HomeVideos

Marc Benioff | All-In Summit 2024

Now Playing

Marc Benioff | All-In Summit 2024

Transcript

1055 segments

0:00

it is the center of the technology world

0:03

right now it's not what Mark did it's

0:06

when he did

0:08

it and the king of the cloud is

0:11

Salesforce please welcome Mark

0:16

Ben one of the things that really

0:18

matters me is having a positive Global

0:20

impact technology is not good or bad

0:23

it's what you do with it that matters in

0:25

your quest to change the world don't

0:27

forget to do something for other people

0:31

and that was a moment in time when I

0:33

said wow when I start a company I'm

0:36

going to make sure that philanthropy and

0:38

giving and generosity and these values

0:41

are in the culture of the company from

0:43

day

0:47

one you want to stay

0:50

here are you want the couch you want to

0:53

sit on the couch where you want to sit

0:55

I'll give you the couch I'll sit

0:57

here you deser you deserve the couch you

1:00

deserve it the big couch

1:03

okay it's little little too close it's

1:07

nice to see you also I I warned you that

1:09

I I'm not the the interviewer in the

1:11

group you but this is you chose me so

1:14

I'm

1:15

honored but you're the nice one oh okay

1:18

thank

1:20

you well it is am I right is he the nice

1:24

one and you're the one that all the

1:26

women really like like I'll talk to my

1:28

friends at dinner there like you know

1:30

sax what's he like he's amazing let's

1:33

just say we're honored to have Mark

1:35

Benny off here and uh truly who's a

1:37

Visionary in the world of software and I

1:40

would say you know there's probably a

1:43

lot of and I thank my mother for writing

1:44

that video by the way is well Mom thank

1:47

you for writing that for me great you

1:50

know in the in the world of business

1:51

offer in particular we don't have that

1:54

many people who you can describe as

1:55

Visionaries but you consistently have

1:57

been one you really it's true

2:01

you got the I think the whole we're

2:02

sitting now on the edge of the couch

2:04

okay here we go maybe we'll end up on

2:06

the floor I don't know what's going to

2:08

happen I'm trying to keep it engag

2:11

around a lot oh okay all

2:15

right is this how it's going to be the

2:17

whole

2:18

time way worse way worse okay all right

2:22

let me finish the this little intro here

2:25

um I forgot where I was so can I just

2:29

before we start you know

2:31

listen so I want to just do something I

2:34

would not normally do and this is like

2:36

going to be a little bit of a thing but

2:37

I just have to do a little Riff on this

2:40

but we just heard some an extraordinary

2:42

presentation on an extraordinary man and

2:45

there's somebody who's amazing that most

2:47

people don't get to hear of and we just

2:48

heard his name quite a few times his

2:50

name is shin yamanaka Yaman nakasan he

2:53

is um based in Kyoto Japan but he works

2:58

halime at UCSF

3:00

and it's amazing what his vision for the

3:02

world is that he thinks basically that

3:05

we're salamanders and we're going to be

3:06

able to regenerate ourselves and that's

3:09

amazing and so I've been friends with

3:12

him maybe for a decade but I fund his

3:14

research and so a lot of these things to

3:17

watch him have these breakthroughs and

3:18

you heard about the yamanaka factors the

3:21

yamanaka factors which are basically

3:23

this idea that

3:24

yamanaka had this breakthrough in Kyoto

3:28

you know basically hanging out there in

3:30

his lab eating the sushi the whole thing

3:32

and

3:33

then boom and he goes if I take these

3:37

four things I can take an ordinary skin

3:40

cell just any little skin cell and turn

3:43

it into a stem cell which is like the

3:46

the heart of human existence and he did

3:49

it and he was able to repeat it and

3:50

repeat it and repeat it and he won the

3:53

Nobel Prize for it pretty cool and then

3:57

he and I'm going to get the

3:58

pronunciation of this wrong but but he

4:00

then was able to take that stem cell put

4:03

it into your eye if you have tacular

4:06

generation and

4:09

boom healed the eye because the eye

4:12

regenerated then he worked with a buddy

4:14

of his in the lab next door and he took

4:17

the same thing took the stem cells and

4:20

he turned it into on a cookie

4:22

sheet and it looked like it was like a

4:25

plastic thing on the cookie sheet it was

4:27

really cool and then he's like took out

4:30

somebody's cornea that was all screwed

4:32

up cut the material out of the cookie

4:35

sheet popped it in the eye and the guy

4:38

could see it was like amazing then he's

4:41

like listen this is

4:44

amazing I bet I can grow a

4:46

brain so he took the step cells and he

4:50

started growing brains called

4:52

organoids and he's like got a cookie

4:54

sheet of brains and I'm like really he's

4:57

like this is amazing look at all the

4:59

brain brains and I's like and then I

5:02

went and saw him and had lunch with him

5:04

and like I'm like what's happening with

5:05

the brains he's like I stopped the

5:07

brains I'm like why did you stop the

5:09

brains I think they can feel the pain

5:12

I'm like oh

5:14

scary then he's like then I said to him

5:17

now what are you doing oh I'm growing

5:21

intestines I'm like whoa intestin is

5:24

that good he's like huge idea I can now

5:27

grow intestines on the cookie sheet and

5:30

taking the you know stem cells I've got

5:31

a whole intestine here and then like he

5:34

can do like turn it into a lab for all

5:36

the horrible things that people get in

5:38

their gut and all these diseases that

5:40

have never been cured but now you have a

5:42

real simulated environment this is an

5:44

incredible person anyway where do you

5:47

want to go with that I'm going I'm going

5:50

with this you got to stay with me trying

5:52

to help bring the energy up in here okay

5:54

listen follow just hold on hold on hold

5:57

on hold on hold wait wait wait this is

6:00

going to get good okay so then I'm like

6:02

you heard the story like at the end they

6:04

said listen how do I get these

6:07

regenerative factors going inside myself

6:09

so UCSF just published research based on

6:12

funding grant that I and others have

6:14

given them and they had a breakthrough

6:16

that the regenerative Factor inside your

6:18

own blood is called

6:20

pf4 and the way you get pf4 and I'm not

6:23

going to get this exactly right because

6:24

you know I'm in software I'm not a

6:26

doctor so just follow with me I thought

6:28

that's what we're going to talk about

6:29

today that I know but I got to tell you

6:31

this cuz I'm got so jacked watching that

6:33

one is it was either that or those crazy

6:36

shots you have backstage I don't know

6:39

okay number one is P4 you get more

6:43

regenerative factors in your body like

6:44

what you know calorie restriction and if

6:47

you know David and I that does not sound

6:50

very

6:51

good two working out with weights also

6:55

not exactly our top thing

7:01

parabiosis do you know what that is so

7:04

parabiosis kind of came out of research

7:06

published a decade ago in New York Times

7:07

and others which is came from salval

7:10

another person I work with at UCSF where

7:13

they took the blood of a young Mouse and

7:16

put it into an old mouse and then the

7:19

old mouse got young again and that was

7:22

moving the pf4 into that old mouse so

7:26

that's and the fourth thing is cloth

7:27

theapy which is a genetic therapy that I

7:30

don't really understand and these four

7:31

things can start to generate more of

7:33

these things inside your body so then

7:35

I'm like getting excited I'm like God I

7:37

have these problems maybe I can

7:38

regenerate different parts of myself

7:41

whatever and so I'm talking to my doctor

7:44

at UCSF because I'm going through my own

7:46

serious problem where I'm like my left

7:48

leg is like a half an inch shorter than

7:50

my right leg and I'm running on the

7:52

treadmill and I'm always ripping my

7:54

achilles ripping ripping and all of a

7:56

sudden my achilles looks like it has a

7:57

dut and in fact I went to UCSF and

8:00

there's like an MRI you know well how

8:03

many of you have had an MRI raise your

8:04

hand so you know what as horrible it is

8:06

anyway you get in this big machine

8:07

they're looking at my Killers they're

8:09

come out they're all like this oh sorry

8:12

about this really horrible and I'm like

8:15

so I kind of took this thought and I'm

8:17

like talking to my doctor I'm like why

8:19

can't we like do use some of this figure

8:21

out what we can do so he's like all

8:24

right I'm come back on Wednesday so I

8:27

come back on Wednesday at 5: CL you know

8:30

I'm in Mission Bay at UCSF and I'm like

8:34

hey Anthony where is everybody we're

8:37

going to talk about it come into my lab

8:39

so I come into the lab they've got like

8:41

a centrifuge there all this stuff going

8:44

on I'm like well this is interesting

8:46

it's like did you work out today yes I'm

8:48

work out do you're following the P4 I'm

8:50

doing it okay this is what we're going

8:53

to do it's going to be very

8:53

straightforward because we have two we

8:55

have two things we can do with you mark

8:57

number one we can just take your kill

8:59

and we bring you into surgery right now

9:01

we'll just shave off half your Achilles

9:03

and then put you in a boot and see where

9:04

you are in 6 months I go doesn't sound

9:07

great second idea what we're going to do

9:11

is we're going

9:13

to we're going to take a scalpel B right

9:17

here we're going to cut into your

9:18

Achilles like 20 times and into your

9:21

ankle I'm going to take your butt I'm

9:22

going to spin it I'm going to try to

9:24

find the pf4 in your plasma I'm going to

9:26

inject it into your Achilles and into

9:28

your plasma slice into it with my

9:31

scalpel I go sounds great goes one

9:34

problem I go with that we don't use

9:35

anesthesia to do that why because it

9:39

destabilizes the PRP and the plasma and

9:41

all the pf4 and all that I'm like let's

9:45

rock let's rock so he did the whole

9:49

thing and then boom like an I'm like a

9:52

salamander they grew me a new Achilles

9:54

right in place so that thing that you

9:56

just heard that is real

10:00

and you know it's pretty awesome what

10:03

can what and uh it's yeah I have a

10:06

question for you um so if he can sell

10:10

three billion into his startup I should

10:11

probably start I'm ready to go got the

10:13

pitch did do you ever consider that you

10:16

missed your calling as a scientific

10:18

researcher definitely not definitely not

10:21

you're happy with the choices you made

10:22

well no that's an incredible story so

10:23

you are one of the first to actually try

10:25

using the yamanaka factors on yourself I

10:29

wouldn't think I'm one of the first but

10:30

I think that it's very real and it's

10:32

going to going to have a huge impact on

10:34

our whole on our lives and I think that

10:36

we should be supporting these medical

10:38

researchers I think it's it's one of the

10:40

reasons that I've you know put almost a

10:42

billion dollars into UCSF of

10:43

philanthropy because I believe in these

10:45

people who have absolutely yeah they've

10:49

dedicated their lives you know to basic

10:53

science and doing and meeting them they

10:55

so inspiring to me and like I just had

10:58

lunch with yamanaka and salval and

11:01

Anthony Luke and another incredible

11:02

researcher Mark moiser at my house and

11:05

like we're talking about the

11:06

intersection between oncology and

11:08

regenerate medicine which is like two

11:10

completely different worlds that don't

11:11

talk to each other and it's what

11:13

inspires me that you know we can you

11:16

know work with others to kind of get

11:19

give them the entrepreneurial push to go

11:20

do something incredible and these people

11:22

are just awesome each one is amazing

11:24

that is that is incredible so let's

11:26

shift gears and talk about something

11:28

else nice coincidence with the yeah no

11:29

it's incredible it's a great story I

11:31

know you you're very philanthropic and

11:33

do a lot with UCSF so uh kudos to you

11:36

for encouraging that type of research uh

11:38

let's shift to to another thing that's

11:39

having a huge impact in our lives which

11:41

is the cloud and software where you were

11:43

a Pioneer you started Salesforce back in

11:46

1999 25 years ago 25 years ago and how

11:50

long have you been a public company for

11:51

at this point

11:54

24 to 20 years 20 years and one of the

11:58

things I noticed count 80 earnings calls

12:01

yeah well actually speaking of earnings

12:03

let's here let's see if we have this

12:04

slide do we have earnings well first oh

12:07

yeah oh boy what's what slide is that

12:10

this is your stock chart over 25 I think

12:13

25 years 20 years yeah you're almostly

12:15

your all I guess there's no linear

12:17

success exactly right yeah really good

12:19

point yeah we had a keep basically had a

12:21

bubble we had a bubble inate 21 we had a

12:24

huge correction in 22 there was I need

12:25

to make a note of that we should talk

12:27

about that but you're your basic back to

12:29

where you were this is one of your

12:30

tweets actually this is one of the

12:32

things I appreciate about the way you do

12:34

earnings calls is you just put out this

12:36

really simple tweet and it just and it

12:38

shows a

12:39

progression and if you know if you like

12:42

looking at numbers the way I do and

12:43

seeing patterns in them one of the

12:45

things I noticed a while ago was that if

12:48

you start at the bottom and work your

12:49

way to the top that Salesforce is

12:51

growing by about 20% a year and if you

12:56

look at it over 3 years that's roughly a

12:58

dou

12:59

so every 3 years Salesforce was doubling

13:02

and that means that over a decade it's

13:04

growing 10x and so every decade is

13:07

basically an exponential if you can

13:09

stick with it long enough that was one

13:11

of the patterns I noticed with sales

13:13

look I I think that you know that the

13:15

growth obviously is incredible to 38

13:17

billion and obviously the cash flow is

13:20

incredible you know it's more than C

13:22

Coca-Cola did I think last quarter and

13:24

the margin is incredible but let me just

13:26

say probably the best decision we made

13:28

in not on the slide which is the day we

13:31

started the

13:32

company um we put 1% of our Equity 1% of

13:37

our profit 1% of our product 1% of all

13:40

of our employees time into a 50613

13:44

foundation now at the time it was very

13:46

easy because we had no employees we had

13:48

no equity we had no profit we had no so

13:51

wasn't very complicated but that idea

13:55

though really kind of created the

13:57

foundation of the company because we're

13:59

able to do now and I think you know the

14:01

numbers right where you know almost 10

14:03

million hours of volunteerism we've been

14:05

able to give away almost a billion in

14:07

Grants we run almost 100,000 nonprofits

14:09

and NOS for free on our service and I

14:12

think it really set the stage that

14:14

business could be the greatest platform

14:15

for change when it came for Salesforce

14:17

it un it gave it that philanthropic

14:19

platform so is there two billion in

14:22

equity sitting in that 501c3 at this

14:24

point a lot well there's more I think

14:27

there's about a half a billion in the

14:28

foundation and a lot of has been already

14:30

given out and then we give out more

14:32

every year and every month every day

14:35

whatever but like on Monday we'll give

14:36

another $25 million approximately to the

14:40

San Francisco and Oakland Public Schools

14:42

and that is you know we've given them

14:44

about $150 million I mean it's it's

14:48

obviously I went to Public Schools it

14:50

was very important to me but all my

14:52

mother was a teacher in the San

14:54

Francisco Public Schools but also our

14:58

employees you know have 75,000 employees

15:00

their kids are in the public schools and

15:03

so it's a key part of our Mantra and our

15:06

culture that we're trying to support

15:09

public education I adopt a public school

15:12

I really think that each one of us can

15:14

needs to focus more on the public educ

15:16

education system in the United States

15:18

it's something I encourage in not all my

15:21

employees but whenever I do a

15:22

presentation I'm like you know my public

15:25

school is like a block for my house

15:27

procedo middle school and I just went

15:28

down there and knocked on the door and

15:32

they're like who are you and I'm

15:34

like what do you how can I help you and

15:37

what can I do to support you they need a

15:39

new playground they need this they need

15:41

that and maybe they just need some me

15:44

some support moral support um but uh

15:47

it's been a great thing to really anchor

15:50

the company in those values and I think

15:52

it's an important thing uh for every

15:54

company so what did you think when you

15:56

saw that open AI started with a non

15:59

profit not as 1% but as 100% but then it

16:01

became a for-profit what did you think

16:03

of that Innovation confusing I you know

16:08

I mean 18,000 companies have now

16:10

followed our 111 model you can find out

16:12

about it at pledge

16:14

1.org that other model I don't really

16:17

understand I think we've proven our

16:18

model this is important you know we came

16:20

out with three models the cloud model

16:22

which you also have been part of that

16:25

the subscription model you've also been

16:27

part of that and the philanthropic model

16:29

and you've been part of that and those

16:31

ideas that we're doing three models

16:33

that's continues to be the fuel for the

16:36

company and extremely important and I

16:38

think that for a lot of these companies

16:40

that have followed us that have gone

16:41

onto scale and have had huge IPOs and

16:44

whether it was slack or whether it was

16:46

at lasan or whether it was eilo or

16:49

whatever they have these huge

16:51

foundations and have had huge impact and

16:54

business can be the greatest platform

16:56

for change and you can do a lot with

16:57

your business and you you know we are

16:59

all building great products okay that's

17:02

great and we're selling them that's

17:05

great too but we can also do a little

17:07

more with our business and we can use it

17:09

in a positive way and try to move the

17:12

world maybe a little bit more in the

17:14

right direction okay so let's talk about

17:16

the cloud part of that Innovation where

17:19

do you think we're at right now I mean

17:20

is it it's is it all AI all the time how

17:24

how are you thinking about it we're at

17:26

the precipice of the greatest moment in

17:29

the history of enterprise software and

17:31

of cloud computing there there's no

17:34

question we you know I had a moment I

17:37

would say more than a decade ago which I

17:39

call my kind of AI freakout moment where

17:41

I really felt I mean maybe it's you know

17:44

obviously we've all spent how many of

17:45

you watched Minority Report all right we

17:49

saw that movie and what about war games

17:51

War Games anybody remember that from

17:53

okay uh

17:54

her um yeah we all saw these movies

17:57

Terminator Okay that one's a little

18:00

scary but we all seen the movies and you

18:03

know like Peter Schwarz who wrote or was

18:05

a key part of writing Minority Report

18:07

and um also war games uh you know as our

18:11

chief futurist at Salesforce and a

18:13

decade more than a decade ago I had this

18:15

moment where I was like okay this is

18:17

really happening here we go and bought a

18:19

bunch of companies and put together

18:21

Einstein and Einstein has done amazing

18:23

you know it's doing trillion

18:25

transactions trillion and a half

18:26

transactions a week predictive

18:28

generative I really thought okay this is

18:30

was going to be the moment but now I'm

18:33

really convinced that we are now really

18:36

at the moment right now where enterprise

18:40

software is going to be completely

18:42

transformed with artificial intelligence

18:44

and we're going to see it and obviously

18:46

I'm getting tuned up for dreamforce

18:49

which is going to be Tuesday of next

18:50

week how many of you are coming to

18:52

dreamforce not enough

18:55

anyway sad these aren't my people I'm

18:58

leaving now

18:59

well it's good you well they look but no

19:01

let me just tell like since you're not

19:03

going to be there let me tell you what's

19:04

going to

19:06

happen thanks for being part of my team

19:09

anyway number one

19:11

is you know we're going to you know we

19:14

really see a moment right now where we

19:17

are 100% focused on one thing and one

19:20

idea and I can tell you why that is if

19:22

you're interested but it's agent force

19:24

and agent force is the most exciting

19:27

thing I have ever worked on in my

19:30

career um it's the culmination really of

19:32

everything that we've done at Salesforce

19:34

because to make agent force really

19:36

deliver we had to have all of our

19:38

customer touch points wired up which we

19:40

do we have to have an Amalgamated data

19:43

Cloud because we need the data

19:44

especially to achieve the AI accuracy

19:47

and the metadata as well and we have to

19:50

have the agents it's these three layers

19:53

that are really going to deliver this

19:55

next generation of capability and I was

19:56

just with Disney last night and Disney

19:59

has agent force they have the newest

20:01

version which we call Atlas which is our

20:03

most accurate not just model but we have

20:05

an extremely unusual technique that

20:07

we'll talk about and Atlas delivers for

20:10

for Disney for their cast members which

20:12

are their employees through extremely

20:16

complex uh problems that it's solving

20:18

for them more than 90% accuracy and

20:21

almost no hallucinations and in some

20:23

cases 95% accuracy and almost no

20:26

hallucinations and that idea that we can

20:29

kind of come into a very difficult and

20:32

complex and sophisticated data set now

20:34

with now with Disney if you go to

20:37

disneystore.com that's Salesforce if you

20:39

go to the Disney parks do you still go

20:41

to Disneyland sometimes yeah okay you

20:42

ever get a Disney guide sometimes yeah

20:45

it's great because you get to cut around

20:46

the lines and all that how many of you

20:48

have done the Disney guides thing we got

20:50

like a lot of poor people here

20:54

actually sad anyway she get these Disney

20:57

guides cuz they're like get you around

20:59

the lines you got do 30 rides a day and

21:02

it's much better than having to wait

21:04

okay but

21:06

anyway Disney guides run on sales force

21:09

they just SL they have slack too they've

21:10

got we do disneystore.com we have Disney

21:13

plus because you know the service now

21:16

like fell over and we had to like

21:18

replace that inside the Disney plus call

21:20

center we have we're do the Disney

21:23

Cruises and the Disney real estate and

21:25

we have every Disney customer test Point

21:27

all wired up so The Amalgamated data set

21:30

that we have around Disney is awesome so

21:32

when we can take that Disney data set

21:34

and then we apply Atlas and agent force

21:37

okay so how do you define we are able to

21:38

deliver a level of accuracy that has

21:41

been incredible and I've got a couple

21:43

more examples I can tell you that are

21:44

just blowing my mind and I never thought

21:46

it was really possible but now it really

21:49

is go ahead yes well I just wanted to

21:52

you want me to ask you a question

21:54

also no no no um what do you mean by

21:59

agent because we're starting to hear

22:00

this term a lot but I think a lot of

22:01

people here may not know what that means

22:03

in the context of AI did you see the

22:05

movie The Matrix yes I did so are we

22:07

talking about agent Smith or what are we

22:08

talking about well we're at some level I

22:10

mean I think like I'll give you an

22:12

example that you know um we're working

22:14

with a large medical company not so far

22:17

away from her Kaiser they've got 20

22:19

million patients they have a super

22:21

complex data set they have all of the

22:23

data from epic they are the largest epic

22:25

customer in the world and more than 90%

22:30

of all patient inquiries and scheduling

22:32

requests and schedule my doctor schedule

22:35

my CT scan my MRI my this my that are

22:38

being resolved by agent force and Atlas

22:41

that idea that we can resolve through a

22:44

autonomous agent a deep and complex

22:48

customer interaction is a breakthrough

22:50

thought obviously we have to do a few

22:52

things to make it really work for our

22:53

customers number one is it's got to be

22:55

trusted because our customers Trust

22:58

we're running the largest banks

23:00

insurance companies media companies cpg

23:02

companies blah blah blah blah blah in

23:04

the world number two is it's got to be

23:06

easy for them it can't be some separate

23:08

team that they're going to spin up it's

23:10

their existing Salesforce team it's

23:12

happening within the Salesforce platform

23:14

it's got to be open it has to have be

23:17

able to work with and interoperate with

23:18

other systems it's going to have to be

23:21

multimodal so it's going to have to

23:22

speak to them and have voice and video

23:25

and do all of those kind of incredible

23:27

capabilities and one other key thing

23:29

because evidently the humans have not

23:32

gone away the doctors have not gone away

23:35

from Kaiser and the cast members have

23:37

not gone away from Disney and on and on

23:40

so we're going to have to handshake

23:42

seamlessly with our apps so even though

23:44

we have all these apps and we've wired

23:46

up all these customer touch points the

23:48

agents are autonomously interacting with

23:51

and building the data and metadata and

23:53

extending it and by the end of this

23:56

month we'll have more than a thousand

23:57

customers on our agent force platform

24:00

the efficiency and productivity that

24:03

we've been had with agent force is like

24:05

nothing I have ever seen with any of our

24:07

customers or technology in the history

24:09

of software but there's a second point

24:12

it isn't just about this kind of ease of

24:14

use it's that that they have the ability

24:16

to do things that are truly astonishing

24:21

and that is also generate Revenue so

24:23

they can go out and like on a day like

24:25

today like it's 10 something degrees

24:27

outside or if you been out there it's

24:28

pretty

24:29

hot and Disneyland may not be as full

24:33

today as it's going to be and they knew

24:34

that was going to be true two days ago

24:36

that a heat wave was coming Disney can

24:38

proactively go out to their consumers

24:41

and their customers and say hey come

24:43

enjoy the heat with us all you know

24:45

Disneyland and we're going to give you a

24:47

special promotion or Price or contest or

24:49

whatever it is to come to Disneyland so

24:52

we want to be able to proactively go out

24:53

and generate revenue and we also want to

24:56

be able to kind of bring that customer

24:58

service in I think last night I had

25:00

dinner at Beverly Hills at the grill

25:02

have you been there great right cream

25:04

spinach well I did something different B

25:06

what do you want what do you prefer um

25:09

well I like like potatoes you know

25:11

potatoes okay po you know any kind of

25:13

potato you like any kind of potato baked

25:15

potato steak fries all so I'm on Open

25:17

Table Right anybody here use Open Table

25:21

not it's very weird group

25:24

anyway

25:26

so oh I'm using anyway you can use Open

25:28

Table to make restaurant

25:31

reservations and there's 160 million

25:34

consumers on Open Table they're not in

25:36

this room but they're somewhere and

25:39

they've got also 60,000 restaurants and

25:42

they've got a lot of complex issues you

25:45

know in regard you know I didn't get my

25:46

table or my food wasn't right my potato

25:48

didn't get cooked whatever it is these

25:51

things are going to get worked out but

25:52

also all of a sudden the restaurant's

25:54

like oh look we're not as full tonight

25:55

as we want to be and we're willing to do

25:57

let's go out to our customer base and

25:59

bring them in but let's do it through a

26:01

complex conversation you know an

26:04

empathic conversation as an agent with

26:06

our customers I think it's going to be a

26:08

rocket ship okay so so how long will it

26:12

be until when you call a customer

26:14

support center you're talking to an AI

26:18

that sounds like a human and you can't

26:19

tell the difference are we there yet or

26:22

we're there yet we are already at that

26:23

point we already have that live and we

26:27

will have that scaled for thousands of

26:30

customers before the end of um live for

26:34

with thousands of customers live before

26:36

the end of this year and we just I just

26:39

demoed it I was just just at a

26:40

conference and spoke uh mile couple

26:42

miles away from here at KPMG and we

26:44

showed them that exact situation where

26:47

you know through you know we used to

26:49

call you know this kind of voice

26:51

response system whatever but you would

26:54

kind of hit a wall pretty quickly with

26:56

your Bot you know but these aren't Bots

26:59

these are not the Bots you're looking

27:01

for these are like we're really getting

27:04

to like another level capability and I

27:07

think that it's pretty impressive and I

27:09

think in the example of Disney you know

27:10

Google has some great products I know

27:11

Serj was here yesterday and they've done

27:13

a great job with AI as you know but in a

27:16

head-to-head Benchmark of sales forces

27:18

agent force against Google's AI uh we

27:22

twox them on accuracy and the reason why

27:25

is we'll explain it next week um you

27:28

know it's a couple of things not only is

27:30

there our NextGen models but it's also

27:32

new techniques involving Next Generation

27:35

retrieval augmented generation rag

27:37

techniques that no one has seen before

27:39

and it's really incredible what's

27:40

possible so you're kicking Google's ass

27:43

I'm cool with that well they're good

27:45

partner also customer and I love them

27:47

but yeah it's it's competitive just yeah

27:49

let me keep let me keep holding on this

27:51

we're trying to all make AI a little

27:52

more accurate and a lot of less a little

27:54

few less hallucinations along the way

27:56

let me give the audience a little update

27:58

about something we just heard at open AI

28:00

they just did a a a day where they

28:03

brought in relatively small number of

28:06

investors and kind of give gave us all a

28:07

update on their product road map and it

28:09

sounds kind of similar because

28:10

everyone's moving in the same direction

28:12

so there are three big takeaways number

28:14

one was that they said that llms would

28:17

soon be at PhD level reasoning right now

28:20

it's more like a smart high school or

28:23

college student in terms of the answers

28:24

we're going to be at the next level

28:26

shortly behind that is agents like

28:29

you're talking about and then third and

28:31

closely related is that agents will have

28:33

the ability to use tools and a tool can

28:36

be a website so if you think about it

28:39

now you've got this llm it's really

28:41

smart it's got you know it's like a

28:43

PhD it you can give it an objective it

28:46

will break that objective into a list of

28:48

tasks and those tasks can include using

28:52

other pieces of software and thanks to

28:55

things like open AI just launched the

28:58

audio API which developers can use it's

29:01

in private beta we have some companies

29:03

using it the llm can now basically

29:06

pretend to be a human and you know

29:08

there's it won't be hard to find a piece

29:10

of software to enable a phone call so

29:12

you can imagine telling a a personal

29:15

assistant agent that and it could be you

29:18

know it could be open table that hey

29:21

book me book me a a dinner reservation

29:23

at the grill and it could place a phone

29:25

call on your behalf and actually talk to

29:27

the grill it could also go on open table

29:29

and just use open table and book it but

29:31

if for some reason that didn't work it

29:33

could literally place a phone call on

29:34

your behalf and the person picking up at

29:37

Open Table wouldn't even know that your

29:39

agent actually isn't a human it's an AI

29:42

but here's where I think it gets really

29:44

crazy is when the phone gets picked up

29:47

on the other end that could be an AI too

29:49

pretending to be a human so you could

29:51

have two AIS pretend to be humans

29:53

talking to each other and resolving

29:55

tasks on your behalf and I I literally I

29:57

think that's where it's at it we're

29:59

definitely moving in this direction but

30:01

there's a cautionary tale here and I

30:03

think that I'll just tell you the real

30:05

world experience with my customers and

30:07

what I'm the problems that I'm trying to

30:08

solve for them I I think in the last few

30:11

years we've kind of heard and you know

30:13

some of it has come from open AI but

30:15

especially from Microsoft that we're in

30:17

this co-pilot world and these co-pilots

30:19

have universally failed the level of

30:22

accuracy the spillage of information the

30:24

lack of trusted environment co-pilot has

30:26

been a complete disaster

30:28

and that idea that this kind of amount

30:31

of Technology got you know released and

30:34

sold into these very large customers

30:36

telling them that the all promise of AI

30:38

is here but didn't do it in a trusted

30:41

way didn't do it with the level of

30:42

accuracy didn't do it with the level of

30:44

security needed and one of the things

30:46

that was interesting because I was with

30:47

one of the customers would trying to do

30:49

this exact technique that you're talking

30:51

about which is a large

30:52

telecommunications company in

30:54

Seattle and what this company did is

30:57

take model of nope going to just tell

31:01

you training a model retraining a model

31:04

building their own model Mark we have to

31:06

have our own models we're going to DIY

31:08

it we're going to DIY our Ai and it's

31:11

going to be awesome then we're going to

31:13

write our own agents and we're going to

31:14

do this we're going to do that the other

31:15

thing and I'm sitting there and I'm

31:17

going through it and whatever and I'm

31:18

then I finally I'm like now show me your

31:19

benchmarks and show me all these

31:21

different pieces and and you know for

31:23

them it's a bit of a science project and

31:24

I've seen this now with a number of our

31:26

customers that they're kind of DIY and

31:28

their Ai and you know DIY I think it's

31:32

fine if you're like Neil Young and it's

31:34

homegrown and it's Canada and it's you

31:36

know Ontario but this is not what you

31:39

should be doing with your artificial

31:40

intelligence but what are you guys using

31:43

as your foundation model is it llama 2

31:45

like what do you guys use we have a lot

31:46

of our own models our own techniques our

31:49

own and then we let you bring in the

31:51

model that you want but we are all about

31:53

achieving your accuracy because what

31:55

I've seen with these kind of approaches

31:57

especially the one that you just

31:58

outlined is that yeah you can get maybe

32:01

30 or 40% accuracy you know in this case

32:04

this customer is 25% you had somebody on

32:07

the stage yesterday I won't tell it is

32:09

who's a common friend of both of ours

32:11

who tried to take this approach for a

32:13

large telecommunications company that he

32:15

owns and he said he was getting about a

32:16

25% accuracy with this homegrown model

32:19

and I'm like why are you doing that

32:21

instead in our platform the platform is

32:23

building the model for you you're not

32:25

having to train and retrain your own

32:27

models you're building your own models

32:29

in our platform and we're going to

32:31

deliver much higher levels of accuracy

32:33

for you and we're going to deliver AI

32:36

this is the AI that you want this is

32:39

this next generation of AI and I think

32:42

that we'll have to prove that with

32:43

benchmarks and with bake offs and to

32:45

show customers because the promise is

32:48

amazing but at a very deep level

32:50

customers are going to need you know

32:52

what you and I have done for the last

32:54

you know 20 years of our life which is

32:55

build professional Enterprise software

32:57

and delivered to them and a capability

32:59

And in regards to an agent running

33:01

enterprise software I mean you just saw

33:03

like that was the

33:04

fundamental business model of adept

33:06

which was David Lewan's company you know

33:09

and that's he built gpt3 then he left

33:11

open AI to start Adept and this idea to

33:13

build agents that are going to drive

33:15

apps I'm sure that all of those things

33:17

are going to happen but again you have

33:20

to get to a level of accuracy because

33:22

everyone in this room and you and I

33:24

we've all had this experience where on

33:26

these models and it's like this is not

33:29

really more than hallucinations and

33:32

that's no good or as we say here in Los

33:34

Angeles no es bueno when it comes to

33:38

okay Kaiser and you're dealing with

33:40

health care you know when you're dealing

33:42

with health care and you've got a

33:43

patient and you're reading their medical

33:45

records you better be delivering more

33:47

than 90 or 95% accuracy because the 50%

33:50

accuracy thing is no good well I can see

33:54

you're ready for dream Force I'm trying

33:55

to go find it I'm testing mat out here a

33:58

little bit what do you guys think I'm

34:04

like are you are you guys excited for

34:06

the rise of Agents yeah it's going to be

34:10

a really big deal I think that

34:12

everything we've seen so far with llms

34:14

has been again about reasoning and and

34:17

generating but with agents the AI is

34:20

going to be able to take actions and and

34:22

they're going to know how to use tools

34:24

which until now it's been some the only

34:25

human I got to tell you a really good

34:26

story because you're like inspiring me

34:28

around uh you know Steve Jobs had a huge

34:32

impact on my life and at I worked at

34:34

Apple in 1984 when I was a I was uh in

34:37

high school and coming into college and

34:40

I was an assembly language program I

34:42

wrote the first Native Assembly Language

34:44

um on this Macintosh know the 68,000

34:48

assembler and sitting there in the cubes

34:50

and Steve is running whatever it is and

34:54

thank God you know I have this

34:55

relationship and influenced me so much

34:57

in my life and then called me on a

34:59

series of times and after I started

35:01

Salesforce gave me really key advice

35:03

anyway it was 2010 and he calls me come

35:07

down here I need to talk to you I'm like

35:09

what the hell what did I do this

35:11

time so I go down there to his office

35:13

and I always bring a few Salesforce

35:15

employees with me and I've got some

35:16

great folks with me and we're sitting

35:19

there and he's like I'm going to show

35:20

you this and I'm like all right let's go

35:23

and he brings out the iPad and he's got

35:25

two of them he's got the big one and the

35:27

small one

35:28

and he's like yeah Mark here it is you

35:30

know but I don't like the small one I'm

35:32

only going to have one size you know

35:34

that I'm like yes sir and he's like uh

35:37

listen you know I've been working on

35:39

this concept for a long time and you

35:40

know in

35:42

2007 um I introduced a iPhone and you

35:45

know I said thank you for sending me one

35:47

I love it it's great like but do you

35:49

know why now we're doing the iPad I'm

35:51

like no because I know you had that too

35:54

in 2007 oh yeah but you know what the

35:56

real situation here is that couple I'm

35:58

like what is it Steve he's like we only

36:00

have one a team here one a team so we're

36:03

only focused on one thing at a time and

36:05

then he lays out like five or six

36:08

products on his coffee table and he goes

36:10

and we will never have more products

36:11

than can fit on my coffee table and I'm

36:14

like well that's really awesome and he's

36:16

like I've been focused on 2007 on the

36:18

iPhone and now I'm gonna zero in and I'm

36:20

only going to do iPad one focus at a

36:24

time remember that Mark that's the way

36:26

you need to rent

36:28

Salesforce and I'm like okay is that why

36:31

you brought me down here yes you may

36:34

go and and that's how I feel right now

36:38

about agent force this is all I am doing

36:40

just try to take our company you know we

36:42

have a great company 38 billion Revenue

36:46

75,000 employees hundreds of thousands

36:48

of customers and one Focus agent force

36:51

this is the because of what you're

36:53

saying this is the moment this is the

36:56

greatest opport opportunity in the

36:58

history of enterprise software and it

37:00

must be executed with absolute Acuity

37:02

and excellence and that is what I think

37:05

we all need to

37:08

do you know so so I agree with you I

37:12

mean I think the agents are going to be

37:14

huge and Elon say something kind of

37:16

similar to the other day um he said we

37:18

was we got him talking about Optimus you

37:20

know his his robot heard about the farm

37:22

animals I didn't know about the what was

37:24

was there another part of the

37:25

presentation he said well he was talking

37:26

about he's talking about Optimist and um

37:30

oh you the the thing thing about these

37:32

jokes are all each one is kind of dying

37:34

very fast it's sad took me a second to

37:36

to realize that you were talking about

37:38

his

37:40

um but now I got it okay um no he was

37:44

referring to um it's great how you bring

37:46

this humor into uh the Allin yes it's

37:49

very subtle I understand

37:52

um so what Elon mentioned that really

37:55

stuck with me is he said that humanoid

37:58

robots the creation of these humanoid

37:59

robots are the biggest Economic

38:01

Opportunity in the history of the world

38:05

the average person is he making some of

38:06

them by any chance he is but well it's

38:09

kind of like you saying that uh agents

38:11

are the biggest opportunity in the

38:12

history of Enterprise sofware let me

38:14

write that down thank you for letting me

38:16

know that it's it strikes me that

38:18

there's something similar here which

38:19

isan is a good salesman is that your

38:20

point well I'm I'm saying there's an

38:22

analogy here between he didn't give you

38:25

the regenerative pitch is that my

38:28

well no what here me the point is this

38:30

is that is that where we're going with

38:33

AI is it's going to be able to take real

38:35

actions and in the case of Optimus is in

38:37

the physical world and it's going to be

38:38

the brain for these humanoid robots in

38:41

the Enterprise it's basically the brain

38:43

for these agents I think these things

38:45

are actually pretty they're on Parallel

38:49

tracks I wouldn't say they're competing

38:51

and these are the droids you're looking

38:53

for so I think anyway I think that uh I

38:56

I think you're right about this

38:57

opportunity and what I'm saying is I

38:59

think it's analogous to what Elon is

39:00

seeing with robotics I think there's no

39:03

question and I think that for our

39:05

customers they're going to augment their

39:07

employees they're going to make things

39:08

lower cost they're going to increase

39:09

their revenues they're going to increase

39:11

their margins we're going to take some

39:12

customers and just turn them into margin

39:14

machines and I think that the

39:17

opportunity in the Enterprise is

39:18

unbelievable he's also directly

39:20

addressing the consumer Market which I

39:22

think is very exciting obviously he's an

39:24

expert in that area and yeah we're about

39:27

to move into this new world of AI of

39:29

droids of all these things and you know

39:32

it's a bunch of waves of you know where

39:34

you know look technology is getting

39:36

lower cost and easier to use it's a

39:39

Continuum and we're all rioting that

39:41

Continuum this is extremely important

39:43

but also what's very important is

39:45

especially as we move into this we all

39:48

have to think about what are the values

39:50

that are going to guide this technology

39:52

because each of us have seen the movies

39:54

we all watched the movies that was the

39:56

one place where I got the hands to go up

39:58

right so we know how it can go really

40:00

wrong right everybody saw that part of

40:02

the movie so what are the values what's

40:06

going to be really important to us will

40:07

it be trust is it customer success is it

40:09

Innovation is it equality is it

40:12

sustainability is what are the values as

40:14

we kind of guide into the next level of

40:16

the future because those core values

40:18

that we need to manifest and really

40:20

focus on that is I think still out there

40:23

as a major discussion item it's got to

40:25

be figured out and that is why we're

40:27

very lucky that you are one of the great

40:29

Visionaries of our industry because

40:31

you're not just a great entrepreneur and

40:33

CEO but you're a great human being so

40:35

thank you

40:37

Mark thank you

Interactive Summary

This video features an interview with Salesforce CEO Mark Benioff, discussing his philanthropic work in medical research, the evolution of Salesforce as a company, and the future of artificial intelligence. Benioff shares a personal anecdote about his innovative treatment for an Achilles injury using stem cell research, then transitions to discuss how Salesforce is focusing on 'Agentforce'—an AI platform designed for autonomous enterprise tasks. He emphasizes the importance of trust and accuracy in AI, distinguishing his approach from 'DIY' methods that often result in high error rates. The discussion concludes with a reflection on leadership, values in technology, and the future of AI agents in the enterprise space.

Suggested questions

3 ready-made prompts