AI Workers for the Enterprise

0:00

>> Okay. So hi everyone and welcome to our session.

0:02

I'm super excited to be joined by Sanjana Perulekor,

0:05

the VP of AI Product Marketing at Salesforce.

0:08

Welcome, Sanjana. Thank you so much for having me.

0:12

Thank you for joining us.

0:13

So I'm excited to get your perspective because we've talked to

0:16

a wide variety of folks at today's AI workforce summit.

0:19

We've talked to folks from smaller companies,

0:21

from the VC community.

0:23

What I really want to dive in today is your perspective on how

0:26

enterprises can embrace AI workers in the AI workforce to

0:30

really help have a sound strategy with their implementation.

0:33

So we'd love to kind of zoom out and start at the beginning.

0:36

What are you seeing as far as enterprise companies trying to

0:39

adopt AI and AI workers right now?

0:42

>> Yeah, absolutely. I think the conversation right now on

0:46

the industry is very centered around agents and digital workers and AI workers.

0:52

For me, just zooming out a little bit,

0:55

the AI conversation has been going on for so many years,

0:58

so many, many years, and everybody has been trying to grapple with,

1:02

how do we get value out of AI?

1:05

The key thing that we're seeing with enterprises is that they don't know where

1:09

to start.

1:10

You might think, okay, large enterprises means using AI to solve big problems.

1:16

I want to optimize my entire supply chain or I want to make sure

1:22

that every single application we build is AI powered and conversational.

1:27

The reality is that every company just wants to start small.

1:32

Everything from using AI to create a personalized sales email,

1:36

or to help their service reps increase CSAP by delivering a more targeted

1:41

service experience,

1:42

or creating a chatbot, right?

1:45

Chatbots are something we interact with all the time,

1:47

but a chatbot that's more conversational that can really take action

1:51

automatically in the autonomous.

1:53

The use cases are actually quite so,

1:56

they're very similar to the use cases we've seen for years.

1:59

It's just that AI workers, agents, conversational AI,

2:04

these are just new tools in the toolkit to get started.

2:08

I think that's a great perspective,

2:10

and we do feel like there's this rush to be reactive right now.

2:14

What are you doing for AI workers?

2:15

How are you employing AI workers?

2:16

What's your strategy here?

2:18

For all the enterprise folks who are listening,

2:21

how can they avoid having those knee-jerk reactions

2:24

and building out that strategy that's really impactful in long lasting?

2:29

You talked about identifying the use cases.

2:31

Can you expand on that a little bit more?

2:34

Yeah, I think use cases can feel like such a buzzword, right?

2:40

Use cases are the name of the game, though.

2:42

You don't have a use case,

2:44

you have technology in search of a problem.

2:46

We see that all too often, especially in AI, right?

2:50

AI has been around for a really long time,

2:52

but why haven't people gotten value out of it?

2:54

It's because oftentimes,

2:56

they just want to implement a technology without knowing,

2:58

how am I going to measure the success?

3:00

How am I going to know that this is working?

3:02

I think when companies are thinking about use cases,

3:06

it's important to recognize sometimes the KPIs are the same.

3:11

If you look at the service space,

3:12

it's all about measuring case deflection and response time,

3:16

and case resolution time.

3:17

The KPIs are the same when you're implementing

3:20

an AI use case for service.

3:21

That's not different whether you're using a rules-based engine

3:24

or an AI worker,

3:26

but you've got to look at those KPIs with some consistency

3:29

as they relate to your use case,

3:31

otherwise you're just implementing things to implement them.

3:34

I think as I mentioned, these use cases can be everything

3:38

from as simple as summarizing a record.

3:41

I mean, our most widely adopted AI product right now

3:45

is Slack AI.

3:46

I use it every single day.

3:47

I use it to summarize channels.

3:49

I use it to keep up with everything that I do at work.

3:52

I use it to just keep up with my daily workflow.

3:56

And it's so important to have AI that is in a use case

3:59

that people are going to use every single day

4:01

that they relate to.

4:03

I think that's fantastic.

4:04

In one of our earlier sessions,

4:06

we were talking about learning to treat AI as a teammate

4:09

and not a tool.

4:10

So just something that's kind of riding alongside you

4:12

in your every day to help you be faster and skilled.

4:16

There's a big craze that's going on right now

4:19

and probably this sense of urgency to go

4:21

and just buy all of the AI technology

4:24

you can get your hands on.

4:25

We've talked with a lot of folks today

4:27

who feel the sense of urgency to have a strategy in place

4:29

for how they're implementing AI and AI workers

4:32

would love to get your perspective on to build or to buy.

4:37

Where should people look?

4:38

Or how should people approach kind of investing

4:41

in AI technology?

4:42

That's a great question.

4:44

It isn't all you can eat AI buffet right now.

4:47

Everything feels available, possible, purchasable.

4:52

And it just feels like a really confusing space.

4:56

Everything from the way we're seeing VC funding

4:59

to the way that companies are buying products,

5:02

where they're investing, it's super overwhelming.

5:06

And we get into this build versus buy concept a lot

5:09

because it can be hard to understand,

5:12

should I just sit back and pay for a bunch of compute power

5:16

and pay for a bunch of AI credits

5:19

to use some sort of large language model

5:21

and create my own use case?

5:23

Or do I buy a platform?

5:25

And I think from the Salesforce perspective,

5:27

we have always been customer company.

5:30

We've always been focused on connecting with customers

5:34

and enabling customers to do that

5:35

in new and interesting ways.

5:37

And so for us, AI is about inserting that technology

5:42

into those daily workflows.

5:44

I try to recreate that.

5:45

You try to recreate CRM.

5:47

You try to recreate a large platform.

5:51

Could you do it?

5:52

Sure.

5:54

Should you do it?

5:55

Probably not.

5:56

That is a waste of time and resources

5:58

when you could go faster with something

6:00

that your existing employees can use,

6:03

aka a low-code platform,

6:05

and something that already speaks

6:06

the language of your business.

6:08

Already knows the workflows that are important to you.

6:10

Already has access to the right data

6:12

because why recreate the wheel

6:14

when you can just use a platform

6:16

that does all of that AI goodness for you?

6:19

I'd love to expand on that a little bit

6:21

because in one of our earlier sessions,

6:23

we talked with Matt Millen,

6:24

he's the founder of Reggie,

6:26

and he talked about this concept of agent washing

6:29

to be aware of over the next year,

6:30

which is a lot of AI workers and agents that are out there

6:34

that are saying they can do all of these things,

6:36

but they might not have all the data,

6:38

all the infrastructure that they need

6:40

to be really successful.

6:42

Would love to hear your point of view

6:44

if a company, an enterprise company

6:46

is looking to make a longer-term investment in an AI worker,

6:49

what else should they be looking out for?

6:50

We talked about integration into your data.

6:52

I know trust is obviously always a P1 for Salesforce.

6:55

What are some of those things they should be thinking about?

6:58

Oh, there's so many elements.

7:00

And as a marketer myself,

7:02

the marketing is wild around agents.

7:05

I mean, it's just everywhere.

7:07

And I think there are some really key things to look out for.

7:11

I mean, number one, first and foremost,

7:13

above everything is trust.

7:15

These AI workers, we have no idea

7:18

if we're honest with ourselves

7:19

what the next five, 10, 20 years are gonna look like.

7:22

I mean, we couldn't have predicted the last two years

7:26

with the best AI technology,

7:28

helping us predict things in our business.

7:29

We could have never predicted what LLMs would do.

7:32

And so I think trust just has to be at the center

7:35

of your strategy and trust can take far from Sony.

7:39

I think for us, trust means a few different things.

7:41

It's not sharing our customer data,

7:44

making sure that people know that their data is not our product.

7:47

We don't use their data for any sort of creation

7:50

of our own products.

7:52

It's about keeping our customers' customer data safe.

7:56

So knowing what they can use,

7:57

the latest and greatest,

7:59

coolest cutting-edge AI technology without compromising.

8:03

And it's also in some of the stuff that people don't think about

8:06

as they're seeing these exciting AI innovations.

8:09

It's like testing AI solutions in a sandbox.

8:12

You gotta do that.

8:13

You can't deploy these things in a production

8:14

without testing them first.

8:16

Absolutely have to do that.

8:18

And then there's, of course, governance and regulatory

8:21

kind of measures included there.

8:22

And I think a quick plug from Salesforce

8:26

is one of the things that I've always found

8:28

really inspiring is that AI regulations continue to change.

8:31

We don't know what's gonna happen.

8:33

And so it's important that companies

8:35

are a part of that global conversation

8:37

of how will we audit and regulate these AI systems.

8:40

And Salesforce is very much a part of this

8:43

in our kind of ethical, humane use group at the company,

8:46

which is really cool.

8:47

And so trust is a big part.

8:49

I'd say the other part is making sure

8:52

that you're looking for a platform

8:54

that will scale with your business.

8:55

And what I mean by that is having people

8:58

that you don't have to hire,

9:00

but that already at your company,

9:01

that are already at your company

9:03

that can take advantage of these tools.

9:05

I mean, I remember back in the day,

9:07

my first job out of college, I was a Salesforce admin

9:09

and that was my day job.

9:12

Yeah, my day job was actually, I was a marketing analyst,

9:15

but I was at a teeny time startup

9:16

and being a Salesforce admin was a part of my job.

9:19

And I didn't know it at the time,

9:21

but I was building kind of enterprise-wide

9:24

workflow automation for our entire sales team.

9:27

And our sales team grew like crazy while I was there

9:29

and I was building dashboards and workflows and dew alerts

9:33

and it was small, but when I got to Salesforce

9:36

and I was like, well, I could have been a part of this

9:38

like trailblazer community of people

9:40

that are building out all of these amazing things

9:44

for their business with real ROI with low-code tools.

9:46

And so what inspires me is seeing all of these customers

9:51

diving in the co-pilot, diving in a prompt builder

9:54

and they are building AI.

9:56

They're data scientists, they're Salesforce admins

9:58

and Salesforce developers and they are able to kind of add

10:01

that feather in their cap.

10:02

They are a conversational designer.

10:04

They are a prompt engineer now.

10:07

And I think, you know, whether it's Salesforce platform

10:10

or another, it's important to kind of look at the team

10:13

you have and think about how can I adapt their skillset

10:16

to build on these AI platforms?

10:18

Because that's ROI right there, right?

10:22

And I think the last part, and I will use the most

10:24

foreign term of all, is change management.

10:27

I mean, I've been in AI for my entire career

10:31

and I have seen companies create the coolest stuff

10:36

and if you don't put people at the center, no one cares.

10:39

So I mean, I gave the example of Slack AI.

10:44

I mean, it is designed with the human in mind

10:46

everything from where the buttons are

10:48

to the way I use it every day.

10:51

And the same is true for our co-put.

10:52

I mean, we've really thought through

10:54

how are people gonna use this in their daily workflow

10:57

when they log into Salesforce,

10:58

when they use another application

11:00

and you've gotta bring your employees along for the ride.

11:03

Otherwise you'll build this kind of beautiful thing

11:06

that no one's used and then the ROI

11:08

is somewhere down the road that you can't see

11:09

because no one's using what you've built.

11:11

So there's the three things that really come to mind.

11:14

I think that's really insightful

11:15

and we talk about this all the time internally

11:17

is a year from now what AI technology is shelfware

11:21

and folks have invested in it,

11:23

but they're not using it, they're not seeing the impact.

11:25

And then what's the AI technology

11:26

that's gonna become a teammate

11:28

and it's gonna become part of their daily lives?

11:30

I'd love to get your point of view for your customers

11:33

that are successful with adopting

11:35

kind of this new way of thinking.

11:37

It's really hard to steer a big enterprise, right?

11:40

If you're a small company,

11:41

buy new software, get rid of software,

11:43

you can roll it out to your team quickly.

11:46

Do you have any advice for your customer champions

11:48

who are bullish on AI workers in this AI technology,

11:51

but they don't know how to advocate it for it

11:53

to their manager, to their company?

11:55

I think like anything you have to show people

11:58

what's in it for them.

12:01

That's my personal opinion.

12:02

And I think what I've always believed about AI

12:06

is that we are not at the stage

12:08

where we're just gonna go and replace the human with AI.

12:12

Like if we're all honest, like that's not,

12:14

that's not what's going on, right?

12:16

We are deploying AI in situations

12:19

where humans don't necessarily shine, right?

12:22

What does alerts like what model do?

12:24

It generates language, not logic,

12:27

from a corpus of data, right?

12:29

We're really good at creating logic.

12:31

We're really good at this,

12:32

talking back and forth, empathy,

12:34

like understanding each other.

12:36

But are we really great at digesting

12:38

large amounts of information

12:40

and summarizing that information?

12:41

No, take us some time, right?

12:43

Have to read all the books

12:44

and give my point of view and write it down.

12:47

Not really great at spotting anomalies

12:50

in a large corpus of data either,

12:52

but all of these kind of skills

12:56

are things that AI is really, really, really good at.

12:59

And so if you tell your employees,

13:01

hey, we're gonna take all of that grunt work

13:04

and busy work that frankly, you're not awesome at it.

13:07

And we're gonna make you 100x more awesome at your job

13:10

because we're gonna automate that for you AI.

13:13

That is the best, right?

13:16

There's something in it for them.

13:17

There's a clear use case

13:19

and there's a clear KPI

13:20

'cause it's already tied to the work they're doing.

13:23

So I think when we really focus on the here and now,

13:26

what AI is good at,

13:27

which is frankly the things that we're bad at,

13:30

it creates more of that trust, personally,

13:33

'cause they know it's gonna be good for them.

13:36

I love that.

13:36

And it's cool to think about the journey

13:38

'cause Salesforce has been ahead of the curve.

13:39

I remember when you guys brought Einstein to market

13:42

and you guys have been preaching this for a while,

13:43

but I feel like we're at this inflection point

13:45

where AI is becoming demystified

13:47

and we're realizing just how powerful it can make all of us

13:50

and we're seeing it really put into practice.

13:52

As we wrap up, we'd love to just ask one more question,

13:55

which is how is Salesforce thinking about bringing AI workers

13:58

and agents to market?

13:59

What's next for you guys?

14:00

Well, we have had some announcements lately

14:06

around the agent space.

14:08

We most recently kind of had our Einstein service agent

14:11

out there for folks to see from a marketing perspective.

14:15

Dreamforce is right around the corner

14:18

and you know Salesforce, there's AppStory,

14:21

there's a platform story, so wink, wink.

14:24

There'll be a lot to talk about at Dreamforce,

14:27

but I think when it comes down to

14:29

where we're going as a company,

14:31

our prominent strategy, it has been the same

14:34

since day one when it comes to AI.

14:36

It's AI for CRM and AI that is always gonna be tied

14:40

to the flow of work.

14:41

It's where we see the most value.

14:43

We're not trying to solve every problem

14:45

in the world with AI.

14:46

We want to solve our customer problems that they have.

14:49

With AI.

14:50

And so that's something you will not see a steer away from

14:53

if anything, agents are just another way

14:56

that our customers will get to achieve.

14:58

So I'm super excited for the future.

15:00

Well, awesome.

15:01

Well, this was such a great discussion

15:02

and I hope that everybody who's watching,

15:04

who's in the enterprise space,

15:05

can walk away feeling empowered,

15:07

like they can make an impact in starting

15:08

with those use cases and starting small,

15:10

even if you're big, I think is such a perfect mindset.

15:14

So Sanjana, thank you so much for joining us today

15:16

at AI Workforce Summit.

15:17

And I'm looking forward to the Dreamforce announcements

15:19

and we will see you soon.

15:21

Awesome. Thanks for having me.