Finally Trust Your Data with Clearbit AI

0:00

[MUSIC]

0:05

>> Hello, everyone, and welcome to Go to Market AI,

0:08

the future of your Go to Market tech stack.

0:10

I'm your host, Sarah McConnell.

0:12

These days, it seems like every company has AI,

0:14

but on this show, we want to go a level deeper and show you

0:17

first-hand how businesses are actually applying AI to solve your business

0:21

challenges.

0:22

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

0:25

the latest and greatest in AI technology.

0:28

Today, I'm joined by Matt Sorenson, CEO at Clearbit.

0:31

Matt, welcome. Thank you so much for joining us.

0:34

>> Yeah, thanks, Sarah. I'm very happy to be here.

0:36

>> Okay. Tell us a little bit about Clearbit.

0:39

Who are you guys? Who are you helping?

0:42

What do you guys do in the market?

0:44

>> Yeah, absolutely. Clearbit is a B2B data company,

0:48

meaning we create datasets that are useful for B2B teams,

0:51

specifically marketing teams, sales teams, sales ops,

0:54

that's really the center of our bullseye.

0:58

We create, curate, and maintain datasets

1:02

that basically every company in the world that has a website,

1:05

keep buyers at those companies,

1:07

and then a bunch of different intent signals.

1:09

Help you identify where accounts are ready to buy,

1:11

what high-quality ICP accounts are signing up,

1:15

and basically trying to help you to be companies,

1:17

find the signal in the noise that is

1:19

our massive market that we all play at.

1:21

>> Yeah, and I know at Qualified,

1:23

we are happy Clearbit customers.

1:24

I know we use it in our own product,

1:26

and we find a lot of value in it,

1:27

and you guys helping us sort of unveil

1:29

and understand accounts and who they are and where they're at.

1:31

So I'm excited to get to the demo,

1:34

because I know we use parts of your product,

1:36

but not, I haven't seen some of your new AI functionality,

1:39

which is obviously the whole point of the show.

1:42

So Matt, with that being said,

1:43

I want to get to the good stuff,

1:44

and let's go behind the scenes in Clearbit,

1:46

and learn a little bit about your AI functionality

1:49

and what you guys are doing.

1:50

>> Yeah, absolutely.

1:52

So I think one, I'll take half a step back here,

1:55

and I'm actually relatively recently--

1:56

>> Back to Clearbit.

1:57

>> I'm one of the Clearbit founders,

1:59

and I came back just almost five months ago.

2:02

A few days shy of five months ago,

2:04

specifically to rebuild Clearbit,

2:07

you see LLM, Slarge Language Models.

2:09

LLMs are going to change a lot about

2:13

how would you business work,

2:15

this podcast exists because people have

2:17

so much curiosity about how AI is going to change our work,

2:21

but in the data space,

2:22

LLMs have the ability to change how we do things

2:26

like Berry-Verry.

2:27

They're really good at two things,

2:29

more than two things,

2:31

but two things specifically that impact us.

2:33

One is categorization, categorization.

2:36

So choosing between options,

2:38

and the other is extraction.

2:39

So pulling data attributes out of unstructured text,

2:42

audio, video, et cetera.

2:43

So I'll show you kind of what Clearbit is at its core,

2:49

and then how we're using AI to improve our core data set.

2:53

'Cause that's really where almost all of the AI

2:55

is being applied to our business and our products,

2:58

is on the data itself.

2:59

So here, basic Clearbit lookup,

3:03

let's just do me for sake.

3:05

It's a simple lookup here,

3:13

and we pull back information about me,

3:15

we do this for any email address,

3:17

any domain name, any IP address,

3:19

and a couple things here, so roll and seniority.

3:23

These are pulled out of titles,

3:26

and we're using AI and LLMs

3:28

to recognize all of the variations of roll and seniority

3:32

that you can pull from titles.

3:34

- Oh, that's cool.

3:35

Okay.

3:36

- And so can you question,

3:37

can you push this data into your systems

3:39

so leadership and executive,

3:41

all of this is usable data for your teams?

3:43

- Exactly, so it's basically just structured data

3:46

that pushes down into every single system, every tool.

3:48

And so this data exists in qualified,

3:50

for people that are using qualified,

3:51

it exists sales for a sub-spot wherever you want.

3:54

- I want this as someone who's had to build

3:56

in previous roles in ops,

3:58

the weird formulas of if title contains this and this and this,

4:02

then make their seniority or their role,

4:03

like sales or marketing.

4:05

So being able to use AI to do that is super helpful.

4:09

- Totally, and this is something we've done

4:11

for a long time using traditional machine learning.

4:14

Now with LLMs and specifically vector embeddings,

4:17

we're able to do this at like a scale

4:18

that we've never done before and in any language.

4:22

So previously, there's like a lot of writing,

4:25

rules and mapping rules and red jacks

4:27

to figure out what title equals leadership.

4:30

- Yeah.

4:31

- They, you have a much, much, much more robust version of that.

4:35

You do it in Japanese, you do it in Chinese,

4:36

you do it Arabic, you do it in English,

4:38

and it's pretty amazing, pretty amazing what we can do there.

4:42

- Yeah.

4:43

- So one example, another really good one here

4:45

is all of this stuff down here.

4:48

So basically everything below this line is categorization.

4:52

It is what industry, in any different type

4:55

of industry categorization.

4:56

So whether you use NAICS, whether you use SICK,

4:59

whether you GICS tags, we have over 1500 tags now

5:04

that are like hyper, hyper chosen

5:07

for what B2B companies carry the most about.

5:09

You have technologies, technology categories.

5:12

With LLMs, we can really just expand

5:16

the amount of categorization we do,

5:17

and we can translate it into any other categorization type.

5:22

So if you have a specific set of industries

5:24

that you really care about, this makes it easier and easier.

5:27

- It's a customized app for you.

5:28

- Again, I got it.

5:30

- Oh no, I just, I'm really, I think this is really interesting.

5:33

I think data cleanliness and data clarity

5:36

is just something I think that gets overlooked a lot.

5:38

My GoToMarket team's obviously not by like ops people

5:40

who are in it every single day,

5:42

but I think having this level of granularity

5:45

to Sigma and out audiences as a marketer,

5:46

that's my first thought is I'm like,

5:48

this is incredible because I don't just have

5:50

internet and software services,

5:51

I can go so much deeper and create so much better segments

5:54

for my advertising, for my sales team,

5:55

for my outreach that before took so much manual work

5:59

for my ops team to try to get me these segmentations.

6:02

So I love seeing this on the back end of Clearby

6:04

because this is something that I personally would find

6:06

so valuable for our team.

6:08

- Yeah, it really just expands the service area

6:11

of how much we can do and how customized we can get.

6:13

- Awesome.

6:15

- So this is the basics of what we do.

6:17

Like you kind of said earlier,

6:18

this gets pushed down everywhere.

6:19

So, in our world, this gets sent down

6:24

to Salesforce, HubSpot, Marketo, Pardot.

6:27

You use this data to shorten your forms,

6:29

to fire off sales alerts.

6:32

We use it in our segment and our CDP,

6:35

pulls in G2 data, like we basically,

6:38

if you can think of Clearby as like your data

6:41

data control center, where all your

6:44

beautiful data flows in,

6:45

enriched with hundreds of different data attributes

6:47

by Clearby and then pushed back down

6:49

into your engagement systems or your go-to-market systems.

6:52

- Very cool.

6:53

All right, well, I think in terms of places

6:58

that AI gets used the most, it is mostly in the data.

7:01

There's a couple things I want to show you

7:03

that are coming soon and depending on when this goes out,

7:06

they already be live.

7:09

So, we've built an all new prospecting data set

7:12

and prospecting UI that has kind of AI at its core.

7:17

And we're using AI in a handful of different ways here.

7:20

The biggest way that we're using AI

7:21

is to recommend the right prospects at companies.

7:26

So now we're recommending prospects,

7:29

not just based on your search criteria,

7:32

but also based on the data in your go-to-market systems.

7:35

So, your close one opportunities,

7:36

your lead active leads,

7:39

your target accounts that you may have set

7:41

in the system as well, or your target ICPs.

7:45

And we can recommend based on the website activity patterns

7:50

we're seeing, the CRM patterns we're seeing,

7:52

the buying power, potential new prospects,

7:56

all of this using LMs and some kind of standard

8:01

machine learning that doesn't even feel like AI anymore.

8:04

But to really create personalized and predictive prospecting

8:09

versus just role-senority and employee count.

8:13

- I feel like this would be so helpful

8:15

as like I imagine a new SDR or AE joining the team

8:18

and they have to build up their pipeline

8:20

and they're doing prospecting.

8:21

And I know using ourselves as an example,

8:24

like we have multiple personas that we can sell into

8:26

that are involved in our buying process.

8:29

So having AI and clear bit being able to ingest

8:33

all of the data that a new SDR might not know,

8:36

which is like who is the most common champion on our deals?

8:38

Like who is most likely to influence them?

8:40

Who's on our website the most frequently?

8:42

And recommending that to them would save them so much time

8:45

and like wasted effort because they're outbounded

8:46

to the wrong people or people who aren't buying

8:48

as they're learning and ramping.

8:50

So I have to imagine this helps any SDR and AE ramp time

8:53

to generating pipeline just so much faster.

8:56

- Yeah, and it's like a virtuous circle as well.

8:59

The idea here is the ops team gets to set.

9:03

This is our ICP, this is our target accounts.

9:04

This is who we're going after and that gets fed

9:06

to your plant reps, but also the contacts

9:10

and the prospects that they're choosing to outbound to

9:12

or create gets fed back into the system too of,

9:16

this is who the sales team wants to go after.

9:18

This is who we're watching the behavior

9:21

of our frontline folks.

9:22

I've been calling it in some ways,

9:24

it's like it's Tinder for the reps

9:27

and it's like the Netflix recommendation algorithm

9:30

for the ops team.

9:31

- I love that so much.

9:32

I feel like any of those B2C examples

9:33

are just really understandable.

9:35

And I really like that analogy.

9:37

I do think from a sales reps perspective,

9:41

I just think AI is really cool when it comes to taking out

9:43

variability of error, which I think you have here,

9:47

which is like as a rep, you have all this insight

9:49

and like who's actually responding and who's engaging,

9:51

but then there's all this data

9:52

and bringing those two together can be really, really difficult.

9:54

And once you have large teams,

9:56

there's so much variability in what they think

9:58

and who they think is the right fit.

10:00

Having AI help guide them and sort of create guardrails,

10:04

I think is just so beneficial

10:05

for generating high quality pipeline

10:07

that's actually gonna close into, close when we're having nail.

10:11

- Could not agree more.

10:12

And actually that's a really nice segue.

10:14

I'm gonna show you something

10:15

that we haven't really shown anyone yet.

10:17

It's an internal tool we're working on,

10:20

but it really does a lot of the same things

10:21

that clear what does for our customers,

10:23

which is taking unstructured, vast amounts of unstructured data

10:27

and turning it into structured useful data

10:29

for your go-to-market team.

10:30

- Awesome.

10:32

- So for you guys at Qualified,

10:35

when, what do you guys use for sales qualification?

10:37

You like MedPik or Bant or--

10:39

- I think it's MedPik.

10:40

And if my sales team is listening

10:42

and I said it wrong, I'm so sorry.

10:43

(laughs)

10:45

- And do all of your reps fill out every field

10:49

completely all the time in Salesforce?

10:51

- No, I can almost say that was certainty,

10:54

just any field in general and sales team, we love you.

10:57

What I know is that I know Salesforce Cleanliness

10:59

is not your cup of tea

11:00

or the places you like spending your time.

11:03

- Yeah, and no one does.

11:04

It's like in some way, there's a waste of time,

11:06

but for the ops team or the marketing team

11:08

or product marketing team,

11:10

that stuff becomes really, really valuable,

11:11

especially at scale.

11:13

So we've been building this little internal tool

11:16

that takes, I don't know if this drop down is shown,

11:19

I don't look like it,

11:20

but basically it takes one of our reps,

11:21

all of their different calls,

11:23

and then builds these.

11:25

So it takes the transcripts,

11:28

all these calls, the data from Clearbit

11:31

about the company and the people that are on those calls.

11:33

So what their rules titles are,

11:35

and then the data from Salesforce or HubSpot.

11:37

So you can build a perfect men pick answers,

11:42

so fill them out pretty completely

11:45

based on the output of all these different calls,

11:48

automatically write them down to your CRM,

11:50

Salesforce and R, in our case,

11:52

and build some really, really nice,

11:54

kind of like human readable summaries.

11:57

So what the customer cares about,

11:58

what we know, what they're looking for,

12:01

we've even been playing around with,

12:02

drafting that next email.

12:04

Are every one of our reps is using this now,

12:08

and are enterprise reps,

12:10

or those with the most complicated deals,

12:11

say it saves them somewhere between 60 and 90 minutes a day.

12:16

- I'm very jealous that your reps

12:17

are getting to use this,

12:18

because I know our entire ops team would love it

12:20

if our reps had something that would do their men,

12:23

build in more and better.

12:24

- Another pretty cool thing here

12:27

is because of the way this is built,

12:30

say you changed the qualification framework you were using,

12:34

or added a new field that you really cared about,

12:36

you could run this, run a backfill, basically,

12:39

which runs through everything you call,

12:41

everything you call transcript,

12:42

and pulls out those answers.

12:43

- That's awesome. - Super useful

12:44

for product marketing,

12:46

advertising copy, all that stuff.

12:49

- That's really cool.

12:51

- I think in terms of AI things,

12:54

that's most of what we can show you.

12:56

I'll give you one last little example here.

12:58

So, hey, hey, big Japanese company,

13:02

might be somewhat hard if someone signs up

13:05

with this domain name to basically understand

13:09

what they're doing.

13:10

And so, let's see how clear it does with this.

13:14

Go to look up, live demos are the best.

13:17

- I was going to say live demos are stressful,

13:19

which is why we warn people ahead of the show.

13:21

- You know, exactly.

13:23

So, I think about three ish months ago,

13:25

three, four months ago,

13:27

this would have come back with a Japanese description

13:31

and no categorization.

13:32

- Industry is an everything like that.

13:33

- Because we weren't able to basically translate that.

13:37

- Yeah.

13:38

- Let's give it a go.

13:39

All right, so it has written us a English description

13:44

based on the content of the site, other places.

13:48

We've been able to match it to industries and tags

13:51

and everything else that is kind of dependent

13:53

on us being able to translate there.

13:55

And this feels, might feel like a small thing.

13:58

Translation's been done.

14:00

This works for any website in any language in any country.

14:05

It has increased our international coverage by 10fold.

14:08

Still ticking up the way clear bit works

14:11

is kind of a live creature.

14:12

As people send us requests, we build these profiles

14:15

and this has had such a massive impact on our customers

14:18

and especially some of our larger like international ones.

14:22

That's one of the things I'm most excited about

14:23

that we shipped.

14:24

- I would say this is really cool.

14:25

I feel like especially for global sales teams,

14:27

this has to be so impactful.

14:29

And I think it just goes back to that like

14:30

on leanness of data and time saving

14:32

of not having to do that manually is,

14:35

this is really cool.

14:36

- Yeah, data drives good decisions.

14:40

And in order for that to work,

14:41

the data needs to be clean and standardized.

14:44

And especially, I'm sure you're running into this

14:46

to some extent in this series,

14:48

but all of these different AI tools,

14:51

whether they're co-pilots or chatbots,

14:53

they need consistent structured data

14:56

to have consistent answers.

14:57

And yeah, we're excited to provide that.

15:01

- I think that's, I'm really excited, Matt,

15:03

that you joined and thank you so much

15:04

for taking us through this

15:05

because I do think data cleanliness

15:06

is something that probably gets overlooked pretty frequently.

15:09

I think at least personally,

15:11

when I'm thinking about AI and we're thinking about

15:12

like generative AI and it can help you with all this

15:14

like writing and sending and...

15:16

But when we think about time saving or productivity,

15:19

I think data cleanliness is something

15:20

that doesn't get talked about enough.

15:22

And I know from my end on a go-to-market team,

15:25

bad data is just such a killer for good marketing campaigns

15:28

and effective sales and marketing alignment

15:30

because the data's not there

15:31

and you ask me to pull a report or build segmentation.

15:33

And if I don't have clean data,

15:35

that is just next to impossible to do.

15:38

So I think into this demo,

15:40

I hadn't really thought about AI

15:41

in the context of using it for data cleanliness

15:44

and the impact it can have on productivity or teams

15:46

and how they're pushing really consistent data

15:51

into their systems.

15:52

And I think this could be,

15:54

I see the power of this for any go-to-market team,

15:57

even our own, just thinking about

16:00

how we think about like segmentation and data.

16:02

- Yeah, I think generative AI is amazing

16:06

being able to quickly write and create campaigns,

16:09

emails, et cetera, is very, very cool.

16:11

However, the art of a really, really, really good email

16:15

or really good ad campaign,

16:17

we can get there faster with AI,

16:19

but AI doesn't necessarily solve it.

16:21

What I do think LLMs and well-designed LLM-based systems

16:25

can do is take out so much of the busy work

16:28

of filling out fields, like generating reports,

16:32

pressing send on internal tools and internal reports.

16:35

There's just so much efficiency gain that's possible there

16:39

and it's been really, really fun to work on.

16:41

- Absolutely, and that's why we wanted to do this show

16:43

is I think there are so many different types of AI

16:45

and how they can be used in our Canadian business.

16:48

So I'm glad that in the shows that we've done,

16:50

I think this is a new type of AI

16:51

that we haven't talked about as frequently.

16:52

So it's good, I think, to bring our viewers

16:54

and start to expand our knowledge base

16:57

of all the things that are out there.

16:58

So this was awesome.

17:00

With that being said, Matt,

17:02

is there anything else in the demo that you wanted to show?

17:03

Otherwise, we can move into our lightning round Q&A.

17:06

- I don't think there's too much to show

17:10

from the demo perspective.

17:12

I think for anyone that hasn't,

17:16

it was a current clear-but customer and hasn't yet,

17:19

which I know a lot of our audience is.

17:21

Come say hi, we'll refresh and reenrich your database

17:26

for you with all the new AI data.

17:27

If you haven't done that in a while,

17:29

then we can show you the difference.

17:32

- Awesome.

17:33

Okay, so moving into lightning round Q&A,

17:35

which is how we're gonna wrap this episode today, Matt.

17:37

I have a couple questions for you.

17:39

The first one is, how long have you been building AI

17:42

in the clear bit?

17:43

And I think you touched on this in the beginning

17:44

and I love that you're kind of like a boomerang story

17:46

that you were there in,

17:47

and then you came back and came back just to build this.

17:48

So I would love for you to expand on that.

17:52

- Yeah, I think we've been actively building

17:54

on top of LMs for about four and a half months, maybe five.

17:58

- And awesome.

17:59

And then everything that you just showed,

18:02

I know we already touched on this a little bit,

18:03

but I wanna clarify for viewers on the show,

18:05

what is generally available?

18:07

And it sounds like there's a few things

18:08

that are coming here in the near future.

18:10

- Everything you've seen except the kind of sales call

18:14

notes and summary is live or about to be live,

18:17

depending on when this goes out.

18:20

And all of the data improvements,

18:21

every time we improve the data,

18:22

that's just immediately going out to customers.

18:24

So we're constantly working on that.

18:26

- That's awesome.

18:27

And speaking of customers,

18:28

who are some of your current clear bit customers

18:30

that are benefiting from Clearbit AI?

18:32

- Yeah, so everyone, every single clear bit customer benefits

18:36

from the better data.

18:37

So we qualify as a great example.

18:40

We've got customers across the board.

18:42

We've got the stripes, the Atlassian, the Asanas of the world.

18:45

We've got some, the smaller startup folks as well.

18:50

We've been in this game, this data game for quite some time

18:52

and I feel very lucky with who our customer base is.

18:55

- Yeah, that's awesome.

18:56

You guys really do, I think span the gamut

18:57

of like customers you talked about.

18:59

Some of your AI benefits that you showed there at the end

19:01

that are gonna help these like really large global

19:03

enterprise type teams,

19:04

but then you also have customers down too.

19:06

I know we've been a clear bit customer since we were

19:08

a much, much smaller company.

19:09

So you guys really span, I think the whole customer base

19:12

and who can benefit because I don't know

19:14

who can benefit from clean data.

19:15

So that's awesome.

19:17

And then what is next on your AI roadmap?

19:20

What is your vision for AI at Clearbit?

19:23

- Yeah, I think I showed a bit of a teaser

19:26

with the prospecting data set

19:27

and doing this personalized prospecting kind of at scale.

19:31

I think that's one of the things we're most excited about

19:33

is the ability to build that system out.

19:37

So prospecting is going live,

19:39

kind of being relaunched right about now

19:42

and really excited to keep building on that,

19:45

out that East case functionality throughout the year.

19:49

The kind of deal intelligence that MedBIC

19:52

piece I just showed you that may be a product we launched,

19:54

we're not sure, the internal tool, we really love it.

19:57

Not positive, it'll make it to GA.

20:00

We have a lot of internal tools that we think about

20:02

turning into public products that never make it.

20:04

So we will see and...

20:07

All right, I'm gonna get really nerdy here.

20:11

- I love it.

20:12

- Vector embeddings are untapped and like unappreciated,

20:18

I think in the data world.

20:20

And they allow us to do categorization at a scale

20:26

and a efficiency scale that we just never

20:30

have been able to before.

20:32

So we're now able to build a custom tag

20:37

based on kind of anything you want.

20:39

So one really good example here is like the pains

20:42

that you solve.

20:43

So you can take all of your jobs to be done,

20:45

turn each one into a vector embedding

20:48

and we can run that across the entire B2B world

20:51

but people in companies and tag every company

20:54

and every person with a KNN like similarity score

20:58

to whatever that job to be done that you solve is.

21:02

And we can give you a prospect list

21:05

or a scoring attribute that is completely custom

21:08

to your business.

21:09

So you probably have a good answer for this as a marketer

21:13

but when you're thinking about your target accounts

21:16

and your ICP, give me the paragraph version up with it.

21:21

- Yeah, so we're going after high tech B2B companies

21:23

and they have to have Salesforce

21:25

and our sweet spot is anyone that has web traffic

21:30

over 10,000 a month.

21:32

- All right, so you just gave me the answer

21:34

of someone that's been really trained

21:35

into like attribute based--

21:37

- Yep.

21:39

- 13 scoring, tell me, talk to me like I'm a fifth grader

21:43

and describe or maybe not fifth grader.

21:45

Maybe I'm in high school

21:46

but describe what qualified does for it.

21:49

- Yeah, so we are helping companies in the B2B space

21:53

drive more pipeline from their website.

21:55

So the way I like to describe it too,

21:57

and I won't say a high schooler

21:58

but someone let's say stops by an event booth

22:00

and is like, I've never qualified, what are you doing?

22:03

Essentially anytime someone is on your website,

22:05

we're trying to figure out what's the best way

22:07

to convert them into pipeline.

22:08

And we can do that.

22:09

There are a number of different things,

22:10

whether it's chatbots or live chats or meeting schedulers

22:14

but at the end of the day, when someone is on your website

22:17

and they're showing interest,

22:18

we want to try to fast track them

22:20

in a really personalized way to turn them into

22:24

a new opportunity for your company.

22:26

- So we could take in this new paradigm,

22:30

we can take everything you just said,

22:32

turn that into basically a paragraph,

22:34

fully describes all of the different ways in all

22:37

the different things you might solve for someone

22:39

and then go score the entire B2B world against that.

22:42

So that's websites, that's job postings,

22:44

that's news articles, that's podcasts,

22:47

it's conference,

22:50

keynotes and fine people that are talking about

22:53

the problems that you solve

22:55

and bringing it back that way,

22:57

a much more fluid human understandable

23:00

and communicable way versus saying B2B uses sales force

23:04

has as many employees.

23:06

- That's so cool.

23:06

So to your point, my first answer,

23:08

and I swear for the people listening,

23:09

we did not rehearse this ahead of time,

23:11

I just happened to give the answer

23:12

but not into the learnings there.

23:16

So like as a marketer,

23:17

I am really used to because I have to

23:20

from a segmentation standpoint,

23:21

like I need to know all of the like technographic

23:23

and geographic data sets of what we consider

23:26

our target accounts,

23:27

but what you're saying is with vector embeddings

23:29

and with clear bit AI,

23:31

we could instead take that so much further

23:34

into what people actually care about

23:35

and when they have problems with this,

23:37

so it's gonna help tell me like,

23:38

yeah, you think these are like the technographic attributes

23:41

that you need for your target accounts,

23:42

but instead like here's the actual people in accounts

23:44

that are having these problems

23:47

and much more likely to convert.

23:49

- Yeah, it lets you do like intent based prospecting

23:52

and pain based prospecting using natural language,

23:55

which I think is how most of us communicate

23:57

and think about these things versus the extractions

23:59

that are data attributes.

24:01

- That is really cool.

24:02

And Matt, I am now gonna turn around

24:03

and put you on the spot

24:04

and this also was a plan

24:06

because you said you wanted to nerd out.

24:07

So I'd love for our listeners

24:08

to also be able to nerd out on this.

24:10

Can you give us just a brief description

24:12

of what vector embeddings are?

24:13

So if someone listening, they don't know what that is.

24:14

I'm like, hey, that sounded really cool.

24:17

What does that mean for you?

24:18

- Vector embedding is pretty straightforward.

24:19

It is a string of text

24:21

that has been turned into a numerical representation

24:24

in three dimensional space.

24:25

All that sounds nerdy, but it's just a number.

24:28

It's a number basically that represents a string of text

24:32

that makes it much, much more efficient

24:34

to find other numbers or text that is similar to that embedding.

24:39

- That's really cool.

24:40

Okay, we learn something new on GoToMarket.ai

24:43

every time we do an episode.

24:45

Matt, last question for you.

24:47

Are there any other AI-powered products

24:49

that your GoToMarket team is currently using

24:51

and loving that you wanna tell listeners about?

24:54

- Yeah, absolutely.

24:55

We have been lucky enough.

24:58

We were part of the chat GPT for business

25:01

kind of alpha beta over the last four months or so.

25:05

- Very cool.

25:06

- The team is using that really heavily

25:07

across sales GoToMarketOps

25:11

and, oh, by far and away,

25:15

a co-pilot.

25:16

I think co-pilot on the engineering team,

25:19

we easily added 20 to 30% in efficiency.

25:22

I'm making the number up my VP of engineering,

25:25

my yellow me, but we are moving faster

25:27

as an engineering organ than we have ever before.

25:29

And a huge part of that is thanks to Co-pilot.

25:33

- That's awesome.

25:34

Well, Matt, that is it for our show today.

25:37

Thank you so much for joining.

25:38

I think Clearbit, like I said,

25:40

we've been customers for a long time.

25:41

I've always found so much value

25:43

in what Clearbit offers to us,

25:44

but going behind the scenes here and being able to see

25:46

not only what you're doing in the back end with AI,

25:48

but what's on your roadmap was very enlightening.

25:51

So thank you.

25:52

Thank you so much for joining us today.

25:54

(upbeat music)

25:57

(upbeat music)