0:07

Welcome, everyone, to the AI

0:09

in Business podcast. I'm Matthew

0:12

D'Amelio, Senior Editor here at

0:14

Emerge Technology Research. Anuj

0:17

Manjar is a Principal

0:19

at Deloitte specializing in

0:21

Financial Services. He

0:24

joins Emerge CEO and Head

0:26

of Research Daniel Fajello on

0:28

today's podcast to discuss the

0:31

dynamics of balancing customer expectations

0:33

and channel strategy for financial

0:35

institutions. Later, they discuss

0:38

emerging AI use cases

0:40

throughout the multichannel space,

0:42

including after-call work in

0:44

customer service. Today's

0:46

episode is sponsored by Deloitte, and

0:49

without further ado, here's their conversation.

0:59

So, Anuj, welcome to the program. Thanks

1:02

for having me, Dan. Yeah, glad to

1:04

be able to dive in to the

1:06

customer service side of financial services. Lots

1:08

of talk about an unpack, but

1:11

I always like to begin with

1:13

trends and challenges. You're seeing generative

1:15

AI enter insurance and investment banking

1:17

and every other portion of financial

1:19

services, and you're seeing new

1:21

dynamics arise and maybe concerns arise. What for

1:23

you are kind of the driving trends that

1:26

are kind of defining the

1:28

adoption of this technology? The

1:30

biggest trends we're seeing is that AI

1:34

and Gen AI

1:36

are really making an impact

1:38

on solving the equation that most

1:40

service executives have dealt with

1:43

since I started operating this space

1:45

25 years ago, which is how do I

1:47

deal with increasing customer

1:50

expectations with increasing

1:53

pressure on my budgets? I

1:55

need to be able to do

1:57

more with less, and the trends we're

1:59

seeing... are the adoption

2:02

of AI to

2:05

really help solve that equation. Whether

2:08

it comes in the form of conversational AI,

2:10

generative AI, or

2:13

simple just brute force

2:15

analytics that tell me more about

2:17

what customer needs are and how

2:20

I can solve them efficiently. Yeah

2:22

and there's multiple dynamics to kind of pick a

2:25

part here. On the side

2:27

of increasing customer expectations you and I were

2:29

talking off microphone about how despite what's going

2:31

on in the financial markets or to companies,

2:34

people's expectations around how

2:36

they're being treated whether it's via phone, via

2:38

chat on an FAQ site or something like

2:41

that are going to be level set by

2:43

all the other great technology companies that they're

2:46

interacting with. Whether it's the Amazon's

2:48

of the world or whatever else and so

2:50

those pressures are increasing. How

2:52

would you sum up I guess the

2:55

the origin of this pressure to increase

2:59

where these increased customer expectations are coming from. Do you

3:01

see it more on text than invoice? Do you see

3:03

it? Is there any detail to that that maybe our

3:05

listeners should be aware of? We definitely

3:07

see it across channels

3:11

and what I'll say there is the biggest

3:15

pressure point is on how

3:19

integrated channels should

3:21

be to solve a

3:23

customer's issue. So you

3:25

know we've all heard the term omni-channel over

3:27

and over and again and and we know

3:29

it means a lot of different things to

3:32

different people. I personally

3:34

believe in what's called

3:36

right channeling and and that

3:38

is the pressure that we're seeing. It's what's

3:41

the right channel or what's the

3:43

right sequence of channels to serve

3:45

as a customer's need. It's

3:47

not just make all

3:49

channels available to serve all types

3:52

of needs. I mean we've had

3:54

clients that have done things like

3:56

made a social media channel available in

3:59

financial services when

4:01

from a servicing perspective, you can't

4:03

really address a customer's issue in

4:05

a social media channel. You immediately

4:07

have to switch them. So

4:10

the question then is, what is

4:12

the real channel strategy? If it's going to

4:14

be an issue

4:16

that requires a human being or

4:19

it's going to be an issue that requires

4:21

a level of authentication that

4:23

needs to push out of a social

4:25

media channel, then orchestrate

4:28

that for the customer. That's

4:30

the pressure we're seeing. It's

4:33

don't do what others

4:36

have done in terms of just

4:38

proliferating channels. Create

4:40

connections between the channels to serve our

4:42

needs. Do you see

4:45

enough bigger organizations doing

4:47

enough of that where if I'm a

4:50

financial services company and I'm not, I'm

4:52

maybe railroading somebody into a channel that

4:54

isn't really right for their question, where

4:56

they're feeling that overtly now is a

4:58

frustration? Is this both a now pressure

5:00

and something they're thinking is going to

5:02

arrive in the future? It

5:04

is a now pressure. It

5:06

usually is a result of

5:09

disconnected channel strategies and

5:11

almost an organizational

5:13

structure issue in many of

5:15

these large financial services companies

5:18

where the digital channel

5:20

is owned by one organization, the contact

5:22

center channel is owned by another organization.

5:24

In a bank, the branch channel is

5:27

owned by the head

5:29

of the retail bank and bringing

5:32

those groups together can

5:35

be challenging in some of these larger institutions

5:37

and even in smaller institutions. Some

5:40

of it is a result of the

5:43

organizational structure within these

5:45

financial services companies. Other

5:48

challenges are really a result of a patchwork

5:51

of technology and

5:53

an underlying data infrastructure

5:56

that can't Create

5:58

that. Right channeling

6:01

strategy that alps. Service.

6:03

A customer. You know,

6:05

sequenced way carrying information from first

6:07

channel to second channel to thirteen.

6:10

Yeah. I'm in. We're going to get a

6:12

little bit into that in the use cases,

6:14

but I do want to touch on the

6:16

other side of the pressure. There's these increase

6:18

expectations is also this crunch for costs arm

6:20

so it's it's It's a little bit of

6:22

a rock and a hard place for folks

6:25

who are leading customer experience, customer support. Your

6:27

which kind of the fast version for the folks

6:29

who din. In. Terms of why.

6:32

Costs. Are being pressured so much for that?

6:34

The customer. Support. Part of of

6:36

Censor: What are some of those big dynamics that

6:38

are driving that that are on people's minds? Serve.

6:41

Or I'll I'll start with the

6:43

insurance sector on the Pnc insurance

6:45

side and we've done a personalized

6:47

side of the carriers are had

6:49

just been really hit hard with

6:51

inflationary pressures because to reflect what

6:53

it costs to replace the bumper

6:55

as got up and up and

6:57

up. And. Those cost pressures you

7:00

know that resulted in you know,

7:02

over one hundred combined ratios that

7:04

the carriers have to deal with

7:07

through looking at their operating expenses.

7:09

On a life insurance side.

7:11

Similarly, years and years of

7:14

pressure from low interest rate

7:16

environment where their conservative investment

7:18

portfolios just haven't. Made.

7:20

Big returns Now there's been a fan

7:22

of of you know of reprise their

7:25

as as rates have increased by it

7:27

on the insurance side. Overall the the

7:29

the the sector has been under tremendous

7:31

pressure. Banking. Banking

7:34

right now is dealing with.

7:36

You know, a series of Christ Disease

7:39

that has quite a ton of

7:41

pressure on the banks. Not.

7:43

Only their balance sheets, but their

7:45

incomes statements. They are. They

7:47

are competing now for deposits in a

7:50

way that they have an addict compete.

7:52

In. A very long time

7:55

raising raising their costs

7:57

on and and reducing.

7:59

Their. net interest

8:01

income. And so there

8:04

is a push to look for where are

8:07

there places in our operations where

8:09

we can reduce costs. And

8:11

then finally on the asset management side, you

8:15

know, again, we were talking off camera, Dan,

8:17

you know, there's mutual fund

8:19

providers, ETF providers now where

8:22

there is no fee on

8:25

those investments, right? There's so

8:27

much downward pressure on

8:29

those fees that you have to

8:31

create a customer

8:33

service organization that's really efficient

8:36

to be able to drive profitability

8:39

off of, yes, increasing

8:41

AUM, but

8:43

decreasing fees on the

8:45

top of AUM. Yeah.

8:48

And I hate to admit, I'm definitely one of

8:51

those vanguard types. So the fact that they still

8:53

have to service me, I can actually sympathize now

8:55

that you're articulating it out loud here at New.

8:57

So a few phone calls into the call center

8:59

and they're losing money on you. That's exactly it.

9:02

I, to be frank, I can't remember if I

9:04

ever have called them, but now I know that

9:06

despite how long I've been with them, man, it's

9:09

going to be, it's a tough road for

9:11

those folks. So God bless them. With that said,

9:13

we've got, so we've got trends that are substantial.

9:16

We also have opportunities that are unlike

9:18

anything we've seen before that the previous

9:20

kind of chat bot AI wave just

9:22

was quite a gag and really didn't

9:24

turn into very much. But now we're

9:26

seeing AI with very robust capabilities in

9:28

text and voice, and it's not quite

9:31

implemented yet, but the possibilities

9:33

are clear. When you think

9:35

about use cases, I know you've got

9:37

a bit of kind of a three

9:39

factor model for weighing a use case

9:41

that you kind of advise financial services

9:44

leaders to think about the idea of

9:46

desirability, feasibility, viability. Maybe we can talk about that

9:48

first and then I'd love to poke into a couple specifics

9:50

after that. Yeah, absolutely. And

9:53

what I'd say is the service

9:55

leaders I've been meeting with in

9:58

financial services, look, they know they're doing

10:00

it. businesses. They know what's driving

10:02

cost. They have no shortage

10:04

of use cases where

10:08

new cloud AI technologies can

10:10

be applied. But one

10:13

of the challenges is really

10:15

understanding what's desirable

10:18

by the customer, what's going to actually

10:20

get adopted. You know, like you talked

10:22

about, you know, there's been no

10:25

shortage of self-service capabilities that have been

10:28

put into the marketplace that

10:30

customers just don't use. So

10:33

is it going to be desirable? Is

10:35

it going to be feasible to deliver?

10:38

Right? And again, going

10:40

back to the chatbot craze, what

10:43

often came to fruition when a

10:48

pilot was launched was, okay,

10:50

we're going to do a chatbot. It's

10:53

going to help answer these seven questions.

10:56

And then you would get into it and

10:58

it was like, wait, but to expose the

11:00

back end platform to actually be

11:02

able to take a payment from

11:04

a customer, we can't expose that.

11:07

It's too difficult. It's some legacy mainframe, like

11:09

we'd have to build an API layer on

11:11

top of it. So like, that

11:14

what's really feasible? And then, oh, how much would

11:16

it cost us to do that? Oh, and we

11:18

deflect 40% of

11:20

the calls at, you know, and the business

11:23

cases would fall apart because of the feasibility

11:25

side. And then, and

11:27

then from a viability, it really does get

11:29

into the business case side. And this is

11:31

where, you know, Gen

11:33

AI, it has a ton of

11:35

promise. But what some of our clients

11:38

aren't factoring in is

11:40

there is a cost. There is

11:43

a real cost of running information

11:45

through a large language model. There's

11:47

a real cost of creating the

11:49

right environment where your data is

11:51

protected, which of course is infinitely

11:54

important to our financial services

11:57

clients. So using

11:59

that thing, Work is really

12:01

critical in evaluating the use cases

12:04

and really looking at all three of

12:06

those dimensions. I have a client who

12:09

talked to me about having a list

12:11

of 200 use cases that they

12:13

think are ready to go for Gen AI. And

12:17

we went through this process with them

12:19

and really whittled it down to

12:21

about 12 that

12:24

made any type of sense or

12:27

that were doable in the next year

12:30

to year and a half given some of

12:32

the feasibility constraints. Yeah, I

12:34

mean, it's a challenge because I

12:36

think, and maybe you're seeing something a little bit

12:38

different in news, but leadership,

12:41

the level of AI fluency is much,

12:43

much, much higher than it was five

12:45

years ago, which is wonderful, but it's

12:47

still not spectacular. And so a lot

12:49

of the use cases we're seeing are

12:51

still inspired by the same kind of

12:54

FOMO motives that maybe drew

12:57

the chat bot thing to come to pass.

12:59

So, hey, you know, well, the boss is

13:01

asking what we're doing about Gen AI when

13:03

it comes to this random use case because

13:05

some other company tweeted it even though they're

13:07

not doing anything about it. And so we

13:09

wanted to have a bunch of use cases

13:11

around that. And I think sometimes you do

13:13

need a measured voice that can come in

13:15

and say, hey, here's what the technology is

13:17

capable of. This is what's realistic given our

13:19

infrastructure. This is what the cost of this

13:21

would actually be. And

13:23

to have a little bit of structure to saying yes,

13:25

no, when

13:27

you think about who needs to be in the room

13:29

to do that, one of the challenging things we see

13:31

in news is we don't always have all the perspectives

13:33

we need. Well, we need somebody that at least gets

13:36

a little bit of, maybe our IT infrastructure, the state

13:38

of our data. We might need somebody that understands

13:40

AI at least a conceptual level,

13:42

maybe even a technical level to

13:45

really give a thumbs up, thumbs down on

13:47

things like cost and viability and whatnot. Who's

13:50

got to have the talk about desirability, feasibility,

13:52

viability, because you and I both know sometimes

13:55

it's just a C-suite and a couple advisors and

13:57

they've been reading Twitter and getting excited. they

14:00

could have this conversation, it wouldn't be productive. Who's got

14:02

to be there? Who's got to be there? We

14:04

firmly believe it needs to be a

14:06

cross-functional team that's looking at these things.

14:09

It's interesting, right? So the phenomenon right

14:11

now is people

14:13

are being tasked with come up with your

14:15

Gen AI use cases, CEO

14:17

mandate down. The

14:20

first to run at that is

14:24

usually somebody from the analytics

14:27

team, the IT team, and

14:29

they're great thinkers at how you could

14:32

apply a technology. And in

14:34

many cases, the operations

14:36

person isn't brought in to way

14:38

too late in the process. And

14:41

really thinking about the implications

14:43

of what does it mean

14:45

to apply that type of

14:47

technology into a large-scale operation

14:49

or a complicated operation, that's

14:52

one of the things that has to be factored in. So

14:56

long way of me saying, it really does

14:58

need to be a cross-functional team that comes in

15:00

and looks at these things. And

15:02

not to say ideas can't be

15:04

generated. Ideas should be generated from

15:07

every part of the organization. Because

15:11

different parts of the organization have

15:13

a different perspective and

15:15

know where the application can

15:17

be done. We've seen some

15:19

great ideas come out

15:22

of tax or legal parts of

15:24

the organization because they understand where

15:26

some of the kind of rote

15:30

human-based work that could be

15:33

automated could be reduced. The

15:36

issue, though, is at

15:38

some point, those teams have to coalesce.

15:40

They have to go through some

15:43

type of framework like the desirability,

15:45

feasibility, viability, and come to an

15:47

alignment on what to push forward.

15:50

Yeah. Sooner rather than later to

15:52

have the right eyes in the room as

15:54

opposed to get excited, brainstorm for months,

15:56

plan to push something forward. And then,

15:59

like you said, that eventually have the operations person come

16:01

in and say, guys, how would we ever

16:03

do this? You know? I'll give you

16:06

a good example of one that we're seeing and

16:10

we're executing with some clients in market.

16:12

So in the service space, after a

16:14

customer talks to

16:17

a representative, a bank or

16:19

a customer service representative, a

16:22

claims representative on the insurance side, there

16:25

is always what's called after

16:27

call work, right? The representative

16:29

hangs up the phone and

16:31

has to take some notes,

16:36

collect some next steps, maybe execute

16:39

something in a workflow

16:41

that passes work on to

16:43

the next person who has to

16:45

help complete the task. One

16:48

of the use cases we are seeing where

16:50

Gen AI has a ton of opportunity

16:53

and impact is around

16:56

that summarization. And

16:59

accurately collecting

17:02

and then indicating what was said

17:04

on the call, what decisions

17:06

were made, and doing it

17:08

very quickly, right, in

17:11

what would usually be a three, four

17:13

minute exercise by a customer service rep.

17:16

It's a 30 second summarization. We

17:19

always focus on having a human in

17:21

the loop. You'll hear Deloitte uses the

17:23

phrase often, human plus machines is where

17:25

we need to be going from an

17:27

AI perspective. And so use

17:30

Gen AI to do that summarization

17:32

off of a

17:34

transcript that's being real time

17:36

created as the conversation

17:38

is happening, have the human in the loop

17:40

to confirm it. That is

17:43

one of those that is great

17:46

for employee experience because the employees don't want

17:48

to do that after call work and

17:51

great for an efficiency standpoint because

17:53

you've probably eliminated three minutes off the back

17:56

end of every call. Yeah,

17:58

OK. Also, it

18:00

seems like and maybe you've got the

18:02

sense of this is wealth. I do

18:04

recall one of the elements. Of.

18:06

The surf chat bot. You.

18:09

Know flimsy that we had years ago that that

18:11

that really made it flap was. The

18:13

idea that this was facing customers, we're going

18:15

to have a i come up with an

18:17

answer in a customer was going to read

18:19

it. We've seen sometimes these new waves be

18:21

more safely experimented with in places where if

18:24

there's an error, know customer. Has

18:26

a has a rough time and this

18:28

use case your articulating of after call

18:30

work if an entire paragraph was missing

18:32

or some goofy word got inserted while

18:34

we're still experimenting with things. The.

18:37

Downsides Feel. All.

18:39

Things considered potentially less consequential or

18:41

or at least overtly embarrassing than

18:43

if we're customer facing. Do you

18:45

think a lot of early experimentation

18:47

is gonna happen with use cases

18:49

of that? Kind of. Absolutely. On

18:51

the j I said, I think

18:53

there will be, especially in financial

18:55

services. There will be a real

18:57

reluctance to expose. Jedi.

19:00

I directly to customers and

19:02

in for it to be

19:05

used as that enabler. That

19:08

augmentations you with human being,

19:10

and I do think we'll

19:12

see a lot of that,

19:14

whether it's. You know,

19:16

The. Up the call work whether it's

19:19

a. A whisper type

19:21

agent that's. Listening. To

19:23

the call. Prompting.

19:26

A customer service representative with ways

19:28

that they could help. A. Customer.

19:32

You know those are the type

19:34

of applications that from a Jenny

19:36

I standpoint, we see. Companies

19:38

willing to experiment with and put into into

19:40

production. Got it? It's and so this. This

19:42

is helpful and I think for the listeners

19:45

tuned in you know. If. They went

19:47

through that last big sees as customer facing

19:49

a i answer of the drawback that happened

19:51

immediately or they might he might behoove them

19:53

to sort of. To. the into

19:56

this idea of experimenting a little bit more and that

19:58

stuff that that doesn't always directly in or with the

20:00

customers first, what are some of the other use cases maybe

20:02

of that ilk that for you are pretty

20:04

exciting and are areas of opportunity that customer

20:07

service leaders should at least be

20:09

taking into consideration? Yeah,

20:11

one of the things that we've always

20:13

talked about from a customer service perspective

20:15

is how do we

20:18

get more proactive with

20:21

our customer service? Instead

20:24

of waiting for a customer to

20:27

go online and

20:30

have to self-service themselves or

20:32

call us, we're

20:34

informing them of what may come

20:36

next. This will be a little

20:38

bit of a blend of what

20:40

I'd call just good

20:43

analytics and customer engagement with a

20:45

little bit of Gen AI. For

20:51

most of our customers, we

20:53

know when they reach out. They

20:55

reach out, let's say, if I've got

20:57

a credit card bill due three to seven days before.

21:00

If I've got my auto

21:03

insurance is due every six months and

21:07

the person is going to start looking, call

21:09

it three to four weeks before that pretty

21:12

hefty bill is due if you're on a

21:15

semi-annual policy. We

21:17

need to do more on proactively reaching

21:19

out to them, informing them

21:21

of things like what are your options?

21:25

Can we spread the payment? Can we

21:27

change the payment date to match when

21:30

your next paycheck is coming in?

21:34

Doing those reach outs, just using a good

21:37

understanding of who's calling and when

21:39

and what

21:41

they're calling about and then

21:44

pairing that with

21:46

the right messages that go out in

21:48

an automated way, whether it's email, text

21:51

if you're allowed and you've gotten permission,

21:54

more effective of course. Those

21:57

type of use cases, which again, Savannah.

22:00

Jenny I use Get this right? Yeah,

22:02

fifty five wouldn't have to be. It's

22:04

a type of things that we should

22:06

be doing and service much more regularly.

22:09

Now. Will jenny

22:11

I have a role in that

22:14

and guess I do think that

22:16

as as our clients or customers

22:18

get more. Comfortable. With

22:21

Jenny I created content. Personalizing

22:23

that I'll read: smutty am

22:25

using Jenny I is a

22:28

perfect useless. And

22:30

and and personalizing at a cost

22:32

that may be. The. Tenth of

22:34

what it would cost to do that

22:36

using traditional methods. That's.

22:39

That's where. The. New

22:41

you mentioned earlier where people going

22:43

like. This. Shift from customer

22:46

service being this reactive. Organization.

22:49

To the to being a proactive

22:51

organization that has some of the

22:54

promise. That. Jenny I

22:56

brings. In I I

22:58

think one of the things that we've been hearing

23:00

about and many conversations we just had to see

23:02

T O of Data Robot are not that long

23:05

ago. We had a great call with Goldman not

23:07

that long over there and leadership Who seeing this

23:09

or of. Idea. Of stitching together Jenny

23:11

I and existing systems and somewhere you mentioned you

23:13

know a great use case it does. Involved in

23:15

a I and I think any lisser who's to

23:18

didn't use the right tool for the job. for

23:20

crying out loud him he needed right. Don't do

23:22

you something so many for it's own sake. mean

23:24

if there's any wardrobe that we beat hear it

23:26

emerges it's exactly that's But I would say to

23:28

your point, maybe there's a way to sort of

23:30

filter for and predicted, maybe notify customer before kind

23:33

of an issue comes about. That could be number

23:35

one and that may not involve any generative ai.

23:37

Maybe that's just. Some. statistics m

23:39

l whatever get a use their but maybe there

23:41

is a next step were generative ai gets layer

23:43

on top of that where we can create serve

23:45

the right message for them you know we know

23:47

they're demographics know the kind of products they like

23:50

to bike satirists and maybe at some point that

23:52

next step can get stitched on so in some

23:54

ways it's not let's have a whole process taken

23:56

over by dna i let's let's think about how

23:58

far we can stretch with predictive and then maybe

24:00

let's tack Gen AI on where it can add

24:02

value and it can be accumulative in that sense.

24:05

Would you concur with that general vision? Absolutely.

24:09

So brute force

24:11

information on who the customers are,

24:14

add on a predictive model that says, hey,

24:16

we think these customers are the ones that are

24:19

going to call in and then get

24:22

to a, and

24:24

let's personalize that message based on

24:26

who we are. That's exactly the

24:28

type of approach that incrementally

24:31

delivers value. Yeah. Each

24:35

step in that process is going

24:37

to create value for the

24:39

customer and the organization. Big

24:42

time. So maybe a good way

24:44

for people at home sort of thinking about what

24:46

use cases they might want to use to go

24:49

through that kind of set of progressive steps, think

24:51

about what could be an extension from here or an extension from

24:54

there, and thinking of getting started and picking use cases, et cetera.

24:57

You've given us a rule of thumb of sort

25:00

of these three dimensions to consider, some

25:02

about the cross-functional team that needs to be involved,

25:05

a good panoply of use cases where there's a

25:08

lot of value in the relatively near term. What

25:11

do you have for sort of parting advice for leaders who are thinking

25:13

about, you know, maybe they run call

25:15

center, maybe they just run customer experience in general

25:17

within Finserve. They're thinking about

25:19

how to get started. What are some of the other considerations

25:21

or takeaway notes for them? Yeah. I'll

25:25

go back to humans and machines. There

25:30

is a real fear among

25:32

customer service folks,

25:35

right? The employees that

25:38

really do the great job that, you

25:41

know, companies ask them to do and interfacing

25:43

with their clients, there's

25:45

a fear that, one, they're going

25:47

to be replaced. Two, every

25:50

word they say is going to be

25:52

scrutinized, or every email they write is

25:54

going to be scrutinized. And

25:57

there is a real opportunity.

26:00

to talk about how AI

26:03

and these capabilities are an

26:05

enabler, not a replacement.

26:09

And leading with that vision, and I'm not trying to

26:11

say, I won't speak out of

26:13

both sides of my mouth and say there's not cost

26:15

pressures. There are cost pressures. There are. Will

26:17

it result in a contact

26:19

center requiring less people?

26:21

Yes, it absolutely should. What

26:24

we've shown when

26:27

somebody in a contact center, for example, gets

26:30

to solve a client's problem and

26:32

a complex problem for a client,

26:35

their employee satisfaction goes

26:37

up significantly. Their

26:39

satisfaction with the job, their retention

26:41

rate, all of those things. So

26:44

if we can start bringing these

26:47

technologies together and

26:49

to our employees in a

26:51

way that they see them as

26:53

enabling them to solve the

26:56

problem of the customer more efficiently. If

27:00

they can move on to

27:02

solving higher level challenges, serving

27:05

clients in bigger and

27:07

broader ways, they're going to be

27:09

happier employees. And we've

27:11

all lived through the last few

27:13

years of 100% attrition within some of our

27:15

service organizations.

27:19

We can't go back to that. It's not

27:21

good for our customers. It's not good for

27:23

us. So really being

27:25

tight on the message of what we're

27:28

trying to achieve and

27:30

not being disingenuous, but

27:33

that what we're trying to

27:35

achieve is, yes, better efficiency,

27:37

better customer experience, a better

27:39

employee experience. I think

27:41

that's a key element that

27:44

organizations need to lean into

27:46

as they embark on conversational

27:49

and AI, Gen AI, good

27:52

predictive analytics, like you said, any

27:55

of those efforts. Yeah, this is what

27:57

kind of comes to mind, Anuj, as your...

28:00

Speaking, you know, in terms of not selling this, I'd

28:02

love to get your feedback. Make sure I'm not putting

28:04

this in the wrong light here. But the ideas that.

28:06

Maybe that the next step forward as

28:09

a sort of vision. For. The

28:11

contact center that, yes, maybe does

28:13

involves. Less you know of certain

28:15

kinds of redundant tas, I do think there's a

28:17

very. Honest frame here that

28:19

hey it made. we don't need as many

28:21

people who do certain things but also maybe

28:23

we want. Kind. Of a

28:25

new normal so to speak that

28:28

involves vastly higher. Employee. Satisfaction

28:30

in terms of success rate with handling customer

28:32

issues, in terms of the ease of doing

28:34

their work, in terms of taking away some

28:36

of the busy work that for the most

28:39

part is annoying for them and saying hey

28:41

yes, there's cost pressures, but can we can

28:43

we share a vision with those boots on

28:45

the ground see members and with leadership to

28:47

say let's get to a new normal where

28:49

customers are happier, people internally are happier and

28:52

we settle to that man machine combo. That's

28:54

that's gonna really hit the sweet spot for

28:56

both of those. Yeah. I

28:58

think he the what the one. Tweet

29:01

that would make his and study using

29:03

gods accent or call center I would

29:05

I would call it our a customer

29:07

experience or customer service organization and like

29:09

am I could spread. The Bank

29:11

France. Because. Accents

29:13

are. Packed. The eight

29:16

yen. Or. The

29:18

insurance agent. The. Contact

29:20

Center Magical experience as we've created

29:22

on the mobile app like that's

29:25

it's a whole gamut. Bet that

29:27

and Asia and that's my. You.

29:29

Know that that's what I would say is.

29:32

That. To really on lot. Of

29:35

power and value of these. Technologies.

29:39

We. Have to think differently. Outside

29:41

of our individual channels, there's

29:43

been so much optimization done

29:45

within each of the individual

29:47

channels that that. Future.

29:49

Optimization and value that's gonna

29:52

come from a I Jenny.

29:54

I've predictive analytics. It. Comes

29:56

from. Those. capabilities

30:00

really enabling the set

30:02

of customer service, customer

30:04

experience channels. Well, I've

30:06

certainly got my fingers crossed that this

30:08

technology can be a big part of

30:10

what makes multimodal come to life. I

30:12

know that's certainly something you believe in

30:14

and it seems like that's inevitably

30:16

the path that we're on. So hopefully the leaders who are

30:18

tuned in kind of take some of that to heart as

30:20

well. And Anuj, I know that's all we have for time

30:22

but I'm really glad we got to unpack some of these

30:24

exciting ideas. Thank you so much for being able to be

30:26

with us. It was great being with you

30:28

and thanks for letting me share, Dan. Drawing

30:39

a close to today's episode, some

30:42

points I think that were brought

30:44

up in the discussion that bear

30:46

some greater spotlight. There

30:49

are increased interactions with tech

30:51

companies raising customer expectations throughout

30:53

financial services. Due to limited

30:56

budgets, operations leaders are turning

30:58

to AI solutions for these

31:01

workflows. Anuj, throughout the episode

31:03

emphasizes the importance of directing

31:06

customers towards the most effective

31:08

channels for an optimal

31:11

customer journey and architectural

31:13

experience. Leaders often prioritize

31:15

the desirability and feasibility

31:17

of AI use cases

31:19

but clients may struggle with

31:21

implementation despite having a high

31:24

level of AI knowledge. Developing

31:26

Gen AI use cases

31:28

requires a cross-functional team

31:30

involving experts in analytics, IT,

31:33

operations and creative thinkers.

31:36

Companies will experiment with using generative

31:38

AI to assist human

31:40

customer service representatives potentially reducing

31:42

after call work time by

31:45

three minutes. Customer

31:47

facing applications of generative AI

31:49

may have less significant consequences

31:52

or be less embarrassing than

31:54

if exposed directly to customers.

31:57

Generative AI can improve customer

31:59

service. service by personalizing

32:01

outreach messages and predicting issues

32:03

before they arise. While

32:07

stretching predictive capabilities with generative

32:09

AI can add value, our

32:12

guest advises against replacing the

32:14

entire processes with that technology.

32:17

Anuj emphasizes furthermore that

32:19

despite significant cost pressures

32:21

associated with generative AI,

32:23

it can lead to

32:25

a reduction in contact

32:27

center staff without compromising

32:29

quality. On

32:32

behalf of Daniel Segella, our CEO and Head

32:34

of Research, as well as the rest of

32:36

the team here at eMERGE Technology Research, thanks

32:38

so much for joining us today and we'll

32:40

catch you next time on the AI and

32:43

Business Podcast. Thank

32:53

you.