Qualified + 16 min

Turn Buyer Signals Into Action with MadKudu AI


In this episode of GTM AI, see how MadKudu's sales intelligence platform can help turn your company's data into action for your sellers.



0:00

[MUSIC]

0:04

Hello everyone and welcome to Go to Market AI.

0:07

I'm your host Sarah McConnell.

0:09

These days, it seems like every company has AI,

0:12

but on this show, we want to go a level deeper so you can see first-hand how

0:15

businesses

0:16

are actually applying AI to solve your business challenges.

0:19

We're going deep into the use cases and getting live demos of

0:22

the latest and greatest in AI technology.

0:24

Today, I'm joined by Frances Ferrero,

0:26

co-founder and chief product officer, Mankudo.

0:29

Frances, welcome to the show.

0:30

>> Yeah, thanks for having me.

0:31

I'm excited to be here.

0:32

>> Yep, super excited to have you.

0:34

So first of all, can you tell us a little bit who is Matt Kudo?

0:37

What do you guys do and who are you helping in the market?

0:40

>> Yeah, so we're a sales intelligence platform and

0:44

our main goal is to help companies turn data into actions for sellers.

0:49

So we typically tend to help the person who owns creating pipeline.

0:54

Actually, I'm just going to steal a quote from

0:57

one of the prospects that we spoke with yesterday.

1:00

I think she summarized pretty well what it is that we do.

1:04

The problem that we solve and she was describing the fact that they have a ton

1:07

of

1:08

tools and a ton of data, they've got Marketo insights, Clearbit,

1:11

Sixth Sense, Salesforce, Tableau, and a bunch of other tools.

1:16

The problem is that they're lacking a command center where everything comes

1:19

into

1:19

one place and they're able to get a 360 of the prospect to make all of this

1:24

data

1:24

actionable by the reps instead of having to log into 17 different tools.

1:28

So Matt Kudo really is the solution that brings all of this together into one

1:33

place.

1:33

And not only just pipes the data in but also makes sense of it by identify what

1:37

's

1:37

really important and how to make an actionable for a rep.

1:41

>> That's amazing.

1:41

As someone who thinks about pipeline every single day,

1:43

I'm really excited to see this demo.

1:46

So with that being said, I would love to jump into the demo and see Matt Kudo

1:49

at work.

1:50

>> Absolutely.

1:52

So we'll focus on three parts of the product where AI is pretty important.

2:00

And bear in mind, this is not meant to be a full on demo of everything the

2:04

product does,

2:04

but rather three little elements where we've implemented AI because I think it

2:08

's

2:09

relevant to the topic of the show today.

2:12

And so essentially for now for this first part, we are viewing Matt Kudo as if

2:18

we were a rep.

2:20

So we're a rep, we're logging into Salesforce.

2:23

Right now we're able to look at all the accounts that are in our instance,

2:28

especially we can look at our book of accounts.

2:30

And let's say I work for Qualified and one of the things I want to do is I want

2:34

to run

2:35

a rip and replace campaign.

2:36

So I'm going to go after some of our competitors.

2:39

Usually running reports in Salesforce is a little bit painful.

2:42

So refs might want to run this with natural language.

2:46

So a rep could come in and say show me accounts that use, let's say, drift,

2:53

because drift is a competitive product, like something that I might want to go

2:57

after.

2:58

And so essentially what this is doing is in the background is looking at all

3:02

the different

3:02

fields that are available in Salesforce, in enrichment tools, in any of the

3:07

data platforms

3:08

that you've connected to Matt Kudo to identify any account that uses the

3:14

technology drift.

3:16

So from there, the rep is basically able to see these are accounts that are

3:20

using one

3:21

of our competitors.

3:22

And then they could say and have a buyer because it's even better if you have

3:28

someone

3:29

who can actually make the decision and already exist in Salesforce.

3:33

So essentially this is like the kind of like typical generative AI use case

3:37

where you're

3:38

reducing the complexity of interaction with the system while keeping the

3:43

ability to run

3:44

pretty complex queries against extensive amounts of data.

3:50

So now what we've done is we've allowed reps to interact with substantial

3:55

amounts of data

3:57

without having to understand which column to use, which field does what.

4:02

And now they're essentially able to go in here and find the buyer for that

4:09

particular

4:10

account.

4:11

That's so cool as someone who builds the reports for our sales team, like a lot

4:15

of times I'm

4:15

helping build those reports, it's been a lot of time in Salesforce helping our

4:18

reps prioritize

4:20

using the natural language processing to be able to help them ask questions is

4:23

something

4:24

that just one makes me my job incredibly easier because I'm not going to build

4:29

those reports

4:30

anymore.

4:31

But also it opens up something that was usually very operationally difficult

4:35

for the entire

4:36

team and now to your point, like I can operationalize this for my entire team.

4:39

They can easily pull these reports for themselves.

4:41

They could then add in something that I own or something.

4:44

But doing it layman's terms in that natural language is just so much easier for

4:48

people

4:48

to grasp.

4:49

So I love this.

4:52

And I think it's an interesting point in time where we have system complexity

4:57

that has

4:57

been increasing pretty dramatically and still is increasing dramatically.

5:01

But for the first time with generative AI, we're able to reduce the complexity

5:05

of interaction

5:06

with the system.

5:07

We're able to increase capabilities without increasing complexity of management

5:12

So that's going to be a boon for any rev ops team essentially because they're

5:16

going to

5:16

be able to do more without meeting either more people or more PhDs to figure

5:21

out how

5:21

to interact with the UIs of the systems.

5:24

Absolutely.

5:25

That's incredible.

5:26

Cool.

5:27

And so a second use case I wanted to show is another thing that reps have to

5:34

deal with.

5:36

So in this case, let's say you have an account that's spiking in engagement.

5:40

So awesome.

5:41

First off, you get a cute little alert from Mad Critter saying, hey, this

5:44

account is showing

5:45

an increase in engagement.

5:46

But now the question is, who should I be reaching out to?

5:49

So you might have a couple people that already exist in Salesforce.

5:53

Then the question is, well, who are the people that I'm missing?

5:56

So what we've built is a level of AI that sits on top of Zoom info, Apollo, and

6:04

LinkedIn

6:05

and sales app to go and look for the right kind of missing contacts in your

6:11

Salesforce

6:12

instance.

6:13

So essentially what it's able to do, it's able to figure out in the past, when

6:16

we look

6:17

at your existing customers, these are the types of people that exist on the

6:22

account.

6:23

Looking at this particular account, these are the people that are missing from

6:26

Salesforce.

6:27

And therefore, these are the people that you should be going after.

6:30

So for us, it turns out that it's a VP of growth.

6:33

It's a VP of sales.

6:35

It's a VP of customer success, like older people that are in the go-to-market

6:39

space and are

6:39

looking to build pipeline either from existing or new business.

6:43

And so again, it's like simplifying the whole process for the rep so that they

6:47

don't have

6:48

to log into Apollo or into Zoom info, figure out what query to run and then

6:53

remember, okay,

6:54

I'm looking for someone, but then having to remember, does this person or the

6:57

existence

6:58

Salesforce or not?

6:59

So it's just like taking a lot of the tedious work that reps have to do and

7:05

simplifying

7:05

it by just looking at historical data and removing three or four steps from the

7:12

life

7:12

of a rep.

7:13

Yeah, absolutely.

7:14

And letting them do it in a system of record that they already spend so much

7:17

time in two

7:17

point now, I'm not bouncing around to other systems.

7:19

I'm not having to go back and reference things.

7:21

It's all done in the same view and in a system that they're already so familiar

7:25

with and

7:25

spend a majority of their day with.

7:27

So this is great.

7:28

Yeah, and it's really about just simplifying the work of a rep.

7:33

Like they just, they don't need to open 17 calves every single time they want

7:37

to send

7:38

the single email and our goal is to make that simpler.

7:41

Totally.

7:42

Their jobs are hard enough as it is.

7:43

I think if anything we can do from a tech standpoint to make it easier, the

7:46

ones out

7:47

there driving pipeline, anything we can do to make it easier, I'm all for.

7:51

Absolutely.

7:53

And so the last part I'll show you is more on the RevOps side of things.

7:57

So one of the elements that you might notice here is like we have this concept

8:01

of who's

8:02

a good fit for the business.

8:04

And essentially this is something that's trained on historical data to

8:07

determine what

8:08

makes for a good customer or not based on who you've closed in the past.

8:14

And so one of the things that the system builds out for our RevOps teams is the

8:20

ability to

8:22

separate good from bad quality leads.

8:27

And so what it does, essentially it builds these little decision trees.

8:31

What's really great about them is that they're very easy to understand.

8:33

So this is saying at this point here, we have a 12.45% conversion rate.

8:40

And now we're saying on one side, we're going to take companies that have more

8:43

than 100

8:44

employees and on this side companies that have less than 100 employees.

8:47

So what we see is that the companies that have more than 100 employees convert

8:51

at a

8:51

slightly higher rate than the folks that are here.

8:56

Then it's going to do another split.

8:58

Okay, now it's like companies that are more than 25 and then it's going to look

9:02

at what

9:02

is the clouds spend that they might have so on and so forth.

9:06

And so essentially what this is doing is looking at all the different

9:10

attributes that

9:11

exist inside your Salesforce to figure out what is the right way to separate

9:16

into homogeneous

9:17

groups, people that are likely to convert from people that are not just simpl

9:21

ifying the

9:21

work of the ICP determination.

9:24

Yeah, this is amazing.

9:25

And I feel like even for companies that are looking to understand their

9:29

business segments,

9:30

like we were just talking about pipeline and forecasting and trying to

9:33

understand like

9:34

what cohort within your accounts, where do you draw the line?

9:39

Like what, where should you make your cohorts to do pipeline modeling?

9:42

And that's where I immediately went with this.

9:44

I'm like, oh, this will help me understand.

9:45

Okay.

9:46

Based on conversion percentages, here's where I need to split my different

9:49

cohorts of my

9:49

accounts and can apply even these conversion percentages to that modeling to

9:53

understand

9:54

okay, like this particular segment where it's over 100 employees, but up to 150

9:59

employees

9:59

has a typical conversion rate of this.

10:01

And I can see where this has a lot of benefits beyond rev-offs because I'm like

10:05

, oh, I could

10:06

use this to help with our line modeling and not guessing at what cohorts I

10:10

should be

10:11

using.

10:12

Yeah, exactly.

10:14

So that's the whole idea, making it simple and visual so that you can, it

10:19

solves the

10:19

blank page problem of it comes up with a segmentation and then it's very easy

10:24

to go

10:25

and tweet things here and there.

10:26

At least you're starting off of something that makes sense.

10:29

That's amazing.

10:30

This is very cool.

10:33

And yeah, that's it for the demo of like three kind of key features that

10:37

leverage AI

10:39

and machine learning with math games.

10:41

Amazing.

10:42

And for instance, I'm assuming for people watching the show, if you're like,

10:43

this was

10:44

great.

10:45

I want to see a full demo.

10:46

They can come to madcuda.com, request a demo and see like the full, the full

10:49

capability

10:49

abilities.

10:50

Absolutely.

10:51

Yes.

10:52

And we even have, we have a demo video on the website if they want to do that

10:56

or they can

10:56

request to talk to someone, we'll give them more in depth of view as to what we

11:00

can do.

11:02

Amazing.

11:03

Well, thank you so much for taking us through that demo.

11:04

I would love to transition into our Q&A section if you're ready.

11:08

Yes.

11:09

Perfect.

11:10

Okay.

11:11

So my first question is how long have you been building the AI that you showed

11:14

today into

11:15

madcuda?

11:16

Thank you.

11:17

So it's been evolving over time, but essentially the AI has always been at the

11:22

very center

11:23

of the company.

11:24

Even the name itself was found through an algorithm.

11:28

Like we struggled to find a name for the company and we ended up writing.

11:32

Yeah.

11:33

So I'm a algorithm to pick the name based on the cost of SEO, the cost of the

11:38

domain.

11:39

We didn't want to end with something that was going to be super competitive and

11:42

yeah,

11:43

having the domain that would cost a ton of money.

11:46

But the core itself, we've been working on this for I would say five years with

11:52

the tree

11:52

system that we showed, the generative AI side.

11:56

It's been about a year as soon as the first kind of readily available GPT

12:02

models were

12:03

out, we started incorporating this because it was very clear that this was

12:07

going to change

12:07

the complexity of interaction with systems like ours.

12:11

And yeah, we're excited to build more of that into the problem.

12:15

I love that you use AI to help pick the name and I will say of all the go to

12:19

market companies

12:20

out there, I love Mad Kudu's name and I've known it for a very long time and

12:23

like heard

12:24

it in the industry.

12:25

Thank you.

12:26

Yeah.

12:27

It's really cool.

12:30

So is what you show today, is this all available to customers right now?

12:34

Yes, it is.

12:36

Perfect.

12:37

And speaking of customers, who are some of Mad Kudu's customers?

12:42

Yeah, we have a lot of different groups of customers.

12:47

I would say on the, there's one group which is on the, I call them the Beat of

12:51

Dev.

12:52

So companies who sell to developers and you could think of like databases,

12:56

companies

12:56

like a Cockroach, MongoDB, then like security companies like a sneak or even

13:02

like core developer

13:04

companies like a Unity.

13:07

So those are extensive users of the platform because they tend to have a

13:12

substantial amount

13:13

of people testing the product for small projects here and there, like

13:17

developers like to spin

13:18

up instances, but they still have a complex enterprise cycle where they have to

13:22

figure

13:22

out who is the decision maker, where the CTO, the CIO is.

13:27

So they're like a typical kind of company is going to have a lot of data from

13:32

different

13:32

places and try to stitch everything together.

13:35

In a similar vein, but maybe on the less technical side like heavy PLG, we have

13:41

the collaboration

13:42

tools.

13:43

If you think of a Lucidchart, Figma, Dropbox, those are all customers that look

13:49

to Mad

13:49

Kudu to identify where are their big pockets of usage within accounts.

13:54

So on the more enterprise plays.

13:57

And then we have the traditional, let's say enterprise cells like a Gong and

14:02

Autodesk

14:03

or an Avallara where you're selling to executives and you potentially don't

14:07

have that kind of

14:08

PLG motion and it's like a longer cell cycles and you want to understand like,

14:13

you know,

14:14

which technologies are they using?

14:15

They use that in your partners and what's the best way to get into an

14:20

opportunity.

14:22

Amazing.

14:23

And my last question for you is what's next on your guys's AI roadmap?

14:27

Where are you taking Mad Kudu's AI in the future?

14:31

So it's either two big components to it.

14:35

One of them is really on the admin side where we are looking to enhance

14:40

capabilities again

14:41

to simplify the interaction with the product, both for our internal users, so

14:46

our support

14:47

team as well as the more RevOps team that tends to configure a lot of this.

14:53

And then for the end user on the cell side, we're building more agents to run

14:59

through

15:00

more of the tedious tasks that reps have to go through, like researching an

15:04

account,

15:05

drafting emails, they soft of all the information that is available.

15:11

Those are some of the big use cases that come to mind.

15:14

And maybe if it is relevant to the folks listening, one of the good frameworks

15:20

that has been

15:21

shared with me that I'm using a lot now is thinking of there's kind of three

15:26

almost personalities

15:27

to AI that you can think about, especially in features.

15:31

It's coach, co-pilot, and agent.

15:33

And the idea is like a coach is really using AI to help you do better.

15:39

And it's identifying a lot of unknown unknowns, like showing you something that

15:43

you might

15:43

not have thought about.

15:45

Whereas co-pilots are more about helping you do something, you're saying, "I

15:48

want to write

15:49

an email, please help me write a better email."

15:51

And then agents are really more about like, "Do this for me."

15:54

Like you give them the task and you expect the full result to be done, like

15:58

researching

15:59

an account or something like that.

16:01

So we're spending a lot of time figuring out where do we need coaches versus

16:06

agents.

16:07

And what's interesting is that reps don't react as well to agents as rev ops

16:13

might.

16:14

So there are specific use cases where you're going to need one more than the

16:18

other in how

16:19

you present the same potential back end technology.

16:22

So it's like a very interesting piece that I think is still being figured out

16:27

by most

16:28

companies out there.

16:29

I love that framework.

16:30

It's such a good encompassing description of all the different stuff we've seen

16:35

on the

16:35

show and what I've seen in products.

16:37

So that's a really great framework.

16:39

Well, Francis, thank you so much for joining us on the show today.

16:42

This was a great demo.

16:43

I love seeing this in action.

16:45

And I also love what you guys are doing at Mad Pudu.

16:47

So thank you so much for joining us today.

16:49

[MUSIC]