speaker: 00:31:20

What tools are you using?

Speaker: 00:31:21

Basically just cloud and chat GPT.

Speaker: 00:31:23

And then any automation like

Speaker: 00:31:25

clay or N8N

Speaker: 00:31:28

or Zapier.

Speaker: 00:31:29

Yeah, we have a lot of things behind

Speaker: 00:31:31

this, but what's interesting is we

Speaker: 00:31:32

actually, the NA8 and the others,

Speaker: 00:31:34

we actually use them less and less

Speaker: 00:31:35

because obviously the capacities of

Speaker: 00:31:37

Shared GPT and Cloud have improved

Speaker: 00:31:39

more and more.

Speaker: 00:31:41

You are able now to have subagents

Speaker: 00:31:42

that can run.

Speaker: 00:31:43

So there is a lot things you can

Speaker: 00:31:44

start doing natively.

Speaker: 00:31:45

And I would expect, honestly, I

Speaker: 00:31:46

think in August SharedGPT

Speaker: 00:31:48

5 will be released.

Speaker: 00:31:49

And I think we're probably going to

Speaker: 00:31:50

see more and more ability for

Speaker: 00:31:52

everyone to just do everything with

Speaker: 00:31:54

directly in their app.

Speaker: 00:31:55

How do you create an employee?

Speaker: 00:31:56

Who creates an employee and then

Speaker: 00:31:58

manages them?

Speaker: 00:31:58

Well, it's very easy.

Speaker: 00:31:59

So it basically it's whoever think

Speaker: 00:32:01

about it, just a normal employee.

Speaker: 00:32:03

So if you have someone is in charge

Speaker: 00:32:04

of marketing, the marketing leader

Speaker: 00:32:06

is going to say, well, who would

Speaker: 00:32:07

hire? Let's say I have unlimited

Speaker: 00:32:09

budget and I can hire as many people

Speaker: 00:32:10

as I want.

Speaker: 00:32:12

What do I need?

Speaker: 00:32:13

And it's very important to think

Speaker: 00:32:14

this way and not thinking that with

Speaker: 00:32:16

one AI tool, you can solve all the

Speaker: 00:32:17

problems. No.

Speaker: 00:32:18

Well, for the case of marketing, for

Speaker: 00:32:19

instance, well, you probably want

Speaker: 00:32:20

someone to write the content of your

Speaker: 00:32:22

website. You also want someone to

Speaker: 00:32:24

write your content of emails,

Speaker: 00:32:25

because it's not the same type of

Speaker: 00:32:27

audience. It's not the same format.

Speaker: 00:32:28

And when it comes to emails,

Speaker: 00:32:30

probably outbound versus

Speaker: 00:32:31

your typical marketing emails, it's

Speaker: 00:32:33

probably not the skill set, right?

Speaker: 00:32:35

And so you just want basically to

Speaker: 00:32:37

divide these roles, this way,

Speaker: 00:32:39

build a job description for them.

Speaker: 00:32:41

Then you need to basically feed the

Speaker: 00:32:42

AI with what we call internally

Speaker: 00:32:44

blueprints, which is all the context

Speaker: 00:32:46

of your company.

Speaker: 00:32:47

So you're offering your customers,

Speaker: 00:32:49

your segmentation, etc.

Speaker: 00:32:50

And then you basically have to

Speaker: 00:32:52

be a bit smart about the prompts, a

Speaker: 00:32:54

lot of iterations.

Speaker: 00:32:55

But eventually you get to a point

Speaker: 00:32:56

where we can build content, for

Speaker: 00:32:58

instance, for a new webpage with

Speaker: 00:33:00

just asking our teammates.

Speaker: 00:33:02

So, you know, we have Emily

Speaker: 00:33:05

and we tell Emily like, well, I need

Speaker: 00:33:06

a new page for this feature we're

Speaker: 00:33:08

launching. Let's about it.

Speaker: 00:33:10

And Emily just dropped the page.

Speaker: 00:33:11

And if she has questions, she will

Speaker: 00:33:12

ask us.

Speaker: 00:33:13

It's like a conversation with a

Speaker: 00:33:14

colleague.

Speaker: 00:33:15

And you'll be surprised by the

Speaker: 00:33:17

quality of the results when you

Speaker: 00:33:18

really spend time onboarding

Speaker: 00:33:20

well your AI tool.

Speaker: 00:33:22

Bethany was just talking about the

Speaker: 00:33:23

fact that product marketing is going

Speaker: 00:33:24

to be replaced by this, essentially,

Speaker: 00:33:27

and my argument back to her was that

Speaker: 00:33:28

I think there's always going to be

Speaker: 00:33:29

that human top-up of it's

Speaker: 00:33:31

going to It always gets to a point

Speaker: 00:33:32

where it's still not quite fit for

Speaker: 00:33:34

purpose basically and somebody has

Speaker: 00:33:35

to look at it with human eyes to

Speaker: 00:33:36

really right size it for whatever

Speaker: 00:33:38

purpose that you have and Bethany's

Speaker: 00:33:40

response to that was well, that's

Speaker: 00:33:41

probably not going to be the case

Speaker: 00:33:42

going forward, Brandon.

Speaker: 00:33:43

I need to hire Emily, the product

Speaker: 00:33:45

marketer here in ChatGBT.

Speaker: 00:33:48

Here's a tremendous amount of

Speaker: 00:33:49

context, blueprints, whatever you

Speaker: 00:33:51

want to call it to get them in a

Speaker: 00:33:52

space where they're highly targeted.

Speaker: 00:33:54

And then feed them your request

Speaker: 00:33:56

that you have for product marketing

Speaker: 00:33:57

purposes, give me copy for the

Speaker: 00:33:58

website or position or whatever I

Speaker: 00:34:00

guess in that case, that's

Speaker: 00:34:01

essentially what you're doing.

Speaker: 00:34:03

Exactly.

Speaker: 00:34:03

And I think, you know, I think

Speaker: 00:34:04

product marketers, many rules like

Speaker: 00:34:06

that are in danger.

Speaker: 00:34:08

But I think what does that mean for

Speaker: 00:34:09

them? For if someone tomorrow is a

Speaker: 00:34:11

product marketer, what should you be

Speaker: 00:34:12

doing? What you should be doing is

Speaker: 00:34:13

really understanding how to build

Speaker: 00:34:15

these.

Speaker: 00:34:16

AI colleagues, because this way

Speaker: 00:34:18

you own the system.

Speaker: 00:34:19

And you will always need someone,

Speaker: 00:34:21

for now, I hope for the next five or

Speaker: 00:34:23

10 years, to actually build that

Speaker: 00:34:25

system because you do need to

Speaker: 00:34:26

onboard that employee.

Speaker: 00:34:27

You do need just to give them the

Speaker: 00:34:29

philosophy of your company when it

Speaker: 00:34:30

comes to the voice and things like

Speaker: 00:34:31

this. It's not something that the AI

Speaker: 00:34:33

should choose.

Speaker: 00:34:34

We still have human beings to do

Speaker: 00:34:35

this. And I kind of tell my

Speaker: 00:34:37

teams this is basically today now,

Speaker: 00:34:39

as they start working with these AI

Speaker: 00:34:41

colleagues they're basically their

Speaker: 00:34:43

own little VP.

Speaker: 00:34:44

It's really what it is. Have a team

Speaker: 00:34:45

of sometimes five, six, ten

Speaker: 00:34:47

colleagues that will be able to run

Speaker: 00:34:49

a different part of their

Speaker: 00:34:51

work, but they still have to

Speaker: 00:34:53

coordinate all of this to make sure

Speaker: 00:34:54

that the work is done according to

Speaker: 00:34:56

the standards of what we want.

Speaker: 00:34:58

And so they become managers, it's

Speaker: 00:35:00

just not real people, it is AI

Speaker: 00:35:02

colleagues.

Speaker: 00:35:02

And so are you using projects for

Speaker: 00:35:05

each, like each colleague is a

Speaker: 00:35:06

project in effect?

Speaker: 00:35:08

Exactly. Yeah, that's the way we

Speaker: 00:35:10

use it in Cloud in particular, I

Speaker: 00:35:11

guess.

Speaker: 00:35:12

And are you using Gemini?

Speaker: 00:35:14

Because Gemini was so bad, I

Speaker: 00:35:16

didn't use it. And now I tried it

Speaker: 00:35:18

the other day and it was actually

Speaker: 00:35:19

quite good.

Speaker: 00:35:19

We use Gemini Mirror on the product

Speaker: 00:35:21

side, not on the go-to-market

Speaker: 00:35:23

teams.

Speaker: 00:35:24

It's changing, and probably Gemini

Speaker: 00:35:26

is going to be out of that very,

Speaker: 00:35:26

very soon. So I think we'll see some

Speaker: 00:35:28

progress. What's really exciting is,

Speaker: 00:35:30

honestly, we've launched that

Speaker: 00:35:31

recently, right?

Speaker: 00:35:32

But we're also doing this because we

Speaker: 00:35:34

know that within a month or two, new

Speaker: 00:35:36

models are coming out.

Speaker: 00:35:37

There is new options all the time

Speaker: 00:35:38

and more things, right.

Speaker: 00:35:40

GPD just released a few weeks ago

Speaker: 00:35:41

the new improvements for

Speaker: 00:35:43

Operator.

Speaker: 00:35:44

And so if things are working

Speaker: 00:35:46

today relatively well for us,

Speaker: 00:35:48

I am absolutely certain that within

Speaker: 00:35:49

a few months it would be probably

Speaker: 00:35:51

even better.

Speaker: 00:35:52

And so I think it's a question of

Speaker: 00:35:54

jumping now on this because

Speaker: 00:35:56

it does require a lot of work to

Speaker: 00:35:58

organize your context to really

Speaker: 00:36:00

understand how to interact with the

Speaker: 00:36:01

AI. But if you do it now,

Speaker: 00:36:03

then when models are getting better,

Speaker: 00:36:05

you're going to be ahead of

Speaker: 00:36:06

everyone. And that's kind of what

Speaker: 00:36:07

we're trying to do.

Speaker: 00:36:08

But salespeople still need to close

Speaker: 00:36:09

deals, right?

Speaker: 00:36:11

They do, and they should keep doing

Speaker: 00:36:12

this, but they should not spend time

Speaker: 00:36:14

anymore sending emails and

Speaker: 00:36:16

follow-ups.

Speaker: 00:36:17

They should not taking any more time

Speaker: 00:36:18

to review an account or do discovery

Speaker: 00:36:21

before a call,

Speaker: 00:36:22

and this should not waste time in a

Speaker: 00:36:24

meeting with a manager coaching them

Speaker: 00:36:26

on their calls.

Speaker: 00:36:27

That's the type of thing that they

Speaker: 00:36:29

should avoid.

Speaker: 00:36:30

I think the way we look at this is

Speaker: 00:36:31

whether it's our customer success

Speaker: 00:36:33

team or sales team, they should

Speaker: 00:36:34

spend most of their time.

Speaker: 00:36:36

And when I say most, it's 80% of

Speaker: 00:36:37

their Thank you for your time.

Speaker: 00:36:38

In touch with customers, whether

Speaker: 00:36:39

it's on a call, for a chat,

Speaker: 00:36:41

there's different ways to interact

Speaker: 00:36:42

with them, right?

Speaker: 00:36:43

But that's the value of a human

Speaker: 00:36:45

being.

Speaker: 00:36:46

It's the relationship that you

Speaker: 00:36:47

create.

Speaker: 00:36:48

And that's one thing that I think AI

Speaker: 00:36:49

isn't going to replace yet.

Speaker: 00:36:51

We'll see in a few years, right, but

Speaker: 00:36:53

I think for now we're safe.

Speaker: 00:36:54

Yeah, sometimes we're safe.

Speaker: 00:36:55

And then I was listening to a

Speaker: 00:36:56

podcast today where they were

Speaker: 00:36:58

talking about chat GPT

Speaker: 00:37:00

therapists and how they're actually

Speaker: 00:37:02

fairly good.

Speaker: 00:37:03

So who knows?

Speaker: 00:37:05

Well, the thing is your therapists

Speaker: 00:37:06

are like AI, right?

Speaker: 00:37:07

If you interact the wrong way with

Speaker: 00:37:09

them, you're not gonna get what you

Speaker: 00:37:10

want. If at the end of the day, you

Speaker: 00:37:11

go see a therapist just because you

Speaker: 00:37:13

want to hear someone telling you

Speaker: 00:37:14

like, that must be hard for you.

Speaker: 00:37:16

Well, sure, but you can also just

Speaker: 00:37:18

talk to your grandmother or to a GPT

Speaker: 00:37:20

and you'll have a lot of the same results.

Speaker: 00:37:21

If you want someone to challenge

Speaker: 00:37:22

you, then you have to find the right

Speaker: 00:37:24

therapist. And that's the thing with

Speaker: 00:37:25

AI. You need to tell the AI that you

Speaker: 00:37:27

wanna be challenged, not, you know,

Speaker: 00:37:28

comforted.

Speaker: 00:37:29

If our listeners can only

Speaker: 00:37:31

take one thing away from

Speaker: 00:37:33

today's conversation,

Speaker: 00:37:35

what would that one thing be?

Speaker: 00:37:37

What we just talked about AI

Speaker: 00:37:38

employees.

Speaker: 00:37:39

I think the misunderstanding still

Speaker: 00:37:40

today in companies and

Speaker: 00:37:42

sometimes tech companies about

Speaker: 00:37:44

what AI can really do.

Speaker: 00:37:46

I think there is a significant gap

Speaker: 00:37:48

right now. And I think that gap is

Speaker: 00:37:50

growing by the day because you're

Speaker: 00:37:51

going to have some companies that

Speaker: 00:37:52

understand this, some companies that

Speaker: 00:37:54

do not.

Speaker: 00:37:55

And I, think we are going to

Speaker: 00:37:57

see a lot of replacements

Speaker: 00:37:59

because once you miss the

Speaker: 00:38:01

train, you're not able to catch up.

Speaker: 00:38:03

And so that's the one thing I would

Speaker: 00:38:04

really encourage everyone to look

Speaker: 00:38:05

into is AI is

Speaker: 00:38:07

10 times more powerful than what

Speaker: 00:38:09

people think it is today.

Speaker: 00:38:11

And there is a lot of things

Speaker: 00:38:13

coming and it's

Speaker: 00:38:15

time to really change and

Speaker: 00:38:17

adapt all the processes and the way

Speaker: 00:38:19

we work.

Speaker: 00:38:21

So on that note, thank you, Matt,

Speaker: 00:38:22

for joining us in the operations

Speaker: 00:38:23

room. If you like what you hear,

Speaker: 00:38:24

please comment or subscribe and we

Speaker: 00:38:26

will see you next week.