WEBVTT
00:00:00.000 --> 00:00:08.369
I often think about AI transformation is like the shift from paper maps to GPS for a long time inside large enterprises.
00:00:08.820 --> 00:00:10.619
Data worked like paper maps.
00:00:11.160 --> 00:00:18.030
We had them everywhere, like dashboards, reports, decks, all beautifully drawn and carefully maintained.
00:00:18.989 --> 00:00:23.234
But you still need an expert to read them, interpret them, and tell you which turn to take.
00:00:23.835 --> 00:00:28.785
And by the time you acted on that information, the situation might have already changed.
00:00:29.385 --> 00:00:34.725
Autopilot flies most of the plane, but no one would argue that the pilot is less important.
00:00:35.325 --> 00:00:42.045
In fact, the pilot's role becomes more critical because they're no longer focused on constant manual control.
00:00:42.554 --> 00:00:46.274
They're focused on judgment edge cases and responsibility.
00:00:46.875 --> 00:00:52.320
In my opinion, organizations that get this right, don't ask what AI will replace.
00:00:53.130 --> 00:00:56.875
They ask, where do we want human attention to be most valuable?
00:00:57.225 --> 00:01:05.444
AI is evolving so fast for expertise to come first, progress comes from trying, failing quickly and iterating.
00:01:18.248 --> 00:01:23.888
Most organizations say they're preparing for ai, but very few are changing how decisions get made.
00:01:24.519 --> 00:01:33.188
The gap between AI ambition and AI reality is widening, not shrinking, and the reason has far less to do with technology than most leaders think.
00:01:33.819 --> 00:01:42.608
Welcome to Data Analytics Chat, the podcast where senior data and AI leaders share the decisions and listen behind delivering real impact at scale.
00:01:43.209 --> 00:01:49.688
Today I'm joined by Sujit Nara Par, head of Data and Analytics at AWS Sales.
00:01:50.228 --> 00:02:04.724
Sujit has worked at the intersection of data leadership and large scale organizations, which makes him well placed to cut through the noise around AI and talk about what changes inside companies are making today.
00:02:05.323 --> 00:02:07.963
So in today's conversation, we're going to explore.
00:02:08.489 --> 00:02:20.519
Where leaders are underestimating what AI really changes, why many organizations stall despite strong technology and what needs to shift if AI is going to become part of everyday work.
00:02:20.878 --> 00:02:22.318
Sujit great to have you here.
00:02:22.919 --> 00:02:23.549
Hello, Ben.
00:02:23.549 --> 00:02:26.868
Thanks for having me and a very happy new year to you and the business.
00:02:27.468 --> 00:02:27.769
Yes.
00:02:27.769 --> 00:02:29.449
And happy New Year to all the listeners as well.
00:02:29.549 --> 00:02:32.098
So I'm looking forward to today's conversation.
00:02:32.218 --> 00:02:40.074
So I guess before we dive in, could you give the listeners a brief introduction to who you are and the work you're focused on
00:02:40.424 --> 00:02:40.993
sure.
00:02:41.503 --> 00:02:46.033
I was born and raised in India where I completed my bachelor's in industrial engineering.
00:02:46.033 --> 00:02:51.584
Before moving to the US I later went on to do my masters in operations research.
00:02:52.514 --> 00:02:59.774
I've spent most of my career at the intersection of data analytics and business station making ever since childhood.
00:03:00.149 --> 00:03:15.598
I love numbers, and that curiosity naturally pulled me towards problem solving, analytical thinking, and also growing up in India instilled a strong sense of competitiveness and drive, and that mindset stayed with me throughout my career.
00:03:16.199 --> 00:03:28.079
Even today, I'm someone who's deeply curious and cares a lot about details, and I like understanding our systems work under the hood and asking why things behave the way they do.
00:03:28.949 --> 00:03:35.998
Over time, my work has evolved from being deeply technical to focusing more on operational efficiency and business impact.
00:03:36.598 --> 00:03:46.199
Today I lead data products, analytics for global sales at AWS, and my focus is on making data and insights truly usable at scale.
00:03:46.769 --> 00:03:53.639
It's not just about building dashboards, but embedding analytics directly into workflows, so sales and marketing.
00:03:54.028 --> 00:03:57.449
Sales and marketing teams can make better decisions faster.
00:03:58.049 --> 00:04:11.848
At this stage, my role is less about doing the analysis myself and more about shaping how data flows through the organization and how teams use AI and agents, and how all of that translates into real measurable outcomes.
00:04:12.449 --> 00:04:18.254
So looking back then, what leadership decision most shaped how you approach your work today?
00:04:18.853 --> 00:04:19.783
Interesting question.
00:04:19.903 --> 00:04:31.024
One of the most important nation that shaped my leadership trajectory was moving from a purely technical role into a nation sciences role at Chase Business Banking.
00:04:31.894 --> 00:04:40.999
At that time, I was doing well as a big data engineer, but I wasn't happy with the fact that I couldn't clearly see the business impact of the work I was doing.
00:04:41.598 --> 00:04:43.428
So I went back into learning mode.
00:04:43.548 --> 00:04:55.968
I picked up data science and finance concepts, worked on some academic projects, and made myself uncomfortable by stepping into a role that was much closer to the Haitian making and the business itself.
00:04:56.569 --> 00:05:03.108
That role turned out to be quite fast-paced, and it fundamentally changed how I think about my work.
00:05:03.814 --> 00:05:26.733
The Lean org structure meant that I had direct exposure to C-suite leaders like the CFO Chief Strategy Officer, and I got to see firsthand how leaders running multi-billion dollar businesses think I learned how they approach problem solving, how they use data to make decisions, and most importantly, how they move forward without waiting for perfection.
00:05:27.334 --> 00:05:29.613
That experience taught me three critical things.
00:05:30.483 --> 00:05:44.944
How to deal with ambiguity under high pressure, how to balance speed with depth, and how to anchor my thinking in first principles and thus focusing on what truly drives outcomes rather than getting lost in the details.
00:05:45.543 --> 00:05:54.754
If I had made that move, I think I would've still stayed very strong technically, but I likely would've struggled later to connect data to outcomes.
00:05:55.324 --> 00:05:59.283
And influence agents or operate effectively at a leadership level.
00:05:59.884 --> 00:06:04.079
So that shift really shaped how I approach problems and leadership even today.
00:06:04.678 --> 00:06:05.069
Brilliant.
00:06:05.399 --> 00:06:14.834
So did you learn leadership through watching people or were you curious and asking like questions?
00:06:14.834 --> 00:06:15.644
How did you go about this?
00:06:15.673 --> 00:06:20.473
'cause obviously stepping up into leadership is a big challenge for a lot of people.
00:06:20.824 --> 00:06:26.473
For me it was more about a learning curve, and it's a journey on how I've learned about leadership.
00:06:26.803 --> 00:06:32.173
It's more about practice across the years on multiple things rather than one single thing.
00:06:33.163 --> 00:06:38.343
I've learned a quite a few things throughout the journey, and there are also some myths that got busted for me.
00:06:39.244 --> 00:06:45.764
See, one of the biggest surprises that for me was realizing that leadership is far more about clarity than depth.
00:06:46.723 --> 00:06:58.884
I learned pretty quickly that business intuition and the ability to connect dots across systems is something that I've seen most effective leaders have consistently displayed and acquired.
00:06:59.483 --> 00:07:02.413
Also, another important learning curve for me was around.
00:07:03.014 --> 00:07:17.324
genuinely believe that leadership meant having all the answers, and leaders were expected to know what to do in every situation, but that's not really true in hyper growth environments, especially leaders are often not the smartest person in the room.
00:07:17.894 --> 00:07:24.973
And a big part of the job is listening to the experts first, weighing the options, applying judgment, and then responding.
00:07:25.574 --> 00:07:34.543
That shift from feeling like you need to know everything to focusing on clarity and asking the right questions is something I've intentionally developed and truly practiced today.
00:07:35.384 --> 00:07:40.079
It's a big role in how I lead and it's contributed to my success so far.
00:07:40.439 --> 00:07:40.559
I.
00:07:41.473 --> 00:07:41.954
Brilliant.
00:07:42.584 --> 00:07:49.814
Then what surprised you most about leadership that I guess no one warned you about when you made the transition?
00:07:50.413 --> 00:07:54.473
Building on top of the business intuition part, that was the biggest surprise for me.
00:07:55.024 --> 00:08:01.004
It's probably the most underrated thing that we don't talk about business in most forums or in the leadership roles.
00:08:01.543 --> 00:08:13.449
Having that knack of understanding the business and connecting the various dots across systems and the ability be, and the ability to understand how various things connect to each other.
00:08:14.048 --> 00:08:20.168
And having that clarity and having that security in yourself as a leader to ask the questions.
00:08:20.559 --> 00:08:25.149
And when you don't know, it's okay to say, I don't know.
00:08:25.639 --> 00:08:29.178
And it's okay to say, Hey, I would love to learn more about this.
00:08:29.449 --> 00:08:39.078
And having that curious mindset is super important and something that really surprised me because I genuinely felt leaders are expected to know everything, which is definitely not the case.
00:08:39.678 --> 00:08:52.278
Yeah, no, I definitely think business domains, even now it's becoming more and more important with excelling in your role because obviously you mentioned on the podcast previously previous episodes, that's.
00:08:52.568 --> 00:08:55.688
You've got a lot of the technologies now do the heavy lifting for people.
00:08:55.688 --> 00:08:59.078
It's now, how can you apply this into the business world?
00:08:59.078 --> 00:09:01.144
How can you get the growth, cost savings?
00:09:01.354 --> 00:09:05.283
How can you improve the efficiencies which are gonna add value to your business?
00:09:05.884 --> 00:09:06.083
Absolutely.
00:09:06.384 --> 00:09:10.724
It's all about normalizing the imperfections because nobody's perfect and it's a learning curve.
00:09:11.024 --> 00:09:12.913
We just have to accept that and embrace it.
00:09:13.514 --> 00:09:13.783
Yep.
00:09:13.788 --> 00:09:16.724
And it's the goalposts always keep moving.
00:09:17.323 --> 00:09:17.683
Yep.
00:09:18.283 --> 00:09:18.583
Brilliant.
00:09:18.614 --> 00:09:20.984
So let's move on to the data topic.
00:09:21.134 --> 00:09:25.693
And obviously there's a lot happening, a lot of businesses are shifting with ai.
00:09:26.323 --> 00:09:33.524
So when you look at the next few years, what do you see AI changing in organizations?
00:09:33.524 --> 00:09:34.033
First?
00:09:34.634 --> 00:09:43.303
See, I often think about AI transformation is like the shift from paper maps to GPS for a long time inside large enterprises.
00:09:43.754 --> 00:09:45.553
Data worked like paper maps.
00:09:46.094 --> 00:09:52.964
We had them everywhere, like dashboards, reports, decks, all beautifully drawn and carefully maintained.
00:09:53.923 --> 00:09:58.168
But you still need an expert to read them, interpret them, and tell you which turn to take.
00:09:58.769 --> 00:10:03.719
And by the time you acted on that information, the situation might have already changed.
00:10:04.318 --> 00:10:08.009
AI is really the GPS movement over the next few years.
00:10:08.308 --> 00:10:12.119
The first thing AI will change is in the strategy or art charts.
00:10:12.839 --> 00:10:14.969
It's the experience of using data.
00:10:15.568 --> 00:10:22.048
See, instead of stopping your work to go back to look at the data, insights will show up while you're driving.
00:10:22.769 --> 00:10:26.249
It'll be embedded directly into your workflows, tools, and decisions.
00:10:26.849 --> 00:10:36.239
Just like GPS, it doesn't eliminate maps, it doesn't remove human judgment, but it'll dramatically reduce the friction between knowing and acting.
00:10:36.839 --> 00:10:41.548
The interesting part here is what does the what does it does to the organizations?
00:10:42.149 --> 00:10:46.948
You no longer need to be a data expert to make sense of massive amounts of data.
00:10:47.879 --> 00:10:52.469
Product managers, operators, frontline leaders, can all ask questions in plain, simple English.
00:10:53.068 --> 00:10:55.168
Get usable answers instantly.
00:10:55.769 --> 00:10:58.229
That shifts the role of the data organization.
00:10:59.188 --> 00:11:11.428
Instead of being the team that hands out directions, it becomes the team responsible for making sure the navigation system itself is trustworthy by putting the right data, right guardrails in place.
00:11:12.028 --> 00:11:20.458
And here's the deeper change that most people underestimate when everyone has GPS access to data is no longer the constraint.
00:11:21.058 --> 00:11:23.489
Judgment on top of the data becomes the constraint.
00:11:24.089 --> 00:11:33.089
So when I look ahead, I think AI will change organizations first by collapsing the distance between insight and action, not just by replacing people.
00:11:33.688 --> 00:11:34.109
Okay.
00:11:34.379 --> 00:11:35.219
Fascinating.
00:11:35.969 --> 00:11:42.418
So do you feel there are any changes that leaders are still underestimating?
00:11:43.019 --> 00:11:46.158
It's hard to generalize but at a macro.
00:11:46.188 --> 00:11:51.948
But what, but at a macro level, I think there is both underestimation and overestimation happening at the same time.
00:11:52.879 --> 00:11:56.149
We only need to work towards finding the right balance in the near future.
00:11:57.139 --> 00:12:07.823
To extend the G Ps analogy, many leaders think AI as just getting a faster route or better traffic alerts that is helpful, efficient, and nice to have.
00:12:08.423 --> 00:12:17.333
What they're underestimating is that once everyone has access to AI, leadership is no longer about knowing the answer.
00:12:17.844 --> 00:12:22.354
It's about understanding what brings the answer, the context behind it.
00:12:23.313 --> 00:12:28.974
See, in the past, senior leaders had an advantage by having the best information.
00:12:29.274 --> 00:12:35.783
They could see the patterns that other others couldn't, and they had answers to the questions faster than the rest of the organizations.
00:12:36.639 --> 00:12:44.259
AI will quietly erode that a vantage When anyone in the organization can ask the questions like, what's driving this?
00:12:44.558 --> 00:12:45.219
What's hap?
00:12:45.278 --> 00:12:52.089
What happens if you change X and all of this is available at your tip of your fingertips?
00:12:52.839 --> 00:12:59.678
Then leadership is less about information access and it's more about direction, judgment and trade-offs.
00:13:00.278 --> 00:13:09.134
So if people can't tell the difference between the insights and the hallucination here, then they will stop using the system.
00:13:09.734 --> 00:13:10.094
Okay.
00:13:10.394 --> 00:13:10.783
Brilliant.
00:13:11.293 --> 00:13:19.394
And then, so there, obviously there's a lot of talk around AI complementing human roles in practical terms.
00:13:20.024 --> 00:13:21.413
So what does that look like?
00:13:22.014 --> 00:13:31.828
So I think AI will complement humans, but it's if I have to give an example, it'll complement the same the way autopilot technology compliments a pilot.
00:13:32.428 --> 00:13:34.494
The technology didn't replace pilots.
00:13:35.214 --> 00:13:40.913
It changed the way they spend their attention on, as we all know, on a modern flight.
00:13:41.484 --> 00:13:46.823
Autopilot flies most of the plane, but no one would argue that the pilot is less important.
00:13:47.423 --> 00:13:54.144
In fact, the pilot's role becomes more critical because they're no longer focused on constant manual control.
00:13:54.653 --> 00:13:58.374
They're focused on judgment edge cases and responsibility.
00:13:58.974 --> 00:14:02.094
That's what AI complementing human roles will look like.
00:14:02.094 --> 00:14:09.474
Inside enterprise organizations, AI can take over continuous, repetitive cogni cognitive tasks.
00:14:10.134 --> 00:14:19.043
Pulling the data, summarizing patterns, testing scenarios, or even answering the basic questions like what happened and what if scenarios.
00:14:19.644 --> 00:14:22.793
And that frees humans to focus on what AI is bad at.
00:14:23.244 --> 00:14:31.524
Like framing the right questions, understanding the context, weighing, tradeoffs, and deciding when not to act in a data organization.
00:14:32.453 --> 00:14:35.063
Analysts will spend less time building analysis.
00:14:35.543 --> 00:14:38.964
More time challenging them in product teams.
00:14:39.293 --> 00:14:44.423
Managers will spend less time chasing metrics and more time debating priorities.
00:14:45.024 --> 00:14:50.903
In leadership, executives will spend less time asking for updates and more time deciding direction.
00:14:51.504 --> 00:14:59.033
The key practical shift is this becomes the first pass, not the final say, just like autopilot.
00:14:59.124 --> 00:15:01.854
It works extremely well in normal conditions.
00:15:02.798 --> 00:15:11.828
But when something unexpected happens, like a pandemic, a policy change, or a cultural nuance, that's when human judgment takes control.
00:15:12.428 --> 00:15:17.109
And one more important thing here, complimenting doesn't mean splitting work 50 50.
00:15:17.979 --> 00:15:24.249
It means AI handles the known parts and humans handle the moments where map no longer applies.
00:15:24.849 --> 00:15:30.293
In my opinion, organizations that get this right, don't ask what AI will replace.
00:15:31.104 --> 00:15:34.849
They ask, where do we want human attention to be most valuable?
00:15:35.448 --> 00:15:37.489
Yeah, you make some interesting points.
00:15:37.489 --> 00:15:43.229
It's, similar to Excel, their stories were, they're gonna take all the accountancy jobs.
00:15:43.379 --> 00:15:44.609
Accountants are still here.
00:15:45.389 --> 00:15:48.958
Also LinkedIn, there's always, 20 years ago, they're gonna take all the recruitment jobs.
00:15:49.558 --> 00:15:50.249
I'm still here.
00:15:50.788 --> 00:15:57.239
So it is, like I said, obviously parts of it will be replaced, but you still need the human element, don't you?
00:15:57.239 --> 00:16:06.953
It is, it's just, I think it's just a lot noise on the market in my opinion, whether people are trying to sell their products or scare people.
00:16:06.953 --> 00:16:10.283
It's just humans are still gonna be here like it is.
00:16:10.464 --> 00:16:12.293
We're not getting replaced in my opinion.
00:16:12.293 --> 00:16:16.734
It's just you're going to get some of your roles gonna slightly adapt.
00:16:17.333 --> 00:16:24.719
So then what's the biggest challenges companies facing right now as they try to integrate AI into everyday work?
00:16:25.318 --> 00:16:27.808
I love giving analogy, so let me give one more on this.
00:16:28.609 --> 00:16:29.318
Let's go for it.
00:16:30.119 --> 00:16:34.769
so implementing AI reminds me of setting up a professional kitchen, right?
00:16:35.399 --> 00:16:48.859
So you can't create anyone who's worked in the food industry knows that you can't create great dishes or an excellent culinary experience unless you know your pantry very well, and it's organized and you understand how ingredients work together.
00:16:49.458 --> 00:16:57.558
That's exactly what I'm seeing with AI adoption today, companies are rushing to deploy AI without building to essential foundations.
00:16:58.188 --> 00:17:06.949
First, they're working with messy, inconsistent data, like just trying to cook when you don't even know what's in your ingredients.
00:17:07.788 --> 00:17:14.263
At AWS, when we started our AI journey, we invested heavily in standardizing data definitions.
00:17:14.864 --> 00:17:21.493
And ensuring clean, consistent information across the organization, but clean data alone isn't enough.
00:17:22.183 --> 00:17:27.193
You also need to understand how pieces of business information relate to each other.
00:17:28.153 --> 00:17:29.804
We call this as a semantic layer.
00:17:30.403 --> 00:17:33.374
It helps a understand business context.
00:17:34.273 --> 00:17:40.453
For example, how marketing influences sales pipelines, or how customer engagement drives revenue.
00:17:41.054 --> 00:17:42.403
And without this layer.
00:17:42.808 --> 00:17:48.269
AI might give technically correct answers, but contextually it could be meaningless.
00:17:48.868 --> 00:17:57.388
So by investing in these foundations, we are basically enabling AI to deliver insights that leaders can trust and act on.
00:17:57.989 --> 00:18:07.469
Yes, this totally requires an upfront investment, but it's a difference between AI that adds value versus the AI that creates noise.
00:18:08.068 --> 00:18:14.128
So in my opinion, the companies that get this right will have a real advantage.
00:18:14.729 --> 00:18:16.064
Yeah, no, I agree.
00:18:16.094 --> 00:18:22.653
And then so what separates organizations that have got the ability to move quickly from those that get stuck?
00:18:23.253 --> 00:18:28.233
See, organizations that move fast will all tell very similar stories.
00:18:28.818 --> 00:18:31.729
They don't start by going in all in on ai.
00:18:32.419 --> 00:18:39.709
They start with one real workflow owned by a real leader where nation get made or real nations get made.
00:18:40.548 --> 00:18:44.689
They focus on operationalizing AI inside an actual business workflow.
00:18:45.288 --> 00:18:50.058
The invest heavy on foundational data and infrastructure.
00:18:50.659 --> 00:18:52.608
It's not just about optimized data.
00:18:53.163 --> 00:19:04.354
It's also about building a data infrastructure that can be trusted, connected, and available at all times, and that one decision alone removes months and months of friction later.
00:19:04.953 --> 00:19:10.564
Another important thing is also here, another important thing here is how they move in small loops.
00:19:11.163 --> 00:19:17.193
They ship something that's good enough, watch how people use it and improve it in weeks, not quarters.
00:19:17.794 --> 00:19:21.243
improves because it's in use, not because it's in theory.
00:19:21.844 --> 00:19:28.023
Organizations that move faster, put light guard rails in place instead of heavy approvals.
00:19:28.173 --> 00:19:35.314
The teams can move faster without fear, and most importantly, leaders use AI themselves.
00:19:36.064 --> 00:19:41.193
When employees see AI shaping realizations at the top, adoptions become cultural.
00:19:41.794 --> 00:19:45.304
On the flip side, organizations that struggle do the opposite.
00:19:45.903 --> 00:19:47.824
They debate strategy endlessly.
00:19:48.153 --> 00:19:50.703
Chase for perfect data that never exists.
00:19:51.483 --> 00:19:56.493
Avoid experimentation and wait for confidence instead of building it through use.
00:19:57.094 --> 00:20:00.903
The difference here isn't ambition or spend.
00:20:01.564 --> 00:20:05.433
It's operation clarity and a willingness to learn in motion.
00:20:06.034 --> 00:20:15.048
And I'm guessing now as a leader, cause there's, must be so much want from especially like AI agents, from business divisions, I guess now.
00:20:15.858 --> 00:20:20.808
A big challenge is gonna be choosing the correct use case to implement.
00:20:20.929 --> 00:20:26.808
'cause as you mentioned, obviously you can't do multiple projects at once or your business is just gonna get unstuck.
00:20:26.898 --> 00:20:30.919
So I guess a big challenge now for leaders is choosing what's gonna have the biggest impact.
00:20:31.519 --> 00:20:32.358
Absolutely, yes.
00:20:32.419 --> 00:20:39.219
It's about understanding where there is value and picking the right use case, not just because it's a shiny object.
00:20:39.219 --> 00:20:41.108
We wanna pick it up and then start using it.
00:20:41.108 --> 00:20:50.108
Meaninglessly, it's about putting it to use where it really matters, and also understand that it's not just about trying to.
00:20:50.709 --> 00:20:52.449
Make the horses run faster.
00:20:52.628 --> 00:20:55.588
In many cases, you may not need the whole workflow itself.
00:20:55.919 --> 00:21:00.778
It could be about finding a combustion engine instead of trying to find faster horses.
00:21:01.769 --> 00:21:08.368
So there's a, there should be a lot of value on time spent in understanding where exactly this fits
00:21:08.669 --> 00:21:08.959
okay.
00:21:09.523 --> 00:21:19.963
Then as AI takes on more tasks, what new skills or mindsets will employees need to stay relevant and valuable with?
00:21:20.564 --> 00:21:31.324
See the, there's a long list, but for me, the most important skill is in technical mastery, but it's mindset First, it's an eagerness to learn and try what's next.
00:21:32.134 --> 00:21:36.663
The people who stay relevant won't wait for perfect clarity or formal training.
00:21:37.068 --> 00:21:39.798
They will experiment early and learn my doing.
00:21:40.398 --> 00:21:46.169
Second is about being comfortable at being bad at something.
00:21:46.169 --> 00:21:46.798
While learning.
00:21:47.398 --> 00:21:55.618
AI is evolving so fast for expertise to come first, progress comes from trying, failing quickly and iterating.
00:21:56.219 --> 00:22:01.979
Most importantly, instead of worrying about what might replace us, it's far more productive to ask.
00:22:02.564 --> 00:22:05.443
What does this free up to do better?
00:22:05.864 --> 00:22:09.044
Where can I add judgment, creativity, and context?
00:22:09.644 --> 00:22:12.284
That's where creativity becomes a real differentiator.
00:22:12.884 --> 00:22:28.213
It's not just about using AI to be faster, but to think differently and create a new value by experimenting and understanding where we could create new business workflows that we couldn't think in the past.
00:22:28.814 --> 00:22:29.324
Brilliant.
00:22:29.413 --> 00:22:37.269
Then how should leaders rethink the way teams are structured when humans and AI systems start working side by side?
00:22:37.868 --> 00:22:47.138
As AI and humans work together, team structures will shift away from rigid roles and more towards orchestration.
00:22:47.739 --> 00:22:50.648
What works well are smaller and autonomous teams.
00:22:51.249 --> 00:23:00.548
Just like at Amazon, we do our two piecer teams that will own a customer problem end to end and irate quickly with AI embedded in the workflow.
00:23:01.148 --> 00:23:04.929
Also, another interesting shift is how engineering roles are evolving.
00:23:05.739 --> 00:23:13.538
The line between product and engineering is starting to blur and it's leading to the rise of a new role called product engineers.
00:23:14.138 --> 00:23:24.909
This role basically stays close to the customer problem and decides how best to solve it, whether through code configuration or AI orchestration in general.
00:23:25.259 --> 00:23:33.959
Going a little deeper on the engineering side, engineering becomes less about writing code and more about designing systems.
00:23:34.558 --> 00:23:45.959
It's going to be about deciding which tools to use, how agents interact, where guardrails belong, and where human judgment matters that I think enables speed.
00:23:45.959 --> 00:23:59.338
Without Kio, we are gonna see fewer handoffs, tighter feedback loops, and faster translation from intent to execution, not just because teams are bigger, but because they're sharper and better aligned.
00:23:59.939 --> 00:24:00.179
Okay.
00:24:00.179 --> 00:24:05.118
It's interesting you, as you mentioned around pro project sorry, product engineering.
00:24:05.749 --> 00:24:10.983
So is there any other early examples of team structures that you've seen work well?
00:24:11.584 --> 00:24:16.983
There's definitely a lot of experimentation happening and different models work better in different contexts.
00:24:17.284 --> 00:24:31.253
Continuing on the product engineer role in, at our organization, we've introduced this role last year and the idea was to bring product thinking and engineering execution close together, especially as AI became part of everyday work.
00:24:31.854 --> 00:24:34.838
What we are seeing so far has been truly encouraging.
00:24:35.513 --> 00:24:43.523
Particularly around rapid prototyping and iation teams are able to go from an idea to something tangible much faster.
00:24:44.003 --> 00:24:48.683
This and the same people are able to shape the problem and work directly with the tools.
00:24:49.284 --> 00:24:54.588
That said, it's still early days and we are very much in a learning phase and super excited about it.
00:24:55.189 --> 00:24:58.249
Yeah, and I think obviously as these new team structures are.
00:24:58.848 --> 00:24:59.479
Blended.
00:24:59.659 --> 00:25:03.919
The challenge is gonna be actually hiring people for these roles.
00:25:03.949 --> 00:25:21.138
'cause even taking sort of the, I get the product role, you mentioned this, obviously now you've got product data scientists getting the right combination of technical skills, product skills, and also business domain expertise is a massive hurdle for cha businesses that wanna progress quickly.
00:25:21.709 --> 00:25:23.568
So I guess that's gonna impact.
00:25:24.423 --> 00:25:32.153
Other sort of new business models or new team structures as like the AI that's within the co within companies.
00:25:32.753 --> 00:25:33.534
Absolutely, yes.
00:25:33.534 --> 00:25:38.784
It's going to see, it's about a paradigm shift in how these roles are shaping together.
00:25:39.084 --> 00:25:43.453
Like you said, it's an intersection between a product and a science and an engineering role.
00:25:44.084 --> 00:26:02.453
That's giving birth to these new roles, either the product engineer, which is at an intersection of an engineer and a product manager or a product data scientist between a science roles and a product manager, or even a super engineer who was at the intersection of a data software and a business intelligence engineers.
00:26:03.054 --> 00:26:12.614
So we are not going to find ready-made skillful professionals with all acquired skills in the industry, but we are gonna work towards it.
00:26:13.003 --> 00:26:21.104
And that's where I think as a leader, love to have the patience to make sure the culture is set for that experimentation.
00:26:21.104 --> 00:26:27.913
People will fail more often than they've seen in the past while experimenting and while acquiring new skills.
00:26:28.453 --> 00:26:37.273
And it's totally fine to make sure as lo as long as we are going in the right direction, as long as there is learning.
00:26:38.219 --> 00:26:41.118
Faster than the pace of the problems that we are facing.
00:26:41.239 --> 00:26:47.648
It's absolutely fine to basically keep experimenting, learning on the go and see how it goes.
00:26:48.249 --> 00:26:48.459
Yeah.
00:26:48.459 --> 00:26:55.544
And I think in the future sort of learning and development like teams, that's gonna be where businesses succeed.
00:26:55.943 --> 00:27:02.663
If you're investing in your team to learn these new skills, they're gonna get up to scratch a lot quicker than just learning on job.
00:27:02.784 --> 00:27:04.969
So I think it's gonna be interesting how businesses do this.
00:27:04.999 --> 00:27:08.058
'cause even now, like obviously I work in recruitment, we're seeing.
00:27:08.114 --> 00:27:10.124
Businesses struggle to hire people.
00:27:10.183 --> 00:27:15.763
'cause a lot of them are just looking for all the skill sets when really you're not gonna get that.
00:27:15.854 --> 00:27:29.534
'cause the industry is so diverse now, so many different roles, so many different positions that where data roles previously were more technical, they now have more business stakeholder, product focus.
00:27:29.534 --> 00:27:33.034
So there's so much out there now and it, and I guess to make it.
00:27:33.864 --> 00:27:34.614
Even more challenging.
00:27:34.644 --> 00:27:37.314
Everyone is trying to become more data driven.
00:27:37.314 --> 00:27:42.743
So the competition, everyone wants these skill sets, so it is challenging for businesses to keep up.
00:27:43.344 --> 00:27:47.213
So true the, it's very interesting to see how the business teams.
00:27:48.159 --> 00:27:52.568
Trying to become more and more technical, and AI is totally helping them to get there.
00:27:52.868 --> 00:28:00.818
And how the technical teams, like engineers and scientists want to become more and more business savvy.
00:28:01.239 --> 00:28:03.009
And AI is helping that too.
00:28:03.729 --> 00:28:05.199
So it's a very interesting phase.
00:28:05.199 --> 00:28:14.108
We are gonna see everyone trying to pick up and acquire something which they were not exposed to or something they, something that was not the skill in the past.
00:28:14.709 --> 00:28:24.798
And that's why I truly think it's going to make everyone a lot more productive and skillful in the next few years with the availability of all the tools in front of us.
00:28:25.788 --> 00:28:26.148
Yep.
00:28:26.209 --> 00:28:26.868
Definitely.
00:28:27.598 --> 00:28:28.679
Moving to the next question.
00:28:28.769 --> 00:28:37.469
When businesses ask how do we prepare for the future of ai, what's the smartest first step they can take right now, in your opinion?
00:28:37.834 --> 00:28:43.864
I usually start by saying, embrace AI in a way that genuinely works your advantage.
00:28:44.673 --> 00:28:51.523
Be clear about what, it creates real value, not just where it's exciting, and sometime that starts with data.
00:28:51.884 --> 00:28:55.874
A strong data foundation matters, especially as you introduce agents.
00:28:56.443 --> 00:29:04.544
Agents will need clear definitions, clear inputs and shared context, the judgment, what a human being could apply.
00:29:04.723 --> 00:29:11.314
When there is ambiguity and when there is multiple definitions agents cannot really apply that judgment.
00:29:12.064 --> 00:29:18.364
So it's important that we have very clear definitions and we call it as metrics store at Amazon.
00:29:18.963 --> 00:29:21.439
And without that, agents will.
00:29:22.263 --> 00:29:28.233
Basically give you inconsistent outputs and there could be hallucinations and that leads to loss of trust.
00:29:28.834 --> 00:29:31.324
And second one, it's also, again, a mindset shift here.
00:29:31.923 --> 00:29:34.443
AI isn't about doing the same work faster.
00:29:34.894 --> 00:29:38.284
That's the faster us faster horse strap I was talking about.
00:29:38.973 --> 00:29:45.273
The real opportunity is rethinking workflows entirely, and that's where the step change impact comes from.
00:29:45.874 --> 00:29:46.564
See, we don't need to.
00:29:47.163 --> 00:29:48.544
out everything from the get go.
00:29:48.844 --> 00:30:02.374
But if we start with clear values and understand what exactly we are trying to do, and with some solid data foundations and a willingness to rethink workflows, we are setting ourself for success.
00:30:02.973 --> 00:30:05.824
Ji, I really appreciate your Thank you for conversation.
00:30:06.423 --> 00:30:09.318
It was awesome being here and thoroughly enjoyed the conversation too.
00:30:10.074 --> 00:30:11.003
Thanks for having me again.
00:30:11.979 --> 00:30:12.338
No problem.
00:30:12.338 --> 00:30:25.598
And I wish you all the best for 2026 and also for thanks for listening, everyone, and I wish you all, you hit your goals for 2026 and I'll see you in the next episode of Data Analytics Chat.
00:30:25.898 --> 00:30:26.259
Thank you.
00:00:00.000 --> 00:00:08.369
I often think about AI transformation is like the shift from paper maps to GPS for a long time inside large enterprises.
00:00:08.820 --> 00:00:10.619
Data worked like paper maps.
00:00:11.160 --> 00:00:18.030
We had them everywhere, like dashboards, reports, decks, all beautifully drawn and carefully maintained.
00:00:18.989 --> 00:00:23.234
But you still need an expert to read them, interpret them, and tell you which turn to take.
00:00:23.835 --> 00:00:28.785
And by the time you acted on that information, the situation might have already changed.
00:00:29.385 --> 00:00:34.725
Autopilot flies most of the plane, but no one would argue that the pilot is less important.
00:00:35.325 --> 00:00:42.045
In fact, the pilot's role becomes more critical because they're no longer focused on constant manual control.
00:00:42.554 --> 00:00:46.274
They're focused on judgment edge cases and responsibility.
00:00:46.875 --> 00:00:52.320
In my opinion, organizations that get this right, don't ask what AI will replace.
00:00:53.130 --> 00:00:56.875
They ask, where do we want human attention to be most valuable?
00:00:57.225 --> 00:01:05.444
AI is evolving so fast for expertise to come first, progress comes from trying, failing quickly and iterating.
00:01:18.248 --> 00:01:23.888
Most organizations say they're preparing for ai, but very few are changing how decisions get made.
00:01:24.519 --> 00:01:33.188
The gap between AI ambition and AI reality is widening, not shrinking, and the reason has far less to do with technology than most leaders think.
00:01:33.819 --> 00:01:42.608
Welcome to Data Analytics Chat, the podcast where senior data and AI leaders share the decisions and listen behind delivering real impact at scale.
00:01:43.209 --> 00:01:49.688
Today I'm joined by Sujit Nara Par, head of Data and Analytics at AWS Sales.
00:01:50.228 --> 00:02:04.724
Sujit has worked at the intersection of data leadership and large scale organizations, which makes him well placed to cut through the noise around AI and talk about what changes inside companies are making today.
00:02:05.323 --> 00:02:07.963
So in today's conversation, we're going to explore.
00:02:08.489 --> 00:02:20.519
Where leaders are underestimating what AI really changes, why many organizations stall despite strong technology and what needs to shift if AI is going to become part of everyday work.
00:02:20.878 --> 00:02:22.318
Sujit great to have you here.
00:02:22.919 --> 00:02:23.549
Hello, Ben.
00:02:23.549 --> 00:02:26.868
Thanks for having me and a very happy new year to you and the business.
00:02:27.468 --> 00:02:27.769
Yes.
00:02:27.769 --> 00:02:29.449
And happy New Year to all the listeners as well.
00:02:29.549 --> 00:02:32.098
So I'm looking forward to today's conversation.
00:02:32.218 --> 00:02:40.074
So I guess before we dive in, could you give the listeners a brief introduction to who you are and the work you're focused on
00:02:40.424 --> 00:02:40.993
sure.
00:02:41.503 --> 00:02:46.033
I was born and raised in India where I completed my bachelor's in industrial engineering.
00:02:46.033 --> 00:02:51.584
Before moving to the US I later went on to do my masters in operations research.
00:02:52.514 --> 00:02:59.774
I've spent most of my career at the intersection of data analytics and business station making ever since childhood.
00:03:00.149 --> 00:03:15.598
I love numbers, and that curiosity naturally pulled me towards problem solving, analytical thinking, and also growing up in India instilled a strong sense of competitiveness and drive, and that mindset stayed with me throughout my career.
00:03:16.199 --> 00:03:28.079
Even today, I'm someone who's deeply curious and cares a lot about details, and I like understanding our systems work under the hood and asking why things behave the way they do.
00:03:28.949 --> 00:03:35.998
Over time, my work has evolved from being deeply technical to focusing more on operational efficiency and business impact.
00:03:36.598 --> 00:03:46.199
Today I lead data products, analytics for global sales at AWS, and my focus is on making data and insights truly usable at scale.
00:03:46.769 --> 00:03:53.639
It's not just about building dashboards, but embedding analytics directly into workflows, so sales and marketing.
00:03:54.028 --> 00:03:57.449
Sales and marketing teams can make better decisions faster.
00:03:58.049 --> 00:04:11.848
At this stage, my role is less about doing the analysis myself and more about shaping how data flows through the organization and how teams use AI and agents, and how all of that translates into real measurable outcomes.
00:04:12.449 --> 00:04:18.254
So looking back then, what leadership decision most shaped how you approach your work today?
00:04:18.853 --> 00:04:19.783
Interesting question.
00:04:19.903 --> 00:04:31.024
One of the most important nation that shaped my leadership trajectory was moving from a purely technical role into a nation sciences role at Chase Business Banking.
00:04:31.894 --> 00:04:40.999
At that time, I was doing well as a big data engineer, but I wasn't happy with the fact that I couldn't clearly see the business impact of the work I was doing.
00:04:41.598 --> 00:04:43.428
So I went back into learning mode.
00:04:43.548 --> 00:04:55.968
I picked up data science and finance concepts, worked on some academic projects, and made myself uncomfortable by stepping into a role that was much closer to the Haitian making and the business itself.
00:04:56.569 --> 00:05:03.108
That role turned out to be quite fast-paced, and it fundamentally changed how I think about my work.
00:05:03.814 --> 00:05:26.733
The Lean org structure meant that I had direct exposure to C-suite leaders like the CFO Chief Strategy Officer, and I got to see firsthand how leaders running multi-billion dollar businesses think I learned how they approach problem solving, how they use data to make decisions, and most importantly, how they move forward without waiting for perfection.
00:05:27.334 --> 00:05:29.613
That experience taught me three critical things.
00:05:30.483 --> 00:05:44.944
How to deal with ambiguity under high pressure, how to balance speed with depth, and how to anchor my thinking in first principles and thus focusing on what truly drives outcomes rather than getting lost in the details.
00:05:45.543 --> 00:05:54.754
If I had made that move, I think I would've still stayed very strong technically, but I likely would've struggled later to connect data to outcomes.
00:05:55.324 --> 00:05:59.283
And influence agents or operate effectively at a leadership level.
00:05:59.884 --> 00:06:04.079
So that shift really shaped how I approach problems and leadership even today.
00:06:04.678 --> 00:06:05.069
Brilliant.
00:06:05.399 --> 00:06:14.834
So did you learn leadership through watching people or were you curious and asking like questions?
00:06:14.834 --> 00:06:15.644
How did you go about this?
00:06:15.673 --> 00:06:20.473
'cause obviously stepping up into leadership is a big challenge for a lot of people.
00:06:20.824 --> 00:06:26.473
For me it was more about a learning curve, and it's a journey on how I've learned about leadership.
00:06:26.803 --> 00:06:32.173
It's more about practice across the years on multiple things rather than one single thing.
00:06:33.163 --> 00:06:38.343
I've learned a quite a few things throughout the journey, and there are also some myths that got busted for me.
00:06:39.244 --> 00:06:45.764
See, one of the biggest surprises that for me was realizing that leadership is far more about clarity than depth.
00:06:46.723 --> 00:06:58.884
I learned pretty quickly that business intuition and the ability to connect dots across systems is something that I've seen most effective leaders have consistently displayed and acquired.
00:06:59.483 --> 00:07:02.413
Also, another important learning curve for me was around.
00:07:03.014 --> 00:07:17.324
genuinely believe that leadership meant having all the answers, and leaders were expected to know what to do in every situation, but that's not really true in hyper growth environments, especially leaders are often not the smartest person in the room.
00:07:17.894 --> 00:07:24.973
And a big part of the job is listening to the experts first, weighing the options, applying judgment, and then responding.
00:07:25.574 --> 00:07:34.543
That shift from feeling like you need to know everything to focusing on clarity and asking the right questions is something I've intentionally developed and truly practiced today.
00:07:35.384 --> 00:07:40.079
It's a big role in how I lead and it's contributed to my success so far.
00:07:40.439 --> 00:07:40.559
I.
00:07:41.473 --> 00:07:41.954
Brilliant.
00:07:42.584 --> 00:07:49.814
Then what surprised you most about leadership that I guess no one warned you about when you made the transition?
00:07:50.413 --> 00:07:54.473
Building on top of the business intuition part, that was the biggest surprise for me.
00:07:55.024 --> 00:08:01.004
It's probably the most underrated thing that we don't talk about business in most forums or in the leadership roles.
00:08:01.543 --> 00:08:13.449
Having that knack of understanding the business and connecting the various dots across systems and the ability be, and the ability to understand how various things connect to each other.
00:08:14.048 --> 00:08:20.168
And having that clarity and having that security in yourself as a leader to ask the questions.
00:08:20.559 --> 00:08:25.149
And when you don't know, it's okay to say, I don't know.
00:08:25.639 --> 00:08:29.178
And it's okay to say, Hey, I would love to learn more about this.
00:08:29.449 --> 00:08:39.078
And having that curious mindset is super important and something that really surprised me because I genuinely felt leaders are expected to know everything, which is definitely not the case.
00:08:39.678 --> 00:08:52.278
Yeah, no, I definitely think business domains, even now it's becoming more and more important with excelling in your role because obviously you mentioned on the podcast previously previous episodes, that's.
00:08:52.568 --> 00:08:55.688
You've got a lot of the technologies now do the heavy lifting for people.
00:08:55.688 --> 00:08:59.078
It's now, how can you apply this into the business world?
00:08:59.078 --> 00:09:01.144
How can you get the growth, cost savings?
00:09:01.354 --> 00:09:05.283
How can you improve the efficiencies which are gonna add value to your business?
00:09:05.884 --> 00:09:06.083
Absolutely.
00:09:06.384 --> 00:09:10.724
It's all about normalizing the imperfections because nobody's perfect and it's a learning curve.
00:09:11.024 --> 00:09:12.913
We just have to accept that and embrace it.
00:09:13.514 --> 00:09:13.783
Yep.
00:09:13.788 --> 00:09:16.724
And it's the goalposts always keep moving.
00:09:17.323 --> 00:09:17.683
Yep.
00:09:18.283 --> 00:09:18.583
Brilliant.
00:09:18.614 --> 00:09:20.984
So let's move on to the data topic.
00:09:21.134 --> 00:09:25.693
And obviously there's a lot happening, a lot of businesses are shifting with ai.
00:09:26.323 --> 00:09:33.524
So when you look at the next few years, what do you see AI changing in organizations?
00:09:33.524 --> 00:09:34.033
First?
00:09:34.634 --> 00:09:43.303
See, I often think about AI transformation is like the shift from paper maps to GPS for a long time inside large enterprises.
00:09:43.754 --> 00:09:45.553
Data worked like paper maps.
00:09:46.094 --> 00:09:52.964
We had them everywhere, like dashboards, reports, decks, all beautifully drawn and carefully maintained.
00:09:53.923 --> 00:09:58.168
But you still need an expert to read them, interpret them, and tell you which turn to take.
00:09:58.769 --> 00:10:03.719
And by the time you acted on that information, the situation might have already changed.
00:10:04.318 --> 00:10:08.009
AI is really the GPS movement over the next few years.
00:10:08.308 --> 00:10:12.119
The first thing AI will change is in the strategy or art charts.
00:10:12.839 --> 00:10:14.969
It's the experience of using data.
00:10:15.568 --> 00:10:22.048
See, instead of stopping your work to go back to look at the data, insights will show up while you're driving.
00:10:22.769 --> 00:10:26.249
It'll be embedded directly into your workflows, tools, and decisions.
00:10:26.849 --> 00:10:36.239
Just like GPS, it doesn't eliminate maps, it doesn't remove human judgment, but it'll dramatically reduce the friction between knowing and acting.
00:10:36.839 --> 00:10:41.548
The interesting part here is what does the what does it does to the organizations?
00:10:42.149 --> 00:10:46.948
You no longer need to be a data expert to make sense of massive amounts of data.
00:10:47.879 --> 00:10:52.469
Product managers, operators, frontline leaders, can all ask questions in plain, simple English.
00:10:53.068 --> 00:10:55.168
Get usable answers instantly.
00:10:55.769 --> 00:10:58.229
That shifts the role of the data organization.
00:10:59.188 --> 00:11:11.428
Instead of being the team that hands out directions, it becomes the team responsible for making sure the navigation system itself is trustworthy by putting the right data, right guardrails in place.
00:11:12.028 --> 00:11:20.458
And here's the deeper change that most people underestimate when everyone has GPS access to data is no longer the constraint.
00:11:21.058 --> 00:11:23.489
Judgment on top of the data becomes the constraint.
00:11:24.089 --> 00:11:33.089
So when I look ahead, I think AI will change organizations first by collapsing the distance between insight and action, not just by replacing people.
00:11:33.688 --> 00:11:34.109
Okay.
00:11:34.379 --> 00:11:35.219
Fascinating.
00:11:35.969 --> 00:11:42.418
So do you feel there are any changes that leaders are still underestimating?
00:11:43.019 --> 00:11:46.158
It's hard to generalize but at a macro.
00:11:46.188 --> 00:11:51.948
But what, but at a macro level, I think there is both underestimation and overestimation happening at the same time.
00:11:52.879 --> 00:11:56.149
We only need to work towards finding the right balance in the near future.
00:11:57.139 --> 00:12:07.823
To extend the G Ps analogy, many leaders think AI as just getting a faster route or better traffic alerts that is helpful, efficient, and nice to have.
00:12:08.423 --> 00:12:17.333
What they're underestimating is that once everyone has access to AI, leadership is no longer about knowing the answer.
00:12:17.844 --> 00:12:22.354
It's about understanding what brings the answer, the context behind it.
00:12:23.313 --> 00:12:28.974
See, in the past, senior leaders had an advantage by having the best information.
00:12:29.274 --> 00:12:35.783
They could see the patterns that other others couldn't, and they had answers to the questions faster than the rest of the organizations.
00:12:36.639 --> 00:12:44.259
AI will quietly erode that a vantage When anyone in the organization can ask the questions like, what's driving this?
00:12:44.558 --> 00:12:45.219
What's hap?
00:12:45.278 --> 00:12:52.089
What happens if you change X and all of this is available at your tip of your fingertips?
00:12:52.839 --> 00:12:59.678
Then leadership is less about information access and it's more about direction, judgment and trade-offs.
00:13:00.278 --> 00:13:09.134
So if people can't tell the difference between the insights and the hallucination here, then they will stop using the system.
00:13:09.734 --> 00:13:10.094
Okay.
00:13:10.394 --> 00:13:10.783
Brilliant.
00:13:11.293 --> 00:13:19.394
And then, so there, obviously there's a lot of talk around AI complementing human roles in practical terms.
00:13:20.024 --> 00:13:21.413
So what does that look like?
00:13:22.014 --> 00:13:31.828
So I think AI will complement humans, but it's if I have to give an example, it'll complement the same the way autopilot technology compliments a pilot.
00:13:32.428 --> 00:13:34.494
The technology didn't replace pilots.
00:13:35.214 --> 00:13:40.913
It changed the way they spend their attention on, as we all know, on a modern flight.
00:13:41.484 --> 00:13:46.823
Autopilot flies most of the plane, but no one would argue that the pilot is less important.
00:13:47.423 --> 00:13:54.144
In fact, the pilot's role becomes more critical because they're no longer focused on constant manual control.
00:13:54.653 --> 00:13:58.374
They're focused on judgment edge cases and responsibility.
00:13:58.974 --> 00:14:02.094
That's what AI complementing human roles will look like.
00:14:02.094 --> 00:14:09.474
Inside enterprise organizations, AI can take over continuous, repetitive cogni cognitive tasks.
00:14:10.134 --> 00:14:19.043
Pulling the data, summarizing patterns, testing scenarios, or even answering the basic questions like what happened and what if scenarios.
00:14:19.644 --> 00:14:22.793
And that frees humans to focus on what AI is bad at.
00:14:23.244 --> 00:14:31.524
Like framing the right questions, understanding the context, weighing, tradeoffs, and deciding when not to act in a data organization.
00:14:32.453 --> 00:14:35.063
Analysts will spend less time building analysis.
00:14:35.543 --> 00:14:38.964
More time challenging them in product teams.
00:14:39.293 --> 00:14:44.423
Managers will spend less time chasing metrics and more time debating priorities.
00:14:45.024 --> 00:14:50.903
In leadership, executives will spend less time asking for updates and more time deciding direction.
00:14:51.504 --> 00:14:59.033
The key practical shift is this becomes the first pass, not the final say, just like autopilot.
00:14:59.124 --> 00:15:01.854
It works extremely well in normal conditions.
00:15:02.798 --> 00:15:11.828
But when something unexpected happens, like a pandemic, a policy change, or a cultural nuance, that's when human judgment takes control.
00:15:12.428 --> 00:15:17.109
And one more important thing here, complimenting doesn't mean splitting work 50 50.
00:15:17.979 --> 00:15:24.249
It means AI handles the known parts and humans handle the moments where map no longer applies.
00:15:24.849 --> 00:15:30.293
In my opinion, organizations that get this right, don't ask what AI will replace.
00:15:31.104 --> 00:15:34.849
They ask, where do we want human attention to be most valuable?
00:15:35.448 --> 00:15:37.489
Yeah, you make some interesting points.
00:15:37.489 --> 00:15:43.229
It's, similar to Excel, their stories were, they're gonna take all the accountancy jobs.
00:15:43.379 --> 00:15:44.609
Accountants are still here.
00:15:45.389 --> 00:15:48.958
Also LinkedIn, there's always, 20 years ago, they're gonna take all the recruitment jobs.
00:15:49.558 --> 00:15:50.249
I'm still here.
00:15:50.788 --> 00:15:57.239
So it is, like I said, obviously parts of it will be replaced, but you still need the human element, don't you?
00:15:57.239 --> 00:16:06.953
It is, it's just, I think it's just a lot noise on the market in my opinion, whether people are trying to sell their products or scare people.
00:16:06.953 --> 00:16:10.283
It's just humans are still gonna be here like it is.
00:16:10.464 --> 00:16:12.293
We're not getting replaced in my opinion.
00:16:12.293 --> 00:16:16.734
It's just you're going to get some of your roles gonna slightly adapt.
00:16:17.333 --> 00:16:24.719
So then what's the biggest challenges companies facing right now as they try to integrate AI into everyday work?
00:16:25.318 --> 00:16:27.808
I love giving analogy, so let me give one more on this.
00:16:28.609 --> 00:16:29.318
Let's go for it.
00:16:30.119 --> 00:16:34.769
so implementing AI reminds me of setting up a professional kitchen, right?
00:16:35.399 --> 00:16:48.859
So you can't create anyone who's worked in the food industry knows that you can't create great dishes or an excellent culinary experience unless you know your pantry very well, and it's organized and you understand how ingredients work together.
00:16:49.458 --> 00:16:57.558
That's exactly what I'm seeing with AI adoption today, companies are rushing to deploy AI without building to essential foundations.
00:16:58.188 --> 00:17:06.949
First, they're working with messy, inconsistent data, like just trying to cook when you don't even know what's in your ingredients.
00:17:07.788 --> 00:17:14.263
At AWS, when we started our AI journey, we invested heavily in standardizing data definitions.
00:17:14.864 --> 00:17:21.493
And ensuring clean, consistent information across the organization, but clean data alone isn't enough.
00:17:22.183 --> 00:17:27.193
You also need to understand how pieces of business information relate to each other.
00:17:28.153 --> 00:17:29.804
We call this as a semantic layer.
00:17:30.403 --> 00:17:33.374
It helps a understand business context.
00:17:34.273 --> 00:17:40.453
For example, how marketing influences sales pipelines, or how customer engagement drives revenue.
00:17:41.054 --> 00:17:42.403
And without this layer.
00:17:42.808 --> 00:17:48.269
AI might give technically correct answers, but contextually it could be meaningless.
00:17:48.868 --> 00:17:57.388
So by investing in these foundations, we are basically enabling AI to deliver insights that leaders can trust and act on.
00:17:57.989 --> 00:18:07.469
Yes, this totally requires an upfront investment, but it's a difference between AI that adds value versus the AI that creates noise.
00:18:08.068 --> 00:18:14.128
So in my opinion, the companies that get this right will have a real advantage.
00:18:14.729 --> 00:18:16.064
Yeah, no, I agree.
00:18:16.094 --> 00:18:22.653
And then so what separates organizations that have got the ability to move quickly from those that get stuck?
00:18:23.253 --> 00:18:28.233
See, organizations that move fast will all tell very similar stories.
00:18:28.818 --> 00:18:31.729
They don't start by going in all in on ai.
00:18:32.419 --> 00:18:39.709
They start with one real workflow owned by a real leader where nation get made or real nations get made.
00:18:40.548 --> 00:18:44.689
They focus on operationalizing AI inside an actual business workflow.
00:18:45.288 --> 00:18:50.058
The invest heavy on foundational data and infrastructure.
00:18:50.659 --> 00:18:52.608
It's not just about optimized data.
00:18:53.163 --> 00:19:04.354
It's also about building a data infrastructure that can be trusted, connected, and available at all times, and that one decision alone removes months and months of friction later.
00:19:04.953 --> 00:19:10.564
Another important thing is also here, another important thing here is how they move in small loops.
00:19:11.163 --> 00:19:17.193
They ship something that's good enough, watch how people use it and improve it in weeks, not quarters.
00:19:17.794 --> 00:19:21.243
improves because it's in use, not because it's in theory.
00:19:21.844 --> 00:19:28.023
Organizations that move faster, put light guard rails in place instead of heavy approvals.
00:19:28.173 --> 00:19:35.314
The teams can move faster without fear, and most importantly, leaders use AI themselves.
00:19:36.064 --> 00:19:41.193
When employees see AI shaping realizations at the top, adoptions become cultural.
00:19:41.794 --> 00:19:45.304
On the flip side, organizations that struggle do the opposite.
00:19:45.903 --> 00:19:47.824
They debate strategy endlessly.
00:19:48.153 --> 00:19:50.703
Chase for perfect data that never exists.
00:19:51.483 --> 00:19:56.493
Avoid experimentation and wait for confidence instead of building it through use.
00:19:57.094 --> 00:20:00.903
The difference here isn't ambition or spend.
00:20:01.564 --> 00:20:05.433
It's operation clarity and a willingness to learn in motion.
00:20:06.034 --> 00:20:15.048
And I'm guessing now as a leader, cause there's, must be so much want from especially like AI agents, from business divisions, I guess now.
00:20:15.858 --> 00:20:20.808
A big challenge is gonna be choosing the correct use case to implement.
00:20:20.929 --> 00:20:26.808
'cause as you mentioned, obviously you can't do multiple projects at once or your business is just gonna get unstuck.
00:20:26.898 --> 00:20:30.919
So I guess a big challenge now for leaders is choosing what's gonna have the biggest impact.
00:20:31.519 --> 00:20:32.358
Absolutely, yes.
00:20:32.419 --> 00:20:39.219
It's about understanding where there is value and picking the right use case, not just because it's a shiny object.
00:20:39.219 --> 00:20:41.108
We wanna pick it up and then start using it.
00:20:41.108 --> 00:20:50.108
Meaninglessly, it's about putting it to use where it really matters, and also understand that it's not just about trying to.
00:20:50.709 --> 00:20:52.449
Make the horses run faster.
00:20:52.628 --> 00:20:55.588
In many cases, you may not need the whole workflow itself.
00:20:55.919 --> 00:21:00.778
It could be about finding a combustion engine instead of trying to find faster horses.
00:21:01.769 --> 00:21:08.368
So there's a, there should be a lot of value on time spent in understanding where exactly this fits
00:21:08.669 --> 00:21:08.959
okay.
00:21:09.523 --> 00:21:19.963
Then as AI takes on more tasks, what new skills or mindsets will employees need to stay relevant and valuable with?
00:21:20.564 --> 00:21:31.324
See the, there's a long list, but for me, the most important skill is in technical mastery, but it's mindset First, it's an eagerness to learn and try what's next.
00:21:32.134 --> 00:21:36.663
The people who stay relevant won't wait for perfect clarity or formal training.
00:21:37.068 --> 00:21:39.798
They will experiment early and learn my doing.
00:21:40.398 --> 00:21:46.169
Second is about being comfortable at being bad at something.
00:21:46.169 --> 00:21:46.798
While learning.
00:21:47.398 --> 00:21:55.618
AI is evolving so fast for expertise to come first, progress comes from trying, failing quickly and iterating.
00:21:56.219 --> 00:22:01.979
Most importantly, instead of worrying about what might replace us, it's far more productive to ask.
00:22:02.564 --> 00:22:05.443
What does this free up to do better?
00:22:05.864 --> 00:22:09.044
Where can I add judgment, creativity, and context?
00:22:09.644 --> 00:22:12.284
That's where creativity becomes a real differentiator.
00:22:12.884 --> 00:22:28.213
It's not just about using AI to be faster, but to think differently and create a new value by experimenting and understanding where we could create new business workflows that we couldn't think in the past.
00:22:28.814 --> 00:22:29.324
Brilliant.
00:22:29.413 --> 00:22:37.269
Then how should leaders rethink the way teams are structured when humans and AI systems start working side by side?
00:22:37.868 --> 00:22:47.138
As AI and humans work together, team structures will shift away from rigid roles and more towards orchestration.
00:22:47.739 --> 00:22:50.648
What works well are smaller and autonomous teams.
00:22:51.249 --> 00:23:00.548
Just like at Amazon, we do our two piecer teams that will own a customer problem end to end and irate quickly with AI embedded in the workflow.
00:23:01.148 --> 00:23:04.929
Also, another interesting shift is how engineering roles are evolving.
00:23:05.739 --> 00:23:13.538
The line between product and engineering is starting to blur and it's leading to the rise of a new role called product engineers.
00:23:14.138 --> 00:23:24.909
This role basically stays close to the customer problem and decides how best to solve it, whether through code configuration or AI orchestration in general.
00:23:25.259 --> 00:23:33.959
Going a little deeper on the engineering side, engineering becomes less about writing code and more about designing systems.
00:23:34.558 --> 00:23:45.959
It's going to be about deciding which tools to use, how agents interact, where guardrails belong, and where human judgment matters that I think enables speed.
00:23:45.959 --> 00:23:59.338
Without Kio, we are gonna see fewer handoffs, tighter feedback loops, and faster translation from intent to execution, not just because teams are bigger, but because they're sharper and better aligned.
00:23:59.939 --> 00:24:00.179
Okay.
00:24:00.179 --> 00:24:05.118
It's interesting you, as you mentioned around pro project sorry, product engineering.
00:24:05.749 --> 00:24:10.983
So is there any other early examples of team structures that you've seen work well?
00:24:11.584 --> 00:24:16.983
There's definitely a lot of experimentation happening and different models work better in different contexts.
00:24:17.284 --> 00:24:31.253
Continuing on the product engineer role in, at our organization, we've introduced this role last year and the idea was to bring product thinking and engineering execution close together, especially as AI became part of everyday work.
00:24:31.854 --> 00:24:34.838
What we are seeing so far has been truly encouraging.
00:24:35.513 --> 00:24:43.523
Particularly around rapid prototyping and iation teams are able to go from an idea to something tangible much faster.
00:24:44.003 --> 00:24:48.683
This and the same people are able to shape the problem and work directly with the tools.
00:24:49.284 --> 00:24:54.588
That said, it's still early days and we are very much in a learning phase and super excited about it.
00:24:55.189 --> 00:24:58.249
Yeah, and I think obviously as these new team structures are.
00:24:58.848 --> 00:24:59.479
Blended.
00:24:59.659 --> 00:25:03.919
The challenge is gonna be actually hiring people for these roles.
00:25:03.949 --> 00:25:21.138
'cause even taking sort of the, I get the product role, you mentioned this, obviously now you've got product data scientists getting the right combination of technical skills, product skills, and also business domain expertise is a massive hurdle for cha businesses that wanna progress quickly.
00:25:21.709 --> 00:25:23.568
So I guess that's gonna impact.
00:25:24.423 --> 00:25:32.153
Other sort of new business models or new team structures as like the AI that's within the co within companies.
00:25:32.753 --> 00:25:33.534
Absolutely, yes.
00:25:33.534 --> 00:25:38.784
It's going to see, it's about a paradigm shift in how these roles are shaping together.
00:25:39.084 --> 00:25:43.453
Like you said, it's an intersection between a product and a science and an engineering role.
00:25:44.084 --> 00:26:02.453
That's giving birth to these new roles, either the product engineer, which is at an intersection of an engineer and a product manager or a product data scientist between a science roles and a product manager, or even a super engineer who was at the intersection of a data software and a business intelligence engineers.
00:26:03.054 --> 00:26:12.614
So we are not going to find ready-made skillful professionals with all acquired skills in the industry, but we are gonna work towards it.
00:26:13.003 --> 00:26:21.104
And that's where I think as a leader, love to have the patience to make sure the culture is set for that experimentation.
00:26:21.104 --> 00:26:27.913
People will fail more often than they've seen in the past while experimenting and while acquiring new skills.
00:26:28.453 --> 00:26:37.273
And it's totally fine to make sure as lo as long as we are going in the right direction, as long as there is learning.
00:26:38.219 --> 00:26:41.118
Faster than the pace of the problems that we are facing.
00:26:41.239 --> 00:26:47.648
It's absolutely fine to basically keep experimenting, learning on the go and see how it goes.
00:26:48.249 --> 00:26:48.459
Yeah.
00:26:48.459 --> 00:26:55.544
And I think in the future sort of learning and development like teams, that's gonna be where businesses succeed.
00:26:55.943 --> 00:27:02.663
If you're investing in your team to learn these new skills, they're gonna get up to scratch a lot quicker than just learning on job.
00:27:02.784 --> 00:27:04.969
So I think it's gonna be interesting how businesses do this.
00:27:04.999 --> 00:27:08.058
'cause even now, like obviously I work in recruitment, we're seeing.
00:27:08.114 --> 00:27:10.124
Businesses struggle to hire people.
00:27:10.183 --> 00:27:15.763
'cause a lot of them are just looking for all the skill sets when really you're not gonna get that.
00:27:15.854 --> 00:27:29.534
'cause the industry is so diverse now, so many different roles, so many different positions that where data roles previously were more technical, they now have more business stakeholder, product focus.
00:27:29.534 --> 00:27:33.034
So there's so much out there now and it, and I guess to make it.
00:27:33.864 --> 00:27:34.614
Even more challenging.
00:27:34.644 --> 00:27:37.314
Everyone is trying to become more data driven.
00:27:37.314 --> 00:27:42.743
So the competition, everyone wants these skill sets, so it is challenging for businesses to keep up.
00:27:43.344 --> 00:27:47.213
So true the, it's very interesting to see how the business teams.
00:27:48.159 --> 00:27:52.568
Trying to become more and more technical, and AI is totally helping them to get there.
00:27:52.868 --> 00:28:00.818
And how the technical teams, like engineers and scientists want to become more and more business savvy.
00:28:01.239 --> 00:28:03.009
And AI is helping that too.
00:28:03.729 --> 00:28:05.199
So it's a very interesting phase.
00:28:05.199 --> 00:28:14.108
We are gonna see everyone trying to pick up and acquire something which they were not exposed to or something they, something that was not the skill in the past.
00:28:14.709 --> 00:28:24.798
And that's why I truly think it's going to make everyone a lot more productive and skillful in the next few years with the availability of all the tools in front of us.
00:28:25.788 --> 00:28:26.148
Yep.
00:28:26.209 --> 00:28:26.868
Definitely.
00:28:27.598 --> 00:28:28.679
Moving to the next question.
00:28:28.769 --> 00:28:37.469
When businesses ask how do we prepare for the future of ai, what's the smartest first step they can take right now, in your opinion?
00:28:37.834 --> 00:28:43.864
I usually start by saying, embrace AI in a way that genuinely works your advantage.
00:28:44.673 --> 00:28:51.523
Be clear about what, it creates real value, not just where it's exciting, and sometime that starts with data.
00:28:51.884 --> 00:28:55.874
A strong data foundation matters, especially as you introduce agents.
00:28:56.443 --> 00:29:04.544
Agents will need clear definitions, clear inputs and shared context, the judgment, what a human being could apply.
00:29:04.723 --> 00:29:11.314
When there is ambiguity and when there is multiple definitions agents cannot really apply that judgment.
00:29:12.064 --> 00:29:18.364
So it's important that we have very clear definitions and we call it as metrics store at Amazon.
00:29:18.963 --> 00:29:21.439
And without that, agents will.
00:29:22.263 --> 00:29:28.233
Basically give you inconsistent outputs and there could be hallucinations and that leads to loss of trust.
00:29:28.834 --> 00:29:31.324
And second one, it's also, again, a mindset shift here.
00:29:31.923 --> 00:29:34.443
AI isn't about doing the same work faster.
00:29:34.894 --> 00:29:38.284
That's the faster us faster horse strap I was talking about.
00:29:38.973 --> 00:29:45.273
The real opportunity is rethinking workflows entirely, and that's where the step change impact comes from.
00:29:45.874 --> 00:29:46.564
See, we don't need to.
00:29:47.163 --> 00:29:48.544
out everything from the get go.
00:29:48.844 --> 00:30:02.374
But if we start with clear values and understand what exactly we are trying to do, and with some solid data foundations and a willingness to rethink workflows, we are setting ourself for success.
00:30:02.973 --> 00:30:05.824
Ji, I really appreciate your Thank you for conversation.
00:30:06.423 --> 00:30:09.318
It was awesome being here and thoroughly enjoyed the conversation too.
00:30:10.074 --> 00:30:11.003
Thanks for having me again.
00:30:11.979 --> 00:30:12.338
No problem.
00:30:12.338 --> 00:30:25.598
And I wish you all the best for 2026 and also for thanks for listening, everyone, and I wish you all, you hit your goals for 2026 and I'll see you in the next episode of Data Analytics Chat.
00:30:25.898 --> 00:30:26.259
Thank you.
