0:00

Bob Safian here with a hard

0:02

truth about business. Oftentimes, saving

0:04

money just isn't enough, which

0:07

is why the most innovative leaders

0:09

today are reimagining how financial

0:11

strategy drives growth. Enter

0:13

Brex, a modern finance platform

0:15

supporting corporate cards, banking,

0:17

expense management, and more. From

0:19

startups extending their runway to

0:22

established companies optimizing every dollar,

0:24

Brex is more than a

0:26

tool, it's a growth accelerator.

0:28

Over 30 ,000 companies are

0:30

already proving it. Discover your

0:32

company's peak performance at brex.com. Hey

0:35

folks, Jeff Berman here,

0:37

co -host of Masters

0:39

of Scale. In a

0:41

time when we are riding a

0:44

roller coaster of news, it really

0:46

pays to take a big step

0:48

back, see the bigger picture,

0:50

then chart the course forward. At

0:53

this year's Masters of Scale Summit,

0:55

we're doing just that. By bringing together

0:57

bold leaders who are shaping the

0:59

future, we all want to see. Join

1:02

speakers like Reid Hoffman, Kara

1:04

Swisher, Andrew Ross Sorkin, Aza Raskin,

1:06

Derek Triseater, and many, many

1:09

more this October in San Francisco.

1:11

Apply to attend at

1:14

mastersofscale.com slash response. Again,

1:17

that's mastersofscale.com

1:19

slash response. Hey,

1:22

everyone, Bob Safian here. Today,

1:24

we're sharing something special and a

1:26

little different for rapid response. It's

1:29

an episode of our sister show,

1:31

Pioneers of AI. Every week,

1:33

host, Rana El -Calubi talks to

1:35

some of the most innovative and

1:37

inspiring AI researchers, entrepreneurs, and investors.

1:39

As a pioneer herself in the

1:41

field of emotion AI and a

1:43

CEO turned VC, Rana cares

1:46

deeply about how AI is

1:48

changing life and work as we

1:50

know it. For the episode,

1:52

we're sharing Today, Rana talks with

1:54

a startup founder who's exploring

1:56

AI for so -called un -bossing, turning

1:59

some management tasks over to

2:01

bots, effectively reshaping how we approach

2:03

middle management and human resources. It's

2:06

a bold, provocative notion. As

2:08

you listen, think about your own work

2:10

and whether at times it might be better

2:13

to seek advice from AI instead of

2:15

from your actual boss. or as

2:17

a boss yourself, whether you let

2:19

AI give advice on your behalf. There's

2:21

lots to chew on, so let's

2:24

get to it. Here's Pioneers

2:26

of AI with Rana L.

2:28

Kalyubi. Picture

2:31

the best manager you've had to

2:33

date. The one who gave

2:35

you the perfect advice, challenged you

2:37

to be better, maybe even

2:39

helped you jumpstart your career. It's

2:41

a rare thing to have a manager

2:43

who can also be a mentor and champion

2:45

of your personal and professional growth. On

2:48

the other end of the spectrum, we've

2:50

all had a bad manager. Those

2:52

that micromanage, parachute in and out,

2:55

or lack communication skills. And

2:57

it's no slight to the individual. Managing

3:00

is often a thankless, difficult

3:02

job, which makes it a role

3:04

ripe for improvement. AI could

3:06

help. I

3:09

have seen some recent study

3:11

that 41 % of Gen Z

3:13

trust AI more than their

3:15

manager. So I think

3:17

one misconception is that people don't really

3:19

want to use it or don't trust it.

3:21

I think actually they do and they're

3:23

using it whether you like it or not.

3:27

Catherine Vaughnian is watching

3:29

this trend closely. She's

3:31

co -founder and CEO of

3:33

Tough Day. A company she

3:35

started after decades working in

3:37

people management at big tech

3:39

companies, including Salesforce. She's

3:41

building AI managers to help

3:43

professionals thrive. Her advice? Try

3:46

new AI tools right now

3:48

at every level. People are

3:50

in sort of analysis paralysis

3:52

over what we should do with

3:54

AI. And I would say

3:56

start using AI, thinking

3:59

through unintended consequences in advance

4:01

as best you can and prepare

4:03

for those is great. And you

4:05

do that alongside the team that

4:07

is actually trying to make the

4:09

solution work. Today,

4:12

Catherine and I are digging into

4:14

questions around the future of work

4:16

and AI. We talk

4:18

about the concept of unbossing, approaches

4:20

to DEI work at this critical

4:22

moment in time, and how AI could

4:24

be a solution to how we

4:26

work moving forward. I'm

4:30

Rana El -Calubi, and

4:32

this is Pioneers

4:34

of AI, a

4:36

podcast taking you behind the scenes

4:39

of the AI revolution. Hi,

4:47

Catherine. Welcome to Pioneers of AI.

4:50

Hi, Rana. I'm excited to be here. Thank

4:52

you for having me. Catherine

4:54

has been in management at tech

4:56

companies for decades. And in

4:58

all those years, she gained some valuable

5:00

insight on how technology can improve

5:02

our work lives. A couple

5:04

of years ago, she saw an opportunity,

5:06

which eventually led her to co -founding

5:08

her company Tough Day. So

5:10

I wanted to know more about her background

5:12

and the aha moment that led her to

5:14

founding her company. I

5:17

started my career at Lotus back

5:19

in the 90s working on collaborative technology

5:21

at a time when the internet

5:23

was brand new for business and trying

5:25

to figure out how that technology

5:27

would empower people to work better and

5:29

smarter and create value and all

5:31

of those things. You know,

5:33

I've rode the tech waves, mobile,

5:36

social, AI. I've been working in

5:38

AI since 2012. So

5:40

I guess, you know, the last

5:42

10 years before starting Tufte,

5:44

I was at Salesforce and I

5:46

was leading innovation projects there

5:48

internally and externally with customers and

5:50

partners and serving our ELT

5:52

over that time and the leadership

5:54

team, ELT leadership team. And

5:57

along the way, one of

5:59

projects that we did internally

6:01

was how to improve belonging

6:03

and inclusion inside using tech.

6:06

And of course, that work

6:08

starts by deeply understanding and

6:10

researching what the challenges are

6:12

for folks internally, not only

6:14

inside Salesforce, but in the

6:16

world. And that research

6:18

was fascinating. We learned a lot

6:20

about what happens when people

6:23

struggle and where they turn. So

6:25

there are two kind of

6:27

common paths in that. One is

6:29

that they go to some

6:31

kind of reporting system internally and

6:33

actually that's quite rare. Only

6:36

about half the people will ever

6:38

report something that actually needs

6:40

to be reported and there's a

6:42

lot more that happens before

6:44

then. So that solution typically means

6:47

we're going into investigation and

6:49

that is scary for everybody. And

6:51

then the other place that people

6:53

go are colleagues and often

6:56

friends and family at home. That's

6:58

what we hear the most even

7:00

today. So when you go to

7:02

friends and family at home, you're

7:04

not necessarily getting great advice. You're

7:06

not necessarily talking to someone who

7:08

knows HR or knows management best

7:10

practices or employment law or any

7:12

of those things. So the big

7:14

insight was that we need something

7:16

before you actually talk to someone

7:18

in HR about a problem, a

7:20

place where people can go. And

7:23

we created that there in something

7:25

called the warm line. A

7:28

warm line instead of, you know,

7:30

a hotline. A place for people

7:32

to call when something felt off

7:34

but hadn't reached any kind of

7:36

critical stage. So

7:39

when we built this warm

7:41

line and staffed it with

7:43

amazing humans who were really

7:45

smart on the topic of

7:48

DEI and lots of other

7:50

things, HR and management, they

7:52

were able to handle a small

7:55

population. The initial cohort got the

7:57

answers that they needed. They got

7:59

guidance. They got help. They got

8:01

empathy. They were able to go

8:03

back to work very quickly. Their

8:05

performance improved. It was a magical

8:07

story. And unfortunately,

8:10

that solution doesn't scale. It

8:13

doesn't scale because the warm line

8:15

needed to be staffed by humans with

8:17

the right skill set. Catherine and

8:19

her team simply could not find and

8:21

train enough people with the right

8:23

mix of skills and experience to serve

8:25

every person at the company. So

8:28

we know that everyone in the

8:30

organization has these moments where they need

8:32

help and the only way to

8:34

scale the solution so that everyone gets

8:36

the help they need is to

8:39

use AI. And

8:41

here was the opportunity. Imagine

8:43

an AI people manager that could

8:45

help employees express their personal

8:47

challenges and concerns and jump in

8:49

on problems so they didn't

8:51

need to escalate. That's what

8:53

Catherine set out to build. It

8:55

was just at the time when

8:57

ChatGPT was born and I knew that

9:00

that technology could be used to

9:02

to do things differently and support workers.

9:05

And I did a couple experimental

9:07

projects on my own, got

9:09

a couple teams together. And

9:11

what we learned in that

9:13

is that actually we could

9:15

get great content from HR

9:18

management, employment law, and create

9:20

this experience with AI and

9:22

be able to serve everyone

9:24

even more safely and consistently

9:26

and expertly than most humans

9:28

can. So let's

9:30

get into exactly what Tough Day does.

9:33

Walk us through how an experience would

9:35

look like for a person using Tough

9:37

Day. Yeah. So first, we

9:39

have created both generative and

9:41

agentic AI to support workers.

9:44

And it is a B2B sale.

9:46

So we're selling to organizations. And

9:48

essentially, the core product has

9:51

already been fine -tuned on

9:53

great management HR, employment law,

9:55

and other kinds of content.

9:57

And then in addition to

9:59

that, we're ingesting the company's

10:01

information. So their employee

10:03

handbooks, their strategy, their values,

10:05

their... organizational wisdom, learning and

10:07

development, anything that the organization

10:10

would use to onboard an

10:12

employee, we would say use

10:14

that to onboard our AI. And

10:16

over the course of a week, that

10:19

AI gets really smart about that organization.

10:22

The point is that once it's really

10:24

home to the organization, then they

10:26

invite their people to engage with our

10:28

AI, which we call Tuffy, and

10:30

that was named by users. So an

10:32

employee will get an email that

10:34

says, Hey, We'd like to

10:36

offer you the opportunity to talk

10:38

with Tuffy. Click your Go set up

10:41

an account. They sign up

10:43

though with their personal email

10:45

and their personal phone number, their

10:47

personal device and we tell

10:49

them and the organization tells them.

10:52

Please do this so that you

10:54

have complete control over all of

10:56

your own data. This is a

10:58

safe place where you can go

11:00

have a conversation about anything and

11:03

not worry about saying things the

11:05

wrong way or asking a dumb

11:07

question. It's really very much a

11:09

place where they can feel comfortable

11:11

with whatever is going on. I

11:14

think this is really important because

11:16

trust is at the center of this.

11:18

We're going to come back to

11:20

trust a little later on. But if

11:22

I have any inkling that the

11:24

company might get access to my conversation

11:26

with Tuffy, I'm just not

11:28

going to go to it. So

11:30

I think it's really smart that you

11:32

did that outside of the company,

11:34

that you could do this just kind

11:36

of on your own personal device. Yeah,

11:39

absolutely. And we're really

11:41

providing the trust to both

11:43

the employer and the

11:45

employee. First of

11:47

all, a lot of people in the

11:50

organization would never say anything. We know

11:52

that people don't necessarily tell the truth

11:54

on employee surveys, and they're afraid to

11:56

say what's really going on. And

11:58

if they did any of this the

12:00

company's devices, they would have access to that

12:02

data, whether they were looking at it

12:04

or not, which just creates an error of.

12:07

concern. Let's just call it that. And

12:10

for the organization, they want to be

12:12

able to understand what's going on, but

12:14

they don't want to have the whole

12:16

mess exposed to them either. That's a

12:18

danger to the organization. It's

12:20

a risk. So for them, having a

12:22

bit of a barrier and saying, go

12:24

here and have the conversations you need

12:26

to have is actually very helpful for

12:28

the organization. Can you share

12:31

some of your key learnings? Look,

12:33

how are people using Tuffy?

12:35

What are they going to it

12:37

for? So first, I

12:39

will say about a third

12:41

of the challenges are

12:43

really policy related, you know,

12:45

explain our how FMLA

12:47

works, or I don't understand

12:49

how how to get

12:51

my corporate card, all the

12:53

kind of operational things.

12:55

Do I get a lactation

12:57

break? Things like that. It

13:00

can get more complicated. You know,

13:02

what are the gender rules now, especially

13:04

with a new administration? What

13:06

are the bathroom rules? So there's

13:08

a lot of interesting. change

13:10

happening where each organization can

13:12

define how do they want

13:14

to address those issues. And

13:16

the one example I use

13:18

often is nepotism. So in

13:21

a family -owned business, nepotism

13:23

is expected and positive. And

13:25

that kind of organization would want

13:27

to potentially explain that. And

13:30

in other organizations that's frowned upon or

13:32

against policy. So there's a customization

13:34

there and I think a lot of

13:36

employees have questions about things and

13:38

what is ethical, what is allowed, all

13:41

of that. We can

13:43

answer those questions very well,

13:45

very quickly. The other

13:47

two thirds of the issue

13:49

are the sticky and interesting

13:51

things. So you could think

13:53

about a manager relationship and

13:55

all the different types of

13:57

managers, maybe a micromanager, maybe

13:59

there's some toxic behavior, maybe

14:01

there's favoritism. Maybe someone

14:03

just has poor communication style

14:05

and choices. And I'll

14:07

just say the alternative there

14:10

often is an employee calling

14:12

a manager and saying, hey,

14:14

can i have five minutes of your

14:16

time can we have a quick

14:19

coffee i'm confused of course that would

14:21

normally happen but think about the

14:23

hours and hours of time that it

14:25

ultimately takes for managers to have

14:27

all of those conversations we're just saying

14:29

tuffy plays the role of being

14:31

the first point of contact when you're

14:33

confused or you're struggling come to

14:35

tuffy first so that when you Go

14:37

talk to your manager or you

14:39

go talk to someone else in the

14:41

organization. You're fully prepared and you're

14:43

focused on the right things. How

14:46

does tough day deal with

14:48

issues that have legal repercussions? So

14:50

for example, a sexual assault, what

14:53

would it say?

14:56

Yeah, so we are a communication

14:58

platform and not a reporting

15:00

platform. So I think this also

15:02

provides the workforce a safe

15:04

place to go. to talk about

15:06

what's going on. And

15:08

if people do come

15:10

in and there's something egregious,

15:12

someone was assaulted, the

15:15

AI will be very, very

15:17

curious and ask for more

15:19

context, collect more

15:22

information. Ultimately, diagnose

15:24

the situation as something that

15:27

should be reported, but we are

15:29

providing the agency to the

15:31

individual employees. So we want to

15:33

build their competence and their

15:35

confidence and help them understand why

15:38

it should be reported, why the company

15:40

really wants them to report it, how

15:42

they should report it, and what to

15:44

expect in the process. So

15:46

as an example, a lot of people will

15:48

worry, you know, if I report this, will

15:50

the other person be alerted? Well, the reality

15:52

is like, yeah, they're going to be part

15:55

of an investigation. But there's a

15:57

lot of ways that you can

15:59

engage with the employee and support them

16:01

in the process so that they

16:03

feel comfortable with that. And ultimately, they'll

16:05

make that decision. So

16:07

we can see the benefit for

16:09

employees. Why would a company

16:11

be incentivized to do this? Is

16:13

it mostly like cost effectiveness and just

16:15

cutting costs or is there, yeah,

16:18

like when you sell a tough day,

16:20

what is the value proposition to an

16:22

organization? Well, there is

16:24

a lot of efficiency. So

16:26

going back to that use

16:28

case of of the manager spending

16:30

a lot of time having

16:32

a lot of conversations and you

16:34

could probably spend 150 % of

16:36

your time answering questions or

16:38

taking the coffees and and having

16:40

some coaching sessions mentoring sessions. A

16:44

lot of what is discussed

16:46

can probably be handled by the

16:48

AI. So if you can cut

16:50

back 80 % of those questions,

16:52

that's a big savings. And

16:54

that's really the value prop. The

16:57

other value prop is retention. So

16:59

if you have these challenges sort

17:01

of festering and people are not

17:03

getting their problems resolved, even if

17:05

they're small, they tend to quiet

17:08

quit or take PTO. What's

17:10

quiet quit, by the way?

17:12

Oh, quiet quitting is when

17:14

you are doing the very

17:16

minimum in your job so

17:18

that you don't get fired,

17:20

but you're not very productive.

17:23

And it also results in regrettable

17:25

attrition, and often organizations don't

17:27

know until it's too late, often

17:29

in the exit interview that

17:31

something was going on. So

17:33

this is an opportunity to

17:36

serve people better listen and

17:38

learn from what's going on we

17:40

provide anonymized data to the

17:42

organization so that they use that

17:45

like a weather vane to

17:47

know where the problems in my

17:49

organization that they can go

17:51

address and as they do they

17:53

improve the employee experience ultimately

17:55

improve retention and. The cost

17:57

of replacing someone who attrits is

17:59

two times their salary. So you can

18:01

think of losing one good employee

18:04

that might be making $100 ,000 a

18:06

year, cost $200 ,000 a year. After

18:08

a short break, we talk

18:11

about the trend towards conscious

18:13

unbossing, popular with younger professionals.

18:16

And this new concept might just change

18:18

the workplace as we know it. Stay

18:21

with us. Stripe

18:35

helps many of the world's most

18:37

influential businesses grow their revenue and

18:39

build a more profitable business. Whether

18:42

it's Hertz making check out a smooth

18:44

ride for their customers, OpenAI

18:46

answering unprecedented demand, or

18:48

PGA chipping away at

18:50

back office inefficiency, Stripe's

18:52

financial infrastructure platform helps

18:54

companies achieve ambitious goals.

18:57

No matter what success looks like for your

18:59

business, Stripe helps ensure the

19:01

complexity of financial systems doesn't get in

19:03

your way. Learn more

19:05

at stripe.com. No

19:08

matter your industry, everyone's looking

19:10

to save money. But the best

19:12

leaders in finance aren't just

19:15

cutting costs, they're working hard to

19:17

drive growth. And Brex's modern

19:19

finance platform helps them do just

19:21

that. Brex offers the world's

19:23

smartest corporate card, banking, expense management,

19:25

and travel all in one

19:27

place. Over 30 ,000 companies

19:29

from scrappy startups to large

19:31

organizations are using Brex as

19:33

a competitive advantage, and you

19:35

can join them. Get the modern

19:38

finance platform that works as hard as

19:40

you do at brex.com. So

19:44

I want to zoom out. Tough Day

19:46

is coming out at this really interesting

19:48

moment when it comes to the future

19:50

of work. So I want to kind

19:52

of dissect some of the current trends

19:54

that you're seeing around the future of

19:56

work and how I guess Tough Day

19:58

is helping address some of these concerns

20:00

or some of these opportunities that these

20:02

trends are creating. So the first is

20:04

this whole idea of conscious embossing, yes,

20:06

which I only recently heard about. And

20:08

it's basically a trend where young professionals

20:10

like my daughter, she's about to graduate

20:12

from college, essentially don't

20:14

want traditional managerial roles and they

20:16

don't also want to have managers,

20:19

it tells more about that. Exactly.

20:21

What is that? Yes, 72 % of

20:23

Gen Z says that they are

20:25

consciously unbossing. And to your point,

20:27

they do not want to have

20:29

a manager. Okay. And they do

20:32

not want to be a manager.

20:34

And the reason is if you

20:36

dissect it, what has been modeled

20:38

for them by their managers is

20:40

burnout, stress. not

20:43

very enjoyable job number one but

20:45

also no matter how hard they

20:47

try a lot of them still

20:49

get fired their employee reviews are

20:51

bad they're just like why would

20:53

I even want to do that

20:55

and the truth of the matter

20:57

is. most

21:00

people who become people managers didn't do it because

21:02

they want to become people managers they did

21:04

it because it was the next sort of rung

21:06

on the ladder it's like what you have

21:08

to do to get promoted or what you have

21:10

to do to make more money. So

21:12

82 % of people are what

21:14

they call accidental managers and they

21:17

just don't have. the capabilities, or

21:19

frankly, many of them not the

21:21

interest. So Gen Z is saying,

21:23

you know, I'm happy to manage

21:25

myself. I can get the information

21:27

that I need. I know how

21:30

to network. I am very

21:32

open to feedback, and I can

21:34

get feedback from anyone. Why does there

21:36

have to be this formal structure? And

21:39

I have to say, just as we make

21:41

our way in the world with Soft Day, in

21:44

one -on -one conversations and

21:46

behind closed doors, everyone

21:48

agrees. The role of

21:51

the managers kind of outdated. And

21:53

great managers do exist. And

21:55

the problem with the great managers

21:57

is they don't scale. And

22:00

so if we could give them a

22:02

way to augment themselves and create their

22:04

sort of digital replica so that they

22:06

have got a partner, they can spend

22:08

their quality time having those coaching conversations

22:10

and great conversations that they should be

22:12

having. with their team

22:14

and then offset the rest

22:17

of it with an AI. That

22:20

will help improve the experience for

22:22

everyone. The other side of

22:24

the coin is the bad

22:26

managers or the mediocre managers who

22:28

are actually costing the company

22:30

time, talent, and money. Yeah,

22:32

we're just going to use Tuffy instead. So

22:35

the second kind of interesting trend that is

22:37

also very related to what Tough Day is

22:39

doing is this idea of the great flattening.

22:41

Yes. Tell us what

22:43

that is. Yes. So

22:45

organizations are flattening out the

22:47

middle layers. And

22:49

Gartner actually has some research recently

22:51

that says 20 % of companies

22:53

are getting rid of 50 % of

22:55

middle management in the next two

22:57

years. And I think

23:00

part of that is because

23:02

we've spent so much money trying

23:04

to train managers to be

23:06

good managers. And we're not

23:08

showing a return on that learning.

23:10

And organizations are looking

23:12

for other answers. So

23:15

I think, you know, it might

23:17

be a little bit challenging right now,

23:19

but actually, if you think about

23:21

the real role of the manager, a

23:23

lot of those things can be

23:25

outsourced to AI. So

23:27

there's another shift happening specifically

23:29

in the HR landscape. A

23:31

LinkedIn survey found that HR had

23:33

the highest turnover rate out of

23:35

jobs that they tracked. Why

23:38

do you think that is the

23:40

case? And does that kind of

23:42

solidify the opportunity for a tough

23:44

day. I will say in

23:46

terms of HR, we're seeing

23:48

a few different things. One is,

23:51

know, everyone who goes into HR goes in,

23:53

I think they're people, people, right? Initially, you

23:55

think like, I'm going to do a good

23:58

thing. I'm going into this field. And

24:00

as you get into it, it

24:02

may or may not be your

24:04

cup of tea. It is largely

24:06

about protecting the organization. And

24:09

that's great. We need people to

24:11

design employee experiences and deliver services

24:13

and measure all of that. It

24:15

may or may not have been

24:17

why you got into HR in

24:19

the first place. So I think

24:21

there's some of that. I

24:23

think there's so much

24:26

stress and pressure on HR.

24:28

The typical first point of contact for

24:30

an employee is their manager. And

24:33

if that piece is broken, one

24:35

of the next calls is

24:37

hr so they're getting bombarded with

24:39

all of these challenges that are really

24:42

hard and emotionally draining so i

24:44

think there's some burn out there and

24:46

and then in terms of all

24:48

the solutions how do you solve for

24:50

all of this it's just it's

24:52

it's hard work. HR business partners are

24:54

the heroes of an organization. They're

24:57

doing lots of great work, but I

24:59

think depending on the state of

25:01

the organization and what resources they have

25:03

available, what their benefits are, every

25:05

organization has to figure out what's the

25:07

right recipe for their culture and organization,

25:10

and that might be part of it.

25:12

Now, we're also kind of in

25:14

this moment of time where

25:17

We're seeing top -down pushback on

25:19

diversity, equity, and inclusion efforts. We're

25:21

seeing this at the federal

25:23

level, of course, where President Trump

25:26

is seeking to end government

25:28

support for programs promoting DEI. We've

25:30

already seen a lot of

25:32

references to DEI taken offline on

25:34

federal websites. And this is

25:36

all being contested in the courts,

25:38

of course. It'll see what

25:41

happens. We have seen

25:43

a ripple effect already at companies like

25:45

Meta and Google where they've rolled back

25:47

some of these efforts. I

25:49

would imagine that a lot of the

25:51

concerns that show up in the workplace

25:53

are related to equity at work and

25:55

inclusion. What do you think

25:57

of all of that? And again, how can

25:59

Tough Day help? On the

26:01

topic of DEI, when an

26:04

organization works with Tough

26:06

Day, We're tracking what is happening

26:08

in the federal government, and we're tracking

26:10

those trends. And this is an area

26:12

where every organization has to do a

26:14

bit of their own way of finding.

26:17

At the federal level, there

26:19

are things going on, but

26:21

then also at the state

26:23

level. So state by state,

26:25

there's new AI law. There's

26:27

also new DEI -related regulations. Right

26:31

now or at a point where

26:33

organizations have to navigate a lot of

26:36

that on their own and you

26:38

see companies like Costco and Apple and

26:40

Microsoft doubling down on DEI and

26:42

saying, you know, this is

26:44

super valuable for our organization is

26:46

just part of our values and

26:48

we have to, we have to

26:50

invite diversity into the organization to be

26:52

a strong resilient organization and to

26:54

serve our customers best. And

26:56

then on the other end of the spectrum,

26:58

you have. organizations saying,

27:01

you know what, I just don't want

27:03

to fight this machine. Like some

27:05

are saying we don't have to call

27:07

it DEI. Like we love the

27:09

diversity and what have you, but maybe

27:11

we've moved beyond this language. So

27:13

I think there's some of that where

27:15

they're kind of in the middle.

27:17

And then you have organizations just saying

27:19

like, it's all out the window.

27:21

So when we work with companies, we

27:23

ask them, how do you want

27:25

to handle things like bathroom issues? And

27:28

there were customizing the experience

27:30

and the answer for each

27:32

organization. Even

27:34

though a lot of the

27:36

inspiration for Tufte came

27:38

from looking at underrepresented groups

27:40

and their challenges, we

27:42

continue to see that none

27:44

of these issues are

27:46

just about one population. And

27:48

it just doesn't matter who the person is.

27:50

I think it matters. How

27:53

you handle the situation and ultimately

27:55

we want to help people

27:57

have better communications and interpersonal relationships

27:59

and working relationships and that's

28:01

the point. But to

28:03

do this tough day needs to

28:05

train toughy with each company's values

28:07

and cultural norms. So

28:09

how do they do that and how

28:11

are they any different than say

28:13

Microsoft's copilot. We get to that after

28:15

a short break. Convert

28:23

more shoppers into buyers with

28:25

Stripe, the financial infrastructure platform

28:27

to help grow your revenue.

28:30

Businesses saw an 11 .9 %

28:32

revenue uplift on average when

28:34

they used the Stripe Optimized

28:36

Checkout Suite. Optimizations like

28:39

intelligently surfacing the right payment

28:41

methods to each customer, thanks

28:43

to Stripe's machine learning, which is

28:45

trained on trillions of transactions

28:47

globally. This means more

28:49

revenue for you and a better experience

28:51

at checkout for your customers. See

28:54

what Stripe can do for your

28:56

business at stripe.com. If

28:59

you're building a small business, you

29:01

know the questions never stop. Join

29:03

us for the latest installment of

29:05

Masters of Scale Strategy Session, the

29:07

small business playbook. A free

29:10

live virtual event presented in alliance

29:12

with Capital One Business. On

29:14

May 15th, our session is co -hosted

29:16

by leaders who know how to

29:18

scale with impact. Amy Errett,

29:20

a four -time entrepreneur and CEO

29:23

of Madison Read, and Sarah

29:25

Rob O 'Hagan, business innovator and

29:27

best -selling author of Extreme U. They'll

29:29

be answering real questions from business

29:31

owners like you, Let's

29:46

go behind the scenes and kind

29:48

of geek out for a bit

29:50

about how you've built Tuffy. So

29:52

the first question I have is, why

29:55

can't just a company, you know,

29:57

takes, I don't know, say a Microsoft

29:59

co -pilot and feed it its own

30:02

policies and employment law and just

30:04

basically replicate what you guys have done.

30:06

So just talk to us a

30:08

little bit about the technology behind the

30:10

scenes and what is your competitive

30:12

mode and differentiator against say a Microsoft

30:15

co -pilot. Yeah, so number

30:17

one, I do think coming back

30:19

to the issue of trust, it's

30:21

really hard to have any of

30:23

these conversations internally and know that

30:25

your organization can see the data.

30:28

You just can't have that

30:30

level of trust with

30:33

an employee on the stickiest,

30:35

trickiest kinds of challenges. So

30:38

I don't think Microsoft Copilot

30:40

is going to be able

30:42

to have that level of

30:44

relationship and trust. So

30:46

that's one. Number two, we

30:49

have partnerships with a lot

30:51

of content partners that have gated

30:53

content. They will not allow

30:55

the big LLMs, including open AI

30:57

or Plexity or with co -pilot.

31:00

They just don't have the

31:02

content. What are examples of that?

31:04

So one of my favorite examples

31:07

is Charter. CharterWorks.com

31:09

is kind of an up

31:11

and coming thought leader. They're

31:13

a bunch of journalists from

31:15

Wired Magazine and Wall Street

31:17

Journal, New York Times, The

31:19

Atlantic. These are serious professionals

31:21

that all they do is research and

31:23

report on work. And they

31:25

have amazing content. And we have

31:27

a partnership with them. So we

31:30

have all of their content that

31:32

is subscription based. So that's

31:34

an example of how we

31:36

make our platform smarter. Yeah.

31:38

Does Tuffy have a

31:41

personality? Is it like culturally

31:43

sensitive or culturally specific? Yes.

31:47

Yes. So first,

31:50

Tuffy has values and Tuffy

31:52

is very curious. So

31:54

our first phase of development, we did

31:56

what's called Wizard of Oz testing. Can

31:59

you explain what that is for people

32:01

who are not familiar with it? That's

32:03

cool. So we had sitting around a

32:05

table at any given time during this

32:07

process. We would have five or six

32:10

experts, always a lawyer, an

32:12

HR person, a manager, and

32:14

a therapist. And we asked

32:16

early testers to interact with the

32:18

platform so that they would come

32:20

in and ask questions. We told

32:22

them, it's not really an AI.

32:24

You'll interact with it like AI,

32:26

but we actually have these human

32:28

experts behind the scenes. That's the

32:30

Wizard of Oz piece, right? Yeah,

32:32

the Wizard of Oz. And the

32:34

really cool thing is most people

32:36

forgot they were actually talking to

32:38

humans. What happened for all of

32:41

us sitting around the table is

32:43

that we would decide very quickly

32:45

who was the best person to

32:47

answer this question. and or

32:49

tackle it. And what that meant was usually

32:51

that person was asking five or six

32:53

different questions to really understand. So just like

32:55

if you go to a lawyer and

32:57

you ask a question, they're not giving you

32:59

advice at first. They're going to ask

33:01

you more questions. That is

33:03

really how we develop

33:06

Tuffy to be curious

33:08

and get more context.

33:11

And then diagnose what the situation

33:13

is because so many people

33:15

will ask one thing and they're

33:18

asking the wrong question or

33:20

they don't really know what the

33:22

issue is. So, so I

33:24

will say Tuffy is generically very

33:26

curious and. wants to understand

33:29

a situation and bring empathy to

33:31

the dialogue. But

33:33

in terms of your question

33:35

about cultural customization, when

33:37

working with the Hawaii Employers

33:39

Council, which is

33:41

a network of 700

33:43

companies all based in

33:45

Hawaii and that represents

33:47

170 ,000 workers and HEC

33:49

is using Tough Day.

33:52

to join its HR consulting team

33:54

and provide advice to those employees,

33:56

but also as a go -to -market

33:58

partner. It's interesting. We started down

34:01

the path and we got some feedback

34:03

that maybe Tuffy sounded a little too

34:05

New York. And they

34:07

asked, could we make

34:09

Tuffy seem more Hawaiian?

34:12

And that was that language. That was

34:14

really culture and value. So they asked

34:16

us if we could apply, quote, a

34:18

lo -ha spirit to our AI. How

34:21

did you do that? So

34:23

first, they did give

34:25

us redacted transcripts

34:27

of conversations between really

34:29

talented and successful

34:31

HR leaders, managers, what

34:33

have you. So

34:35

we could understand the

34:37

cultural tone of

34:39

the dialogue and After

34:41

we did customize the experience, users

34:43

noted it right away. We got

34:46

great feedback in the platform. And

34:48

our helpfulness rating went from

34:50

88 % to 99 .6%. And actually

34:53

back to kind of this,

34:55

because I'm always kind of thinking,

34:57

OK, what is a company's

34:59

competitive mode? That kind of

35:01

training and that kind of data is

35:03

extremely unique. Like you can't scrape the internet

35:05

for this kind of data. And that's

35:08

very powerful. So one of

35:10

the things that we're very passionate

35:12

about on this podcast is how

35:14

do we build responsible and trustworthy

35:16

AI? And

35:18

you're in a very tricky

35:20

space, right? Like people's

35:23

careers are at stake, people's

35:25

livelihood are at stake. And

35:27

a recent Pew study found that

35:29

52 % of employees say they're worried

35:32

about the future impact of AI. use

35:34

in the workplace. So how

35:36

are you building trust

35:38

into Tuffy? Yeah. Well,

35:41

first, I believe, we

35:43

all at Tough Day believe

35:45

that AI is a more human

35:47

positive solution, often than a

35:50

human. And I will say,

35:52

I believe in humans, don't get

35:54

me wrong, I'm a human

35:56

-centered person. But I will say,

35:58

Humans are messy. Humans have

36:00

bad days. Humans can make

36:02

mistakes and do make mistakes.

36:04

But with the AI, we

36:07

want to make sure that the

36:09

AI is not biased, that the

36:11

AI has safety guardrails, that the

36:13

AI has a hallucination defense system

36:15

in certain circumstances, especially when you're

36:17

calling on a policy or a

36:20

law or what have you. You

36:22

want to make sure this is

36:24

not AI making it up. It

36:26

is specific. That's part of

36:28

it. The other part is

36:30

all that red teaming and knowing

36:32

where all the pitfalls can be.

36:35

So there's a lot of scenario

36:37

planning to think through what could

36:39

possibly happen. You want to

36:41

think through what are all the things

36:43

that could go wrong and build the

36:45

right knowledge and guardrails into the product?

36:48

Yeah. So final question. I

36:50

am very passionate about this idea of

36:52

human -centric AI. So AI that It's

36:54

going to help unlock human potential.

36:56

It's going to augment and amplify our

36:58

abilities, not replace us. But I

37:00

do spend a fair amount of time

37:02

thinking about this following question. In

37:05

this age of AI, where AI can

37:07

be patient and curious and empathetic like

37:09

Tuffy, what does it mean to

37:11

be human? I

37:13

do think being human, if

37:16

you think about all the human needs,

37:18

like even Maslow's hierarchy of needs, the

37:22

positive part about having AI

37:25

involved is a lot of

37:27

our physiological needs and even

37:29

I think community needs in

37:31

terms of interpersonal skills and

37:33

understanding one another and being

37:35

curious and all of that,

37:37

we can accelerate or kind

37:40

of meet those needs faster.

37:42

That frees us up for

37:44

those higher level needs of

37:46

learning and innovation and problem

37:48

solving. And so I think

37:52

The communication and community is

37:54

a human thing, but I

37:56

think it's going to

37:58

be accelerated and improved by

38:00

technology, including AI. I

38:02

do think we might actually

38:04

find ourselves being more empathetic because

38:06

we have relationships with AI

38:08

that are empathetic. I also think

38:11

there's another interesting thing in

38:13

talking with neuroscientists. When humans

38:15

are under stress, our brains

38:17

actually shrink. So if we

38:19

think about what's happening to

38:21

people in the world of work

38:23

or even in our communities,

38:25

the more stressed out we are,

38:27

the less capable we are

38:29

to use the most creative parts

38:31

of our brain. And I

38:33

think what we'll see is that

38:35

this technology by solving some

38:38

of the things that cause us

38:40

stress is going to make

38:42

our brains actually grow and be

38:44

more intuitive and more empathetic

38:46

and more creative and innovative to

38:48

solve the biggest problems in

38:50

the world. Like, I'm absolutely a

38:52

human optimist. That's

38:55

the perfect way to end our

38:57

conversation. AI that is going to help

38:59

expand our brains like literally. Thank

39:02

you for joining us, Catherine. This

39:04

was awesome. Thank you. Thank you so

39:06

much, Rana. Such a pleasure. One

39:10

of my key learnings in starting

39:12

and scaling my company, Affectiva, was

39:14

that the biggest challenge was actually

39:16

not creating Emotion AI, but

39:18

it was dealing with people and all

39:20

of the human messiness that comes

39:23

with it. What I love about Tough

39:25

Day is that their AI platform

39:27

handles that human messiness in a safe

39:29

and supportive way. They

39:31

can manage these issues before they really

39:33

escalate, which allows professionals to focus

39:35

on the work. Often

39:37

the conversation about AI and the

39:39

future of work is focused on job

39:41

loss, but the conversation shouldn't stop

39:43

there. It's so important for

39:46

us to ask, how can AI augment

39:48

our abilities at work What are

39:50

the areas of work that slow us

39:52

down, and where would we rather

39:54

be focusing our energy? One

39:56

of these areas is management, but there

39:58

are so many other opportunities. From

40:01

scheduling and automating bureaucratic tasks, to

40:03

maybe even putting in lunch orders

40:05

before you get hangry. AI

40:07

is still evolving, which means that

40:09

we have the opportunity to shape it. How

40:12

do you want to see AI work for

40:14

you on the job? Leave

40:16

us a voicemail at

40:19

601 -633 -411. That's

40:22

601 -633

40:25

-2424. Next

40:27

week on Pioneers of AI, we

40:29

take you live to the Abundant

40:31

Summit Stage in LA, where

40:33

I spoke with some leading investors

40:35

on the crazy -maze evaluations of

40:37

AI companies, what that means

40:40

for venture funding, and where investors

40:42

should place their bets in AI.

40:48

Pioneers of AI is a wait

40:50

-what original. Our

40:53

executive producer is Yves Tro.

40:55

Our producer is Rachel Ishikawa. And

40:57

our associate producer is Jordan

40:59

Smart. Our senior

41:02

talent executive is Stephanie Stern. Mixing

41:04

and mastering by Ryan Pugh.

41:07

Original music by Ryan Holiday. And

41:10

our head of podcasts is Litao

41:12

Moulin. You can join

41:14

the conversation on LinkedIn, Instagram,

41:16

TikTok, YouTube, and X. Just

41:19

search for at pioneers of

41:21

AI. Thanks so much for listening.