- 1 day ago
Category
🤖
TechTranscript
00:00We want to begin this conversation by letting you all into a conversation we were having just
00:05backstage there, talking about how businesses are integrating AI into their organizations.
00:11And I know, having worked alongside the Formula One teams, that they have completely transformed
00:17their organization, the way that they operate, through the integration of AI. What I didn't know
00:22until just now is that it's actually Tata Communications that have completely underpinned
00:27that evolution of the way that they operate. So I'd love to begin just by asking you about that
00:32partnership and how that transformation has happened. Thank you. Very pleased to be here.
00:37Good morning, all. Now, Formula One has been a great partnership. I think the CEO says that prior to
00:45COVID, every race involved taking three Boeing full of equipment, and he called it a traveling circus.
00:55And we've been able to transform that completely. Now, you know, amazing how the teams work,
01:01the 24 races around the world globally, bringing all the signals from the cameras, from all the sound.
01:10We bring it all to London, and then it gets produced here and then taken to millions of viewers around
01:16the world, all in literally a blink of an eye. And that's testament to not just the technology that
01:25we've put in place to make that happen, but also very agile team working. And this is the experience
01:32that we transport to our larger enterprise customers like banks and manufacturers. And we sort of
01:40cheekily tell them, if we can do for F1, then we can do that for everybody else.
01:45So I think that's a partnership that we're really, really proud of.
01:49And as companies now sort of bed into this world of AI, how do they move from the hype
01:57to actually practical integration across organizations from your vantage point?
02:03I think the biggest, as of now, the biggest drawback that we see is people are looking
02:10at AI as a tool, trying to see what use cases that we can work with. But really, in our opinion,
02:21AI needs to be embedded. It needs to be absorbed into the DNA of an organization and needs to be thought
02:29more holistically. We ourselves, we have sort of developed a framework and we are asking our
02:36customers to probably do the same. What this framework does is we are evaluating ourselves on
02:42one axis on the capabilities, where we look at strategy, talent, culture, because those are very
02:49crucial element of how we think about AI. And on the y-axis, we are looking at what outcomes we are
02:56delivering for that. You know, is it in terms of cost savings, in terms of improved experiences for
03:02employees and customers, in terms of actual revenue that brings in. So, you know, real measurable ones
03:08on both the axis. And we have said, you know, you are in the early stage of experimentation. You are,
03:14you know, and the topmost end is real achievement that we do. And we have administered that in
03:20different parts of our organization and looking at our HR teams and how they have, where they fit.
03:25So different parts of the organization are at different maturity levels, but that gives us a
03:30starting point on how we improve. So I think my firm view is you need to look at AI. It's the new
03:36operating system. It's the new DNA that we need to work. We're also doing, you know, for example, in the
03:42HR, you know, what does, how does a future job description look for an employee, right? So what an employee does
03:50today and what the person needs to do tomorrow for the same job would be very different with the help of AI.
03:56So we are actually writing down what the North Star job description is going to look like.
04:01So that's one element. The employees also, and we've implemented a full agentic AI integrated to our
04:08learning systems. So when somebody goes into that, they would know where they are today. And in order to get to the North
04:14Star job description, what are the skills that they need to acquire? And this agent would keep nudging the
04:20person to say, you need to acquire the skills. These are the courses available for you. So you need to do that.
04:26So these are some, something I think, you know, you need to move the entire organization along on the full axis of
04:33strategy, talent, and culture, and constantly measure what outcomes you're going to deliver. That's what is going to
04:40help people to move from the hype to reality. Yeah, because as you say, you need to embed it across
04:46the entire organization. And it seems that maybe some organizations are struggling to move from that
04:52pilot phase into really embedding it across the organization and then being able to scale it. Is
04:57that what you're seeing? What do you think companies need to do to ensure that they are embracing it for
05:03the scale? Absolutely. I think we, one of the things that we advocate is more platform thinking, as opposed to
05:11thinking that in silos. And what I mean by platform is there are three essential elements in platform. So we ourselves
05:21have just launched an agentic AI with voice AI embedded in that. So what I mean by platform is, first and foremost is
05:30integration. So the data is all fragmented in many places. So how do you unify data and get it to a place where it's
05:40accessible by the agentic AI? You don't have to bring all the data to one place. That's another challenge with all the
05:46regulations in different parts of the world. You can't, so, but you still need to be able to access all the data. And
05:52that's what is a seed for the intelligence. So that's the first thing. The second thing that we're doing about when we think about
05:58platform is, it's an intelligent orchestration, right? Because the models bring the intelligence,
06:05and there are different models. And we talked about Mestrel, we talked about other, so different models
06:10that brings the intelligence. But what is missing is more intelligent orchestration across these.
06:16And that is where, you know, the context gets lost. So in our voice AI, for example,
06:22when the voice AI calls a customer, it knows the context of the customer. That this customer,
06:28for example, you know, if it is what we're implementing for a large automotive OEM, is
06:34the customer has been looking at cars, and this person has been looking at more adventure oriented.
06:42They want to travel long distance, and it could be another person who's looking more at safety features.
06:48So when a voice AI calls that customer for giving a nudge, it would know the context of this customer,
06:54and fine tune the conversations according to that, so you can get to the first goal of booking the
07:00test drive, because that's the first goal. So that is something that the whole world is moving towards
07:06now context engineering, and that context gets lost. Even humans lose that context. Whereas with AI,
07:12you can remember the context, you can bring that all together. So that is the second layer that I talk
07:16about, which is, how do you bring more intelligent orchestration once you have unified the data?
07:21And the third is about governance, and the guardrails that you require so the trust can be there.
07:27The guardrails to make sure that the models are not drifting, and you know, you can make it more
07:34self-learning, as well as whatever is required, the human intervention that is required to make sure
07:38the models don't drift. And once you start drifting, then you lose the trust in the model.
07:43Right. So these are the three important elements of a platform, and we've been able to bring that
07:48together. So my, you know, my formula for doing that at scale, besides the first thing that I talked
07:54about, the maturity, is about platform thinking, which brings the integration, intelligent orchestration,
08:01and appropriate guardrails for that. And not the, as part of the governance also, I would put cost as
08:08an important element, because people don't think about the cost. People know that, you know, when
08:13they move to cloud, you know, I say it's Hotel California, you go in and you can check in, but
08:18you can't check out. The cost is very high, the egress costs are very high, the costs are, with AI,
08:23especially when people implement at scale, they're going to be doing so much of inferencing,
08:28the cost of inferencing is going to really skyrocket. And people have to think about it very
08:33consciously when they, when they, you know, choose the platforms, when they choose the models, and
08:39think about how they have to scale. So I would put, you know, these three as critical factors,
08:45if companies have to move from pilot to the scale stage.
08:51And we have to wrap, which is really unfortunate, because we could talk all day, which we can come
08:55and continue to do afterwards. But just real quick, before we do, I just want to understand from you,
09:00do you think organizations are ready? Very quickly, do you think they are ready and what
09:04they need to do to make sure they are prepared across their infrastructure, across cloud compliance?
09:10Honestly, I don't think people are ready for scale. I think the infrastructure is very fragile
09:17and across the networks. And when you talk about networks, the LAN, the WAN, the cloud, the data is
09:24fragmented. And all these fragmentation, you know, means that, you know, AI is just going to add more
09:31stress to this fragile infrastructure. So one of our customers that I was talking to in London, they
09:38said they've transformed the applications, they're doing AI, but network has been the last bastion of
09:43change. And somehow network is seen to be the last bastion to touch. We believe that people have to
09:50start thinking about it, prepare the grounds of the entire digital infrastructure, especially if you're
09:55a multinational company, how do you think about your network, you know, where your data resides,
10:00how are you going to integrate all of this, and make it really, and that's why we came up with this
10:03concept of a digital fabric, a digital fabric that truly helps companies to uncomplicate things,
10:10and then you can innovate from there.
10:19So
Recommended
5:40
|
Up next
12:33
19:10
4:43
0:38
9:42
9:51
4:31
6:33
Be the first to comment