0:00

Welcome back to Data Driven,

0:02

the podcast where we explore

0:04

the ever-evolving worlds of data

0:06

science, artificial intelligence, and the

0:08

unsung hero behind it all

0:10

data engineering. Now, unfortunately, our

0:13

favorite data engineer couldn't

0:15

make it today. But fear not. Frank

0:17

is here, and is joined by a

0:20

fantastic guest, San Jay Anadate, VP

0:22

and business head of EMEA at Layton

0:24

View. With nearly three decades

0:26

in the industry... Sanjay has

0:28

witnessed digital transformation from its

0:31

infancy to its AI-infused present.

0:33

In this episode, he and

0:35

Frank dive into Microsoft's role

0:37

in the data revolution, the

0:39

rise of AI-powered automation, and

0:41

why businesses that don't embrace

0:43

data-driven decision-making are basically living

0:45

in the past. So grab your

0:48

favorite beverage, sit back, and let's

0:50

get data-driven. Hello

0:54

and welcome to Data Driven the

0:56

podcast. We explore the emergent fields

0:58

of data science, artificial intelligence, and

1:00

the massive technology underpinning it all.

1:02

It is data engineering. However, my

1:04

favorite is data engineer in the

1:06

world will not be able to

1:08

make it today, but the show

1:10

must go on and I am

1:12

pleased to welcome Sanjay Anadate, who

1:14

is the VP and business head

1:16

of Amia at Layton View, and

1:18

just looking at his LinkedIn

1:20

profile, he definitely has it

1:22

very rich. history in the

1:24

industry. So welcome to

1:26

the show Sange. Thank you.

1:29

Thank you, Frank. And

1:31

it's a pleasure to talk

1:33

to you today. Awesome. Awesome.

1:35

So, um, you are

1:37

a listed as a

1:40

digital transformation leader. Tell

1:42

me, what does digital

1:44

transformation mean

1:46

to you? Yeah. So, Frank, first

1:49

of all, Once again, thank

1:51

you for the opportunity on this

1:53

podcast. I have been in the

1:55

IT industry for almost three decades

1:58

now and I have seen the

2:00

era before digital transformation

2:02

and the era during

2:05

and post-digital transformation.

2:07

I think if I

2:09

could explain digital transformation

2:11

in my words, mainly

2:13

it covers the core

2:15

areas such as the cloud,

2:17

such as the data, the

2:20

processes, and the technology. And

2:22

mainly what I had seen, if

2:24

I see the last decade, it

2:26

started with mainly first cloud

2:28

transformation, then the large

2:31

companies went through multiple

2:33

systems transformation. And in

2:35

last three, four years,

2:37

we are seeing significant

2:39

uptake of the data

2:42

and analytics transformation. Means

2:44

even when the large

2:46

customers were going through

2:48

huge cloud and application

2:50

transformation. Data transformation was

2:53

still not the top

2:55

priority. They were still

2:57

focusing on trying to

2:59

optimize applications, trying to

3:01

optimize the infrastructure. But last,

3:03

say, three, four years, we

3:05

have seen a significant increase

3:07

in terms of really focusing

3:09

on data-driven transformation. I think

3:12

that's where the change I am

3:14

observing in the market. That

3:16

drives pretty well what I've seen

3:18

too and it's like you know it's

3:21

you mentioned you've been in industry for

3:23

a couple decades three decades and I

3:25

too were roughly the same age and it

3:28

takes it almost takes your breath away

3:30

and that's cheesy to say when you

3:32

think about how far this is calm

3:34

right like you know I remember yeah

3:36

starting my career and you know they

3:39

still had was a 10 base T

3:41

network connections and token ring was still

3:43

a thing and like you know one

3:45

of You know, because, you know, we

3:47

were kids, we were crawling around under

3:50

desk plugging in Ethan, the new fangled

3:52

unshielded twisted pair stuff. And like, you

3:54

know, it's just amazing because you go

3:56

back as like computers, you know, they

3:58

were, you know, I really started

4:01

during the client server error

4:03

and how that has evolved and

4:05

changed. I mean, I remember when the

4:07

internet was this, you know, to a

4:09

lot of IT people was this big

4:12

scary place. And, you know, now for

4:14

a lot of organizations,

4:16

cloud is just a thing, like

4:18

it's just normal, right? Yeah. And

4:20

you don't really think about that.

4:22

And you're right, like over the

4:24

last decade or so, really the

4:27

idea of using data to inform...

4:29

how you're going to digitally transform,

4:31

right? Even the term digital transformation

4:33

isn't that old either, right? Like,

4:36

and it means a lot, that's

4:38

why I ask, because it means

4:40

a lot of different things to

4:42

a lot of different people, to the

4:45

point where I think, I think the term

4:47

is really lost all meaning. And

4:49

I have also seen kind of

4:51

the rebranding of it as, you

4:53

know, AI infused digital transformation. But

4:56

you have a pretty solid background

4:58

in data and analytics and particularly

5:00

with the Microsoft stack if I

5:03

read that right. So what's your

5:05

take on where Microsoft is with

5:07

all of this? Obviously I'm a

5:09

X Microsoft employee, X Microsoft

5:12

MVP, still very friendly to Microsoft.

5:14

the house I live in was

5:16

built by was paid for by

5:18

Microsoft. So I still have a

5:20

lot of love in my art

5:23

for for Microsoft. What's your thought

5:25

on what they're doing now? Yeah,

5:27

so I think Microsoft

5:29

again has been always

5:31

heavily focusing on investing

5:33

into the digital transformation,

5:35

investing into the technologies

5:37

which really helped the

5:40

customers. I think the

5:42

big advantage Microsoft had was

5:44

like so many millions and millions

5:46

of customers were actually using the

5:49

windows and the operating system. Naturally,

5:51

they got that audience. Where Microsoft

5:53

really had changed the game is

5:55

basically trying to really scale up

5:57

the zero investment and trying.

6:00

to drive the capabilities on

6:02

the back of that. And

6:04

naturally in last two years,

6:06

the open AI and the

6:08

chat GPT investments into that

6:10

naturally has significantly. given the

6:12

acceleration and the momentum to

6:14

the Microsoft. So see Microsoft

6:16

is heavily dominating this space

6:18

and I think they'll continue

6:20

to do so because they

6:22

have such a huge market

6:24

share in the across the

6:26

globe. And the thing thing

6:28

is they're also heavily focusing

6:30

on the industries as well.

6:32

It's not just about investing

6:35

in technology but they're also

6:37

investing in the industries, trying to

6:39

really help. multiple industries to

6:41

build capabilities which are more

6:43

industry specific. I think that's

6:45

where they are also trying

6:47

to understand just not remain

6:49

within the cloud era, not remain

6:51

within only the data or or

6:54

the LLLM Sarah, but they are

6:56

also trying to look into every

6:58

industry and making the capabilities or

7:00

the products which are more relevant

7:03

to the industries. So whether it

7:05

is CPG, whether it is retail,

7:07

whether it is industrial, they are

7:09

also focusing on building industry specific

7:12

capabilities. So I think they

7:14

are heavily. heavily focusing

7:17

towards customer base across

7:19

the industries and not

7:21

just not just trying

7:23

to remain very cross

7:25

industry specific technology

7:28

provider. Yeah, I mean, that makes

7:30

sense. And I've seen their

7:32

evolution go from, you know, sequel

7:34

server. It's funny. I remember the day

7:37

I first heard a sequel server and it

7:39

was. It was kind of like, you know,

7:41

I think at the time internet information server

7:43

was out and I was just like, they're

7:45

like, no, you connected to sequel server. Somebody

7:47

said, you can connect us to sequel server.

7:49

And this is like 1996, right? So put

7:52

the thing. And I'm like, what's sequel server?

7:54

And like Microsoft has a database server now?

7:56

And like course, it was, you know, the

7:58

world was dominated by oral. and it was

8:00

kind of like really why are

8:03

they going after that market that

8:05

seemed kind of odd and don't

8:07

send me hate mail listeners like

8:10

this was 1996 when I said

8:12

that now obviously obviously the product

8:14

has changed I'm fascinated by what

8:16

you know how that product has evolved

8:19

in itself right you have

8:21

Azure sequel there's multiple ways

8:23

it's just fascinating what's your

8:26

and this is where I

8:28

wish Andy were here What's

8:30

your take on Microsoft Fabric?

8:33

What's your thoughts on

8:35

that? Yeah, I think

8:38

the... Microsoft Fabric as

8:40

a capability for the

8:42

market is really amazing.

8:45

What it provides, basically

8:47

it significantly simplifies the

8:49

architecture specifically, and naturally

8:52

the global customers have

8:54

got very very complex

8:56

architecture, which where basically

8:58

you integrate multiple applications,

9:01

multiple databases, multiple

9:04

front ends and the servers.

9:06

What fabric provides is an

9:09

ability to really integrate

9:11

multiple, not only multiple

9:13

applications, but multiple clouds using

9:15

one single fabric. And it

9:18

provides that overarching view to

9:20

the to the administrators, to

9:22

the business, to really simplify

9:24

the whole landscape in a

9:26

big way. And I

9:28

think that's what that's

9:30

basically significantly differentiating Microsoft

9:32

at this point of time.

9:35

So there are customers who actually have

9:37

not just as here as a cloud,

9:39

they also use AWS, GCP, but they

9:41

can still use the fabric as a

9:44

as a layer, which can really

9:46

integrate multiple cloud. And I

9:48

think that's what. Hughes is

9:50

giving Microsoft a significant edge

9:52

in terms of not only

9:54

it is reducing the investments

9:56

the customers are making in

9:58

the technology space. It is

10:01

significantly simplifying the architecture. It is

10:03

also helping them accelerate the innovation

10:05

because what has been happening in

10:07

the past, if you have a

10:09

very very complex architecture, naturally it

10:12

impedes your speed to innovation because

10:14

you are not able to really

10:16

scale up the applications at pace

10:18

and at scale because you have

10:21

to really go through multiple architecture

10:23

reviews and tests and everything. But

10:25

what fabric is giving them

10:27

an ability to really integrate

10:29

all this. all these systems

10:31

without really putting another layer

10:34

of applications on top or

10:36

another level of integration on

10:38

top. And that's where it

10:40

is, it is really driving

10:43

the whole pace of innovation,

10:45

pace of the whole transformation

10:48

across the customers base. So,

10:50

no, I think I have seen and

10:52

we are serving already few

10:54

customers in Europe as well

10:57

as. in a pack where customers

10:59

are quite keen to go towards

11:01

fabric because and the last thing

11:03

is now with the open open

11:06

AI many of them are really

11:08

focusing on co-pilot as well and

11:10

because co-pilot comes pretty handy with

11:12

the with the fabric and actually

11:14

they are trying to leverage the

11:16

power of co-pilot on top of

11:19

the fabric so so yeah I

11:21

think it is it is really

11:23

really doing well in the market.

11:25

And obviously the data privacy sovereignty

11:28

etc. laws are much stricter there

11:30

than anywhere else in the world

11:33

really I think the US well

11:35

certainly but people also don't realize

11:37

that Latin America also has has

11:40

some legislation in that regard to

11:42

most people who wouldn't know that

11:45

yeah but how do how do

11:47

customers how do enterprise customers

11:49

kind of deal with the notion of

11:51

you know if I'm in a

11:53

country and I'm going to an

11:56

Azure data center, right?

11:58

But there's no. hazard

12:00

data center within my

12:02

country. I mean, does that

12:04

go over well? Do customers

12:06

respond positively to that? Or

12:09

is it, what's been your experience?

12:11

Like, what have you seen? Yeah, I

12:14

think that's a very good question,

12:16

Frank. See, and I think

12:18

to be very frank, my

12:20

core is also quite aware

12:22

of this regulations. So they

12:24

keep a very close eye

12:26

on this. Specifically for the

12:28

enterprise customers, Microsoft has built

12:30

in multiple data centers across

12:33

Europe as well. They do

12:35

have data centers in Europe.

12:37

So data residency standpoint,

12:39

I think that's generally

12:41

quite well addressed by them.

12:43

And also many times the

12:45

applications or the data. if

12:47

it is not really sensitive

12:49

information, then customers are also

12:51

fine to have it outside of

12:54

Europe. But if it's a sensitive

12:56

information, naturally it has to reside

12:58

either in the country or in Europe.

13:01

And for that, I think Microsoft has

13:03

gated very well. So they

13:05

do have investment. They do have

13:07

data centers in Europe, and

13:09

which does cater to the

13:11

GDPR regulation very well. So I

13:14

think on that front, they really

13:16

are. doing like they don't have

13:18

any challenges as such from the

13:20

data privacy standpoint. Interesting.

13:23

No, because like in my

13:25

day job, you know, you know, our big,

13:27

our big marketing push at

13:29

redhead for redhead AI is

13:31

the idea that not only is it

13:34

closer to the open source, but

13:36

we provide the tools that you could

13:38

work, run this workload anywhere,

13:40

it makes sense for your

13:42

business. And there's a lot

13:44

of concern in places where

13:46

there are these these heavy

13:49

regulations and you know

13:51

Microsoft certainly has been building a

13:53

lot of data centers a long

13:55

time ago when I once wore

13:57

a blue badge Microsoft.

14:00

will get that reference. There was

14:02

Mark Vosinovich was speaking and most

14:04

of what he says is kind

14:06

of you know don't don't repeat

14:08

after this room but one of the

14:10

things is that he kind of revealed

14:12

is again this isn't super secret

14:14

information. I don't think that

14:16

Microsoft ever really planned to build this

14:18

many data centers. I think the thinking

14:21

was they would build maybe a dozen.

14:23

Right. You know, and then everybody would

14:25

run their workloads, but the politicians and

14:28

lawyers came in and kind of ruined

14:30

that plan. That's my phrasing, not his.

14:32

So now you have this notion where

14:34

they're just trying to build these data

14:37

centers in different sovereign territories as quickly

14:39

as they can. And I do think

14:41

that there's probably going to be a

14:43

max out of, you know, you're not

14:46

going to have a data center in

14:48

every country. I don't think that's. only

14:50

anyone's ever going to have that. But

14:53

you're going to get them in

14:55

most places. And you're right,

14:57

Microsoft has been very responsive to

14:59

this. I also think too, like

15:02

the idea that Faber can kind

15:04

of work with, and there's

15:06

another technology that they're,

15:08

that Azure has too, where, you know, Azure

15:11

is the plane of, the pain of this,

15:13

where maybe I live in a country

15:15

where I do business in and there's

15:17

no Azure data center there. But

15:20

there's an AWS one or Google or IBM

15:22

or Oracle or whoever. Like I would be

15:24

able to kind of like have one paint

15:26

of glass control all of that. I think

15:29

that that's also a clever

15:31

mitigation strategy too. Yeah,

15:33

I think that's what the with that

15:35

concept only I think the fabric

15:38

came into means fabric one of

15:40

the key objective of the fabric

15:42

was to to really provide that

15:44

ability to integrate multiple clouds and

15:47

and you don't really have to

15:49

go. It means what Microsoft means

15:51

of course I'm not speaking on

15:53

behalf of Microsoft but my

15:55

view is that Microsoft wanted

15:58

customers to have the ability. to

16:00

go for multi-cloud strategy,

16:02

but still have that

16:04

integration layer with the

16:06

Microsoft fabric so that

16:09

they customers really don't

16:11

need to go for

16:13

every single specialized integration

16:15

ability capability with each

16:17

cloud. And I think

16:20

that also gives Microsoft

16:22

differentiation that. your customers have

16:24

ability to go for multiple

16:27

cloud but still retain as you

16:29

are as a key component key

16:31

key capability in their landscape

16:34

right absolutely and it also

16:36

is clever too because a

16:38

lot of major organizations large

16:40

organizations they already have

16:43

office they already have

16:45

active directory or yeah exactly

16:47

what to call that intro

16:49

or entrata Oh, I is sleeping

16:51

my mind too. I'll look it up.

16:53

But all the, even office has

16:55

changed his name too, like it's

16:57

not Office 365, it's. Office 65,

16:59

I think it was, yeah, I

17:01

think with the copilot there,

17:04

they're changing that as well.

17:06

But you're right, I think,

17:08

and it's copilot becoming

17:10

a standard feature as part

17:13

of Microsoft Office now. It's

17:15

also available to all the

17:18

enterprise users. So. So I

17:20

think it's really driving it

17:22

very very nicely, the adoption

17:25

across the across the businesses

17:27

and many of the enterprise

17:30

customers, whom we are

17:32

talking, it kind of started

17:34

using Co-Pilot already.

17:37

And the feedback has been

17:39

pretty good. I have a

17:41

business level tenant of office.

17:44

whatever it's called now for myself

17:46

and I recently I enabled co-pilot and

17:48

I'm kind of I see its integration

17:50

in all these places and I'm like

17:53

I turned it on just because I

17:55

want to play with it and I

17:57

have to say I'm impressed I mean

18:00

from the obvious, like, helping you

18:02

write a Word document to help

18:04

you write an email to, the

18:06

not so obvious in terms of

18:09

the administrative tasks that it can

18:11

handle. I'm kind of excited about

18:13

it, you know, and the back

18:15

of my mind is that little

18:18

paranoid side of where's my data

18:20

going, but, you know, I can easily

18:22

see this being a core feature and,

18:24

and, and, and, of that. And I

18:26

just looked it up. as your active

18:28

directory is now known as Microsoft

18:31

Entra ID. They made the change

18:33

in 2023. And I guess when

18:35

you get to a certain point

18:38

in your career, the years just

18:40

start flying by, like 2023 was

18:42

two years ago, which I can't

18:45

get my head around. Yeah, so

18:47

I've also been seeing a

18:49

lot of information about how.

18:51

co-pilot exists now in fabric.

18:54

It exists now in some

18:56

capacity in as your sequel

18:59

or that. What's your take

19:01

on that? Because obviously, get

19:03

hub co-pilot, which I'm a big

19:05

fan of, has been around a

19:07

couple of years now, but. It

19:10

seems to be expanding not just to

19:12

the consumer, business user information, worker space,

19:14

but also the developer, but also the

19:16

data engineer slash data developer. What's your

19:18

take on that? Does that, you think

19:21

that's going to replace data engineers? Or

19:23

do you think it's going to just

19:25

make them more productive? No,

19:28

no, it's not going to

19:30

replace the data engineers for

19:32

sure. I think what it

19:35

will help is the data

19:37

engineers to do the job

19:39

much more efficiently and maybe

19:41

much more effectively. It is

19:43

going to be more of

19:46

an assistant to them, support

19:48

them in terms of trying

19:50

to really improve the productivity.

19:53

Also, there is, see the standard.

19:55

Use cases are pretty good with

19:57

the copilot. Actually, you can get.

20:00

forward answers. Naturally for

20:02

custom complex questions you

20:04

still will need to use

20:06

something more customized more. So

20:08

as your AI is

20:10

what you will need to

20:12

really do so utilize

20:14

to build answers to

20:17

complex questions complex queries. So

20:19

there is of course some amount

20:21

of customization will be required for

20:23

the co -pilot for business users

20:25

to really build what do

20:27

you say tailored capability

20:29

tailored response to their

20:32

questions but by

20:34

if you are looking for the

20:36

standard answers like which are straightforward

20:38

queries hey how has been my

20:40

business doing for last three

20:42

months or six months I think

20:44

co -pilot can straightforward give you

20:46

the answers who are my

20:49

top five customers that that's those

20:51

but when there is a

20:53

need for you to integrate multiple data sources

20:55

and build a very complex response naturally

20:57

you will have to build you're

20:59

going to customize those kind of

21:01

questions using the the

21:03

programming. Interesting yeah

21:07

I just think it's you know I kind

21:09

of go back and forth between oh

21:11

no this is the end of all jobs

21:13

and you know no this is a

21:15

new era of productivity and and and

21:17

yeah and whatnot I see it

21:19

in myself right like I you know

21:21

using different AI tools you

21:23

know whether it's you know the

21:25

the co -pilot whether it's

21:27

chat GPT whether it's running granite locally

21:30

and kind of like some of

21:32

the stuff that you know we do

21:34

that's closer to the open source

21:36

side in my day job my productivity

21:38

has gone up like I mean

21:40

and like my same level

21:42

of time that I devote right

21:44

and and when it comes

21:47

to like my own personal stuff

21:49

like Franksworld .com this podcast impact

21:51

quantum the quantum computing podcast

21:53

I'm able to produce content at a scale that

21:55

I think not just a few years ago

21:57

I would have had to hire at least several

22:00

virtual assistance and kind of

22:02

like just production people where now

22:04

I can kind of like you know

22:06

I can copy and paste content or

22:08

upload content from one form to another

22:10

and it generates it. Now the next

22:13

step would be if I can kind

22:15

of have an agent sit above that

22:17

and move stuff around without me being

22:19

involved right other than the original initial

22:21

content but for me it's opened up

22:23

much many more opportunities right like

22:26

in terms of you know people

22:28

I'm not able to reach these

22:30

sponsorships that were able to get

22:32

it's really and it's not impacted

22:34

my you know I can still you know even

22:36

though I work from home I can still

22:38

be home for dinner you know what I

22:40

mean like I can I don't have to

22:43

do kind of all the legwork you used

22:45

to do and yeah I I I I'm

22:47

a big fan of it I really think

22:49

the augmentation story is very real.

22:51

I think there is going to

22:53

be some. Yeah. There's going to

22:56

be some impact on available jobs.

22:58

I think that's unavoidable. But historically,

23:01

as I like to say, every

23:03

wave of automation has over

23:05

the long haul created new

23:07

jobs and new opportunities. And

23:09

I'm hoping that the pattern

23:11

holds. What's your take on that?

23:13

No, I fully fully agree. I think.

23:16

the open AI or the

23:18

chat activity or LLLMs as

23:20

such has massively improved productivity

23:22

for every like almost everyone

23:25

whose over is using it

23:27

and it is really it

23:29

has really acted as a game

23:32

changer I must say and not

23:34

just us but whenever I'm talking

23:36

to the business now there is

23:38

a significant appetite for

23:41

really leveraging the the

23:43

open AI for automating

23:45

multiple proper tasks

23:48

like if I can give

23:50

an example a large bank

23:52

we have recently completed

23:54

a project for

23:56

they are preparing a

23:58

sales briefing document for

24:01

this sales force using LLLM's.

24:03

Now, this used to be the

24:05

task they were performing over four

24:07

days. So if a sales executive

24:09

talks to the customer, he used

24:11

to take four days to prepare

24:13

a briefing document before

24:16

speaking to a customer. And

24:18

that task is now done using

24:20

the LLLM programming. Within a minute,

24:22

he just fires a prompt and

24:25

he gets a PowerPoint

24:27

briefing document. at his desk

24:29

on his on his machine

24:31

within within minutes. And this this

24:33

is the level of automation. These

24:35

are the level of impact. Of

24:38

course, to program that kind of

24:40

LLLM, it is not just it

24:42

doesn't happen on the fly. You have

24:44

to build and train the models

24:46

to really reach that level. But

24:48

once it is done. Then the LLMs

24:51

can take it forward. You don't

24:53

really then need to monitor it.

24:55

It will just read through your

24:58

data. It will read through the

25:00

websites, PDFs, and prepare the document

25:02

which is most up to date. So

25:04

it is mind blowing. Honestly, like,

25:06

if I am the sales executive

25:09

and if I was taking four

25:11

days time and I now get

25:13

that whole document ready within two

25:15

minutes, can imagine I can now

25:17

use those. those four days to

25:19

talk to many customers. So

25:21

absolutely, absolutely. I'm sorry, cut

25:23

you off. I'm so excited.

25:25

Yeah, yeah, that's that's that's

25:27

the level of savings each

25:30

individual is going to get naturally

25:32

the to reach to the level like

25:34

you have to do some 18

25:36

weeks of work to really bring

25:38

the the do the prompt engineering

25:40

to ensure that the capability is

25:42

built. But once it is built. Then

25:45

you can imagine every single sales

25:47

executive can use those that

25:49

application to To really build

25:51

a briefing document and you

25:53

can imagine like the large

25:55

banks has like at least hundreds

25:57

of sales guys and so saving

26:00

they're going to have is massive. No,

26:02

that's true. And, you know, one of

26:04

the criticisms I heard as well, you

26:06

know, that document was not put together

26:08

by humans, right? Yeah. But if you

26:11

did it in two minutes and if

26:13

you're a, if you're a clever sales

26:15

executive, which one would hope that they

26:17

are all clever, you will use that

26:19

extra time you get back.

26:22

to provide even more personalized

26:24

interaction with customers. Exactly. Exactly.

26:26

Exactly. So you can have

26:29

a higher, ironically, the AI

26:31

gives you more opportunities, not

26:33

guarantees, to make a

26:35

more personalized touch with your

26:37

customers. Exactly. And what it

26:40

will also do, Frank,

26:42

actually, let's say you are

26:44

a sales executive for X customer,

26:46

but you have taken up a job.

26:49

very recently, but before you,

26:51

somebody else was doing your

26:53

role. What this capability now does

26:55

is it avoids the dependency on you.

26:57

It will go through the previous

26:59

conversation and bring all the

27:01

data into that briefing document.

27:04

So you really don't have

27:06

to go through all the

27:08

previous conversations and communications and

27:10

then prepare the document. So

27:12

it significantly reduces. the

27:15

dependency on the sales

27:17

executives as well because if it is

27:19

somebody falls sick the next

27:21

guy can pick it up and

27:23

then just fire the query get

27:26

the document and have conversation so

27:28

it is it is really huge hugely

27:30

beneficial for and that's

27:32

what we are seeing across the

27:34

market customers are really focusing on

27:37

hey how can I automate how

27:39

can I really make the

27:41

task much more impactful and

27:43

much more effective using this

27:45

because manually spending four days

27:47

time to prepare we can

27:50

review an document there

27:52

was no option in the path but

27:54

now they have the option so why

27:57

not to use the the capability?

28:00

He can use the time to really

28:02

make it much more impactfully. He can

28:04

use that for his time to now

28:06

add much more insights into it to

28:08

make it much more personalized. No, absolutely.

28:11

And you know, I think it's this notion

28:13

of high value activity versus low

28:15

value activity. And then I mean, there

28:18

it is right there, right? Like where

28:20

before you really didn't have a say

28:22

because you had to do it, right?

28:24

Like you couldn't separate the low value

28:26

activity from the high value activity.

28:28

Now you can. And I think

28:30

it's those organizations that

28:32

realize that, you know, you can, you

28:35

can separate this and then

28:37

focus lean heavily on the high-value

28:39

activities. I think those are the

28:41

organizations that are going to win

28:44

in this in this AI infused

28:46

age, right, or augmented age, whatever

28:48

the, whatever the buzzword is this

28:51

week. I don't know. I haven't.

28:53

Right. Yeah, I think the buzzword

28:55

now is everyone is talking about

28:58

agentic AI. I mean, they are

29:00

now not just focusing on LLLMs,

29:02

but now going toward a step

29:05

further and talking about agentic AI,

29:07

which is going to be further

29:09

more value addition. So LLLMs typically

29:11

have been used to create

29:13

the response, text response or

29:16

images or something like that.

29:18

Agentic AI is actually going

29:20

to automate multiple. task and

29:22

not just automation of the

29:24

task like a BPA, the

29:26

business process automation has

29:29

done, but what Agent I is

29:31

going to do is going to

29:33

infuse AI in it. So it

29:35

is going to comprehend, it's going

29:37

to take actions like a human

29:39

do based on the previous

29:41

interactions, previous actions. And I

29:43

think again, Agent I mean,

29:45

it's very early days so

29:47

far, but When it becomes full

29:49

scale, it is also going

29:52

to make a significant impact

29:54

on the productivity and the

29:56

business processes because it is

29:58

going to really help. businesses to

30:00

really transform their back of

30:02

his front of his processes.

30:05

Yeah I keep going back and forth

30:07

on this because like some of

30:09

the agentic AI stuff I've seen

30:11

is basically they slap an AI label

30:13

on RPA right that's what I've seen

30:16

but but I also say that

30:18

if you do it right there

30:20

is an enormous level of transformation

30:22

that you can have even with

30:24

just my little like content little

30:26

I call it content empire, but you know

30:29

what I mean? But like, there's stuff that

30:31

I have to do that I'm thinking to

30:33

myself, why do I have to, you know, drag

30:35

this file to there or, you know, do

30:37

this or get my involvement. It'd be nice

30:40

if I could just, you know, basically say

30:42

to an LLLM on my phone, hey, can

30:44

you take this original recording that I'm doing

30:46

with Sanjay and convert it, get the transcript,

30:48

convert to a draft blog article,

30:50

oh, and, you know. pipe it

30:53

over to notebook LM so I

30:55

can get kind of like them

30:57

to do the NPR style podcast.

30:59

Like I do that now, you know,

31:01

manually, right? But again, it's funny

31:03

because not that long ago that

31:05

none of that was an option

31:08

and I didn't create any of

31:10

those. So it's kind of funny

31:12

how quickly humans adapt to

31:14

convenience. Yeah, no, I think

31:16

that's that's that's a very fair

31:18

point. What I guess I have

31:21

seen, Frank, Frank, is You

31:23

are right, like I think a lot

31:25

of the, there is a, I

31:27

would say, misconception about agentic

31:29

AI, it just basically

31:31

AIS slapped on RPA means,

31:34

but the, but the true

31:36

differentiation between RPA and

31:38

agentic AI is the

31:40

context setting actions. So the

31:43

agentic AI will not just

31:45

to mechanical task automation, which

31:48

RPA has done in the past.

31:50

So RPA was typically what a rule

31:52

based automation of the business process. So

31:54

you define the rule and you'll just

31:56

keep following that rule hundreds and

31:59

hundreds of times. It will just, if

32:01

the rule fails, the automation, the

32:03

process will fail, otherwise the

32:05

rule will just follow the

32:07

process. What Agent I had does

32:09

is it will apply intelligence on top.

32:11

So in case the process is failing

32:14

due to some reason, it will

32:16

apply, hey, what happened in the

32:18

last time when it failed, apply

32:20

the correction and then continue.

32:22

So this is the difference between RPA

32:25

and the agentic AI. This is

32:27

more context setting. You can define

32:29

your context, you can define

32:31

your objective goal. So it will

32:34

keep learning, it will keep adopting

32:36

to the previous historic corrections and

32:38

it will not be that every

32:40

time the process fails you have to

32:42

correct it. System will correct on its

32:44

own. So that's where the agentic

32:47

AI will, it's not just the... replicate

32:49

like replacement of RPA. It is

32:51

much more than replacing RPA. I

32:53

like that. I think that's the

32:55

best single definition I've heard,

32:57

right? Because it addresses the

33:00

fact that a genetic AI,

33:02

if you define it as it's an AI

33:04

that actually has agency or it does

33:06

something, right? That covers a very

33:08

wide swath thing. But I think you're

33:11

right. I think the self learning and

33:13

self correction. has to be a big

33:15

part of it in order for

33:17

it to be really something new.

33:20

Not everything called agentic does

33:22

that, but I think you're

33:24

right. I think as an

33:26

idea, that is the future. That's

33:28

correct. Yeah. And I think

33:31

so it's again, same thing.

33:33

It's not going to be

33:35

straightforward like RPA where business

33:37

can just. train the RPA tool

33:39

using the business process and once

33:41

you once a RPA learns the

33:43

new process and basically it keeps

33:46

repeating it is not going to

33:48

be like that because this will

33:50

be for the complex business processes

33:52

where which has not been automated

33:54

in the past I think the agent

33:56

again will really help them because

33:58

in those processes you have to

34:01

apply intelligence. You have to really

34:03

apply. It's not just a rule-based

34:05

automation. It is basically more intelligence

34:08

where human intelligence comes in. I

34:10

think that's where the agent again

34:13

will help businesses to automate complex

34:15

tasks, which historically has

34:17

not been possible using RPA. Excellent.

34:19

Excellent. I think that's a really

34:22

good way to look at it. So what

34:24

are you excited most about in terms

34:26

of where this technology is headed?

34:30

So I think the next five years

34:32

I mean in the past we

34:34

used to talk about decade I

34:36

think now even not even five

34:38

years like it's maybe year or

34:40

two is going to be heavily

34:42

dominated by by AI and as

34:45

I said as I said not

34:47

just LLLMs but agent take care

34:49

they will go in tandem so

34:51

more for when you when the

34:53

customers are more looking forward for

34:56

text-based image creation, it's LLLMs, but

34:58

when it is really automation of

35:01

complex processes, they will go

35:03

for for agent I think the

35:05

adoption of AI across the business is

35:07

going to be massive. And we

35:09

already started seeing large

35:12

businesses already cutting multiple

35:14

hundreds of workforce because they

35:16

already started realizing the benefits

35:19

of AI. So I think

35:21

the new era of AI

35:23

is going to start means

35:25

already started rather than and

35:27

it's going to scale up

35:29

every single business will scale

35:31

up the investment. However I

35:34

think in Europe if I see

35:36

the GDPR and and the AI

35:38

regulation will mean that they will

35:40

have to stay within the legal

35:42

boundaries so naturally they will have

35:45

to keep an eye on that.

35:47

Also sustainability is very very crucial

35:49

within the Europe context so they

35:51

will also have to ensure that

35:53

they cannot just take eyes of

35:56

the sustainability target but regardless

35:58

of that investment will be

36:00

there. Adoption standpoint, they will be

36:03

slightly behind the curve than the

36:05

North America, but I think they

36:07

will certainly invest and Europe has

36:10

already committed 200 billion into the

36:12

AI, so actually they cannot stay

36:15

behind. So they have already pledging

36:17

huge amount of investments in the

36:19

AI. And I think that's what

36:22

will be the main thing for

36:24

next two or three years. Interesting,

36:26

interesting. And you're right, things really

36:29

are speeding up. Like, like, I

36:31

gave a talk earlier this week

36:33

and I had it, it was

36:36

about open source AI and ethics

36:38

and I had a whole bunch

36:41

of stuff just based on like

36:43

what happened since I last gave

36:45

that talk in December, right? Yeah.

36:48

I had an entirely new section

36:50

because, you know, all this stuff

36:52

just keeps changing so fast. So

36:55

this is a point of show

36:57

where we're going to switch to

37:00

the pre-canned questions. I don't think

37:02

we have time to do them

37:04

all. But the first one I

37:07

want to know is, because you

37:09

have a very interesting history, CV,

37:11

as you would say in the

37:14

UK, how do you find your

37:16

way into data? Did you find

37:18

data or did data find you?

37:21

Yeah, very, very good question. When

37:23

I was with, my current role

37:26

is purely with data and analytics,

37:28

but when I was with Accenture,

37:30

specifically when I shifted to Microsoft

37:33

Business Group role, which I did

37:35

towards last year and half within

37:37

Accenture, I saw a significant shift

37:40

in terms of the deals we

37:42

were like going after in the

37:45

market, large significant amount of large

37:47

customers were really investing in data-to-to-to-date

37:49

transformation. And what it gave me

37:52

the signal that, hey, data is

37:54

going to be the big market

37:56

because multiple customers are really now

37:59

focusing on leveraging the. insights and

38:01

analytics to transform the business. No

38:04

more, just a system transformation,

38:06

no more just a cloud

38:08

transformation, but proper data driven

38:11

transformation was the one which was

38:13

really going through. And that just gave

38:15

me the inside that, hey, this is

38:17

a space which is going to go

38:19

significantly. And that's why when the opportunity

38:22

came with my current rule, I moved

38:24

on to that. And I. don't regret

38:26

because this this space is

38:28

is really growing so well

38:30

every single customers now really

38:32

want to they everyone wants

38:35

to talk about data driven

38:37

transformation and naturally

38:39

they cannot stand behind and

38:41

with the AI is is going to

38:43

further accelerate the momentum so

38:46

that's how actually to be very

38:48

frank even in my role within we

38:50

prove we were doing all

38:53

digital transformation. But as I

38:55

said to you, when we were

38:57

doing digital transformation 10 years ago,

38:59

it was mainly moving this application

39:02

from on-premise to cloud,

39:04

moving to a higher version

39:06

of SAP or higher version

39:08

of Oracle or implementing Salesforce,

39:11

nobody was really talking about

39:13

data-driven transformation. It's just the

39:16

last five, six years, customers

39:18

have really shifted to... to really

39:20

leverage the power of analytics

39:22

and data. And yeah, that's

39:24

how I shifted to this

39:26

space. So it's both a

39:28

combination of data found me or

39:30

I found data, but I saw the 10.

39:32

Yeah, no. And I think that's a lot

39:34

of people find that it's a

39:37

mixture of both. I think it's

39:39

fascinating how careers kind of find

39:41

us and we kind of find

39:44

them. It's never one or the

39:46

other. It's always some combination of

39:48

the two. What's your

39:51

favorite part of

39:53

your current gig?

39:56

Current role. Current

39:58

role. Yeah. So. What I

40:00

like about my current role, Frank,

40:03

is we are able to really

40:05

solve very complex problems for the

40:07

business using the data and analytics.

40:09

And I gave you the example

40:12

of the sales briefing document, which

40:14

we help them like reduce from

40:16

four days to two minutes. Now, those

40:18

are the kind of capabilities when we

40:21

deliver to customer. We really, I

40:23

really get a lot of satisfaction.

40:25

The other example I could give

40:27

you is we held our large

40:29

consumer, the CPG client, to

40:31

really drive the deterrent even innovation

40:33

when they were trying to launch

40:36

the new products in the market.

40:38

And historically, the innovation

40:40

process, new product innovation, or

40:43

new product development processes to

40:45

take 18 months, but

40:47

leveraging this data and innovation, they

40:50

could reduce it to six months.

40:52

So these are the like a massive

40:54

impact we are able to make to the

40:56

on the business on the

40:58

business performance. I think that's

41:00

what gives us real satisfaction

41:03

that we are able to

41:05

solve complex problems using the

41:08

data and analytics capabilities.

41:10

Interesting, interesting.

41:12

Where can people find more about

41:15

you and your company? So

41:18

we are listed company, so

41:20

we are of course available

41:23

publicly, we have a website

41:25

latent view.com. You can, I'm

41:28

on the leadership team within

41:30

the Layton View website, so

41:32

you can always reach out

41:35

to us. Otherwise, my details

41:37

are Sanjay.anadate at Layton View.com.

41:39

Please feel free to reach

41:42

out. I'm on LinkedIn, so

41:44

people could reach out to

41:46

me from any number of

41:48

sources. Excellent, excellent. Do you have

41:50

a book recommendation? Do you

41:53

do audio books? I'm just

41:55

curious. Audio

41:57

books I have not done.

42:00

But I do read mainly blogs

42:02

and what I do like

42:04

go through most is what

42:06

are the latest innovations in

42:08

the data driven innovation space,

42:11

specifically the large customers,

42:13

how they are really

42:15

adopting the customers who I'm

42:17

not serving. So at least I

42:20

read about them and really find

42:22

out what exactly they

42:24

are doing. Because as I say

42:26

every day some new examples come

42:29

up. And it really is so

42:31

impressive to see that the

42:33

areas which have not been

42:35

which has not been thought

42:37

about are explored by

42:40

customers. And that's

42:42

what basically gives us the

42:44

gives you that, hey, this

42:46

is possible to be

42:49

implemented. And we always

42:51

call it as reimagine

42:53

your business, reimagine your

42:56

job in the new world. And

42:58

when you start reimagining it, and

43:00

I'll just give you a latest

43:02

today, I was talking to one of

43:04

the very senior business leader of

43:06

a large consumer company, and he

43:08

said, hey, I want to reimagine

43:11

my business without assuming that I

43:13

have anything. And if I want

43:15

to start a new CPG company in

43:17

the market, how will I

43:19

do using today's capabilities? And that's

43:22

how they are thinking. And that's

43:24

what really keeps you agile. Wow,

43:26

that's cool. I like that's a

43:28

great way to think about it, right?

43:30

Because the whole, we've always

43:33

done it this way, right? Like, it

43:35

does tend to seek up in, it

43:37

does sneak up on large organizations.

43:39

It seems to be a human

43:42

trait, but if you kind

43:44

of mentally make you keep your

43:46

mind agile, I think that's

43:48

awesome. No, that's really cool. I

43:50

really, really enjoyed this conversation and. If

43:53

you're willing to and schedule allows, I'd

43:55

love to have you and Andy kind

43:57

of do a deep dive on fabric

43:59

and... kind of what the implications

44:01

are, because fabric is something I

44:04

haven't really dealt much into. But

44:06

I feel like I'm missing something

44:08

by not doing that. No, I

44:10

think it's really nice talking to

44:13

your friend. Yeah, likewise. And I

44:15

will let our AI assistant Bailey

44:17

finish the show. And that's a

44:19

wrap on another episode of data

44:22

driven. A huge thank you to

44:24

San Jay Anadate for sharing his

44:26

insights on digital transformation. and Microsoft's

44:28

evolving role in the data-driven world.

44:31

From the power of co-pilot to

44:33

the rise of agentic AI, one

44:35

thing is clear data isn't just

44:37

the new oil, it's the entire

44:40

fuel supply of the modern enterprise.

44:42

And if your organization isn't leveraging

44:44

it properly, well, good luck keeping

44:46

up. If you enjoyed this episode,

44:49

be sure to subscribe, leave a

44:51

review, and share it with your

44:53

fellow data enthusiasts. And as always,

44:55

you can find more at data-driven

44:58

dot TV. Until next time stay

45:00

curious stay data driven and maybe

45:02

just maybe let a I handle

45:04

that boring spreadsheet for you