This episode of The Analytics Edge, sponsored by NetSpring, features an interview with Awinash Sinha, CIO at Zoom and Bask Iyer, former CIO at VMware, Dell, Juniper Networks, and Honeywell & Advisor at Zoom, an all-in-one intelligent collaboration platform that makes connecting easier, more immersive, and more dynamic for businesses and individuals. In this exclusive episode, Awinash and Bask dive deep into the impact of AI at Zoom, and provide advice for data leaders on generative AI, how to enhance the customer experience in product-led companies, and the importance of hiring “outcome enablers."
This episode of The Analytics Edge, sponsored by NetSpring, features an interview with Awinash Sinha, CIO at Zoom, and Bask Iyer, former CIO at VMware, Dell, Juniper Networks, and Honeywell & Advisor at Zoom, an all-in-one intelligent collaboration platform that makes connecting easier, more immersive, and more dynamic for businesses and individuals.
Responsible for the company’s information technology, data science and analytics, and business application organizations, Awinash brings to Zoom over 20 years of experience in enabling business outcomes and scaling fast-growing companies. He has completed an executive program in Business Administration and Management from Stanford University Graduate School of Business.
Having served as an advisor and mentor to Zoom since 2016, Bask is a renowned technology executive, leading enterprise-wide transformations in digital systems at numerous companies in Silicon Valley. He is the CEO of BaskMind.com, an experienced team offering hands-on digital transformation and operations excellence services to traditional companies and advisory services to high growth technology companies. Iyer holds a Master’s degree in Computer Science from Florida Institute of Technology.
In this exclusive episode, Awinash and Bask dive deep into the impact of AI at Zoom, and provide advice for data leaders on generative AI, how to enhance the customer experience in product-led companies, and the importance of hiring “outcome enablers."
-----------
Key Quotes
"In subscription business you have to really closely follow the life cycle of customers. And it's a infinity loop, right? Even existing customers, you may be looking at feature adoptions, you may be looking at churn, you may be looking at how we can create more value, make them aware about new products and offerings coming together. And internally, for the sales department, marketing department, or customer support department, providing the insight, both coming from product. Basically with this cloud architecture we have an ability to look at product telemetry data. As well as business transaction data and intersecting them. The magic is, the real insights are when we intersect these two data, join these two data and do a cohort analysis. Cohort analysis at product level, at segment level, particularly for large enterprise companies our size and bigger, will have a customer segmentation, will also will have some flavor, some in geodes dimensions or other dimensions. Once we connect the data around product and then customer lifecycle from business transactions, it could be slice and dice from many different perspective, and that's where the insights come." - Awinash Sinha
-----------
Episode Timestamps
(01:39) Approaches for CIO’s and Data leaders to consider regarding AR and VR
(03:31) Zoom’s strategy for AR and VR collaboration
(06:29) AI-powered innovations improving user experience on Zoom
(07:49) Zoom IQ explained
(11:25) AI's role in addressing COVID-driven challenges for engagement in online education and business interaction
(16:59) Advice to C-level executives on generative AI strategy
(22:05) Approach of starting initiatives despite data quality concerns
(27:29) Future of the "data scientist" title amid AI techniques and evolving roles
(30:54) Warehouse-centric data approach and its impact on data leaders
(31:57) Advantages of data warehousing being on cloud
(37:03) Strategies for product-led analytics leaders to boost customer experience
(39:52) Value of information and time on data
(42:38) Consumer-centric approach and its impact on business
(45:05) Call to action for data leaders
(51:58) Takeaways
-----------
Links
[00:00:00] Narrator: Hello and welcome to the Analytics Edge, sponsored by NetSpring.
[00:00:08] Thomas Dong: The Analytics Edge is a podcast about real world stories of innovation. We're here to explore how data-driven insights can help you make better business decisions. I'm your host Thomas Dong, VP of Marketing at NetSpring, and for today's episode my co-host is Vijay Ganesan, Co-founder and CEO at NetSpring. Thank you for joining me on the show today, Vijay.
[00:00:25] Vijay Ganesan: Great to be here, Tom. Really looking forward to talking to these two guys.
[00:00:31] Thomas Dong: In today's episode, we'll be exploring several topics that are top of mind for CIOs from the promise of spatial computing to AI and data strategies. We're joined by two very special guests, Bask Iyer, the former CIO of VMware, Dell, Juniper Networks, Honeywell, and GlaxoSmith Klein, and the longtime advisor to Zoom, and Awinash Sinha, Zoom's current CIO. Zoom, of course, grew to global mainstream prominence during the Covid 19 pandemic, but had already for years firmly established themselves as the most reliable and easiest to use and set up web conferencing tool. Bask, Awinash, we're delighted to have you both with us today. Welcome.
[00:01:08] Bask Iyer: Thank you.
[00:01:09] Awinash Sinha: Thank you. Happy to be here.
[00:01:12] Vijay Ganesan: Thank you so much for being on the show. You guys are thought leaders in this space, lots of extensive experience in the business, so really looking forward to hearing your thoughts today.
[00:01:22] Awinash Sinha: Thank you.
[00:01:26] Thomas Dong: All right, let's start with Apple's launch of their vision Pro augmented reality headset. And with this launch, AR continues to hold great promise for transforming the customer experience and how consumers will interact with business and products in the future. Bask, how should CIOs and data leaders be thinking about AR or VR?
[00:01:43] Bask Iyer: I think personally, I made the mistake of betting against Apple a few times. You know, when they came up with the headsets, I thought, how much can you do with a headset on in our headphones? A headphones? And can you make it physically better than a physical headset? I [00:02:00] was wrong. And so I've been, uh, I've been wrong on all these counts. I was just blown away with experience. I think Apple has focused on experience and sometimes corporate CIOs.
Don't care much about experience. They're, they're, they're getting into practicality of things. This product actually allows you to dream a little bit and, uh, you know, you could only imagine version two and three of this and how much value it'll be to even industries, whether it's medical, healthcare to health doctors remotely, uh, diagnose and do things or, you know, [00:02:30] like engineering and a company like, uh, aerospace companies, you know, they can remotely wear this and.
Inspect several things. So I am, I'm not gonna bet against it. It is expensive. But, um, you know, I never thought I would buy an expensive phone. I never thought I would pay a thousand dollars for a Ford. And here we are every other year, we pay to get the next version of a form. So, uh, I would recommend the CIO CIOs as you, as the, you know, our risk over generally, they wait till the technology's all proven out and wait [00:03:00] till somebody else has done a use case and case study before they experiment.
I would urge them to take a chance and go forward and then you create the use case. Don't wait for somebody else. Don't follow. Be a kind of a path leader. It's not that expensive to get for corporation that to try to things up. I'm excited. It's a good toy, but it's gonna be very useful very soon.
[00:03:21] Vijay Ganesan: One of the use cases Apple is featuring is meetings, and obviously Zoom is the leader in the business and they, uh, proclaim, make
[00:03:29] Awinash Sinha: meetings more [00:03:30] meaningful.
Right. So how is Zoom thinking about AR and VR for collaboration? Yeah, that's a good question, which I, I think these two companies are pioneer in their own space and then they are coming together trying to create a synergetic value. For our consumer space as well as for the enterprise spaces, and help you explain this this way, think of this hybrid world when people are coming both in offices, people remotely, how do we [00:04:00] create an immersive experience and inclusive experiences where the quality of experience is no different?
Whether you are in a room or you are at home. That's where the intersection of two best of breed providers in terms of the collaboration softwares and in terms of the devices comes together. And we have done it at, uh, other circumstances. Also, if you look at it like a year ago or something, we launched in Tesla, Tesla car and experience for accessing [00:04:30] seamlessly using the devices and the camera in the car to attend the meeting, right?
Think of on a larger scale, we are doing this partnership to do that. And key thing about is not just the product, the basic core values of these two companies. If you look at it, Zoom's products are experience driven. We have a differentiator in meeting experience compared to our competitions. Apple is always cares about the customer experience.
So there is a value level. Synergy is [00:05:00] also, we care about BA little. We touched around experience design. These two companies care about experience designs. So that's a natural partnership between us. Also, one more point of right from our c e o, we talk about, we use a language of customer happiness. And apple is all about, we don't use the word delight, uh, but customer is happiness and delight is also very critical for Apple.
So on these two terms, if you look at it broadly, [00:05:30] two, best of the breed players coming, adding synergetic value. Creating an inclusive hybrid experience and the core values of the companies are aligned, so there would be a lot of magic people will experience just like any Apple product launch. Let people realize what the experience is versus we talking about now.
[00:05:51] Thomas Dong: Yeah. It's very interesting how the user experience is so central to to both, uh, apple's and, and Zoom strategy. And I'm delighted to hear that [00:06:00] I can now take Zoom calls in my Tesla. Um, so you obviously have many, many, uh, devices that you support over user experience. I'm curious how you might be using AI techniques to, um, make, uh, that experience, uh, delightful for your customers.
Uh, generative ai, large language models, for example, or, you know, it's beginning to gain mainstream attention. Many other AI techniques have been around for decades now. What are some of [00:06:30] the, uh, innovative applications of AI you're using, uh, to help with the user
[00:06:34] Bask Iyer: experience?
[00:06:35] Awinash Sinha: Yeah, Tom, uh, if you look at it, we have been using AI for a very long period of time.
If you look at how did we create the differentiation in the meeting experience itself, to start with the video quality on low, low bandwidth networks, different kind of network settings, the audio quality, the noise, cancellations behind all of these stuff is machine learning models. A lot of ai. [00:07:00] That has been running for last four, five years that we created us the, uh, differentiation in the marketplace.
Now, more recently, if you look at the contact centers that we have, the other modules, the products that we have, zoom virtual agents, intelligence routing, near real time meeting summary, uh, near real time translations. These are complex use cases. Underneath is all ai, AI technologies, all kinds of AI technology.
[00:07:30] AI is a broad term, right? Machine learning and statistical analysis. L L M. Recently, everybody is now pivoting to chat G P T, but this is just one of the tools in the broad AI umbrella. Now, some of the tools we have that's heavily leveraging the large language models, particularly if you look at Zoom IQ as all our conversations are digital these days.
There is an opportunity to do convers conversational analytics on top of it, sentiment analysis on [00:08:00] top of it. And we have created something called Zoom IQ because we have underlying data of this digital interactions amongst us. And that is leveraging both quantitative side of ai, which I call like ml, and then the qualitative side of conversations.
And this is also underneath his ML model though, and we're bringing this to together. So across our product groups now, we have been using AI in the past. Now we added an augmented with L L [00:08:30] M. So we are talking about employee collaborations and, and the customer collaborations where AI is not an add-on or bolt on thing is embedded as a fabric inside it horizontally.
So that's our positioning. We had a hard internal conversations around how do we go about this from a technology architecture perspective. And we have taken an approach where we are going with the hybrid and directed [00:09:00] AI strategy. And I'll explain very quickly what what we mean by that. So we are leveraging our proprietary models.
We are also leveraging, have a potential to leverage any open source public data. Like charge J p T or others models. We have gone into alliance with a third party, uh, uh, partners like Anthropic. We have made investments in them and we are giving customer a choice to leverage their own model to argument with us.[00:09:30]
So it's a open, flexible architecture that not locks the customer into one approach because these things really evolve and we wanna go with the flexibility. So we are calling this as a federated architecture for AI ml. So in nutshell, our product had been using AI for long. We have doubled down on this effort.
It's now embedded as a fabric across our product lines. And then we have now right technical architecture to support evolving marketplace [00:10:00] as we go into this.
[00:10:01] Vijay Ganesan: That's interesting what you said about federated architecture that you don't
[00:10:05] Awinash Sinha: want lock-in and you wanna give, you give
[00:10:07] Vijay Ganesan: people the freedom to plug in their own models and uh, these models are rapidly evolving, so it's very
[00:10:12] Awinash Sinha: interesting and very contextualize as well.
Right? So same, same zoom mike. You could be deployed in healthcare versus financial client, client relationship with the wealth management people in so many context. So then it could be optimized for that vertical with. I wanna ask a follow
[00:10:29] Thomas Dong: [00:10:30] on question. I think everybody on the planet now knows Zoom because of the Covid pandemic.
You know, you have young children who can connect to Zoom even more easily than our parents. Um, now just become so embedded in everybody's lives. And Zoom, of course, has become a noun in a verb. In popular culture, education has driven a lot of new requirements, and one of the ones that I as a parent heard a lot about is like, how do you make sure the kids are engaged?[00:11:00]
When the teachers don't even know how to teach online yet, and we hear about, you know, all hands meetings in a corporate setting now where business leaders can track engagement of their employees during the meetings. How have you applied AI to kind of solve some of these, uh, new emerging requirements that, you know, just.
Obviously popped up and became so, so vital to how we do business or how we interact in collaborating in, in kind of this new normal.
[00:11:27] Awinash Sinha: I think this is a, for our engineering [00:11:30] team, IM possibilities that remains technology already exists. It's a question of whether we prioritize from a backlog perspective.
If we look at it right, it's huge for abstracting and we are partly doing it for sales. So when you are on a sales call, zoom iq. We are giving and developing this, uh, uh, this kind of insight and giving that, how was the pitch customer feel like engaged or not engaged, and what is my peer group, other sales setup are doing?
Why that call was successful, why that transaction was [00:12:00] successful. Whether I listened to you, I didn't ask any questions. If I'm a customer, this means I'm kind of either, uh, not coming across or I'm disengaged and I just give you a time. Right? If I ask you too many questions and all tone of the questions was critical, I'm just trying to find a reason to sort you down now, why I should not move forward with you.
Right. There's a lot can be done right now. I think that's where the next wave of automation is coming. It's gonna be, nobody will be talking as separate i m O kind of stuff. [00:12:30] It'll be embedded. It's the automation 2.0 or 3.0, whatever you wanna call it. Just like we had in, what is that now? Industrial Revolution 4.0.
Something like that. That's, that's where the, we are at cusp of this. I can feel that mask. You may sense it this way. It'll take time to walk, but, and it, these things takes time. I'll have a concrete example. 19 98, 90 nine.com came up, right? E-commerce came up till pandemic companies were brick and mortar [00:13:00] companies were shifting from their on-prem to, uh, what a 20 year cycle.
20 year cycle took to become mainstream, takes time.
[00:13:11] Bask Iyer: The only thing I would, I thought was an interesting question. The only thing I'm saying is, uh, I, I'm truly convinced it took me all this time to realize, I'm truly convinced, you know, the skills I developed were in high school because I had a new good teachers.
It's not my master's, it's not my PhD. It's not like whatever. It's, uh, it's, it's, a [00:13:30] lot of it is high school. The way you go across the soft skills, your English skills, the language skills. Um, unfortunately those teachers are not recognized in Arcade well, and they're doing a lot of, you know, non-value added job, you know, grading people, checking if people is paying attention and so on.
So I am bullish on ai. I'm not scared. I have a feeling we still do a lot of things we don't like to do. Right? And, and with these teachers are doing a lot of work that has got nothing to do with what they went to school for and their passing. I think AI can [00:14:00] take care of all of that. You know, the take the mundane things out or on people, whatever.
So they could focus their passion and, and I don't, I'm still an idealist. I don't think AI is gonna have the passion to have that thing that the teacher can teach. That's exactly. Sparking you to say, I'm gonna study math. Right. You know? So we all know that story where some teacher inspired you to do something.
They're beginning, paid nothing in any country. This is the universal truth that nobody gets paid and they're doing like really awful [00:14:30] stuff, right? Instead of teaching. So I think things like Zoom made it in. Interesting. Like, I mean, I remember my sister, a teacher, she said, oh, thank God I can connect in a, in a so quickly and not worry about if my microphone set up, camera set up, we set up.
He's not gonna the school to do all he wants to click the button on his phone and start. But there's still a lot of things that she does now I'm sure, that are better than useless. So I think the promise of AI is removing Mindly jobs that all of us do, so we can [00:15:00] work on what is truly our passion and what we were actually put all the earth to do.
So I'm pretty optimistic about. I love
how
[00:15:08] Thomas Dong: you stated that. So we can all focus on our passions and let's automate all the mundane things that we hate doing, and especially for teachers who are so undervalued, but so critical to, uh, to our development as a society.
[00:15:20] Bask Iyer: So that fantastic, uh, perspective.
[00:15:25] Vijay Ganesan: Bas, I'm, I'm sure you're seeing this a lot.
You know, every company, [00:15:30] CIOs, CDOs, they're tasked with coming up with the generative AI initiative. I mean, that's the, that's the charter. And everybody is waking up to this and realizing there's a lot of potential and they know they need to do something, but they don't know what, where do, where do we start?
Right. You know, this is, this is big. We have to get into this if our competitors can disrupt us if we don't get into it. But where do I start? So what would be your advice to a a c-level exec at a large company [00:16:00] that is tasked with coming up
[00:16:01] Awinash Sinha: with a strategy around generative ai?
[00:16:04] Bask Iyer: So, I lost my bet already.
My wife told me, can you do one podcast, one meeting without mentioning Chad G B g I locked in the first five minutes of the conversation, we've lost at that. Yeah, there's so much buzz. Right? So, but the problem is, at, at this stage, the buzz is coming from, at a board level, at the c e o level, right? So they're asking, the question you're asking me is, what are we doing with, uh, [00:16:30] generative ai now that they kind of use the consumer version of chat gt or, so they're, they're asking the question.
Um, unfortunately, a lot of my, uh, I, I'm advising a few CIOs, but some, a lot of the CIOs are at a wait. Attitude, which, which is a mistake with this tech, right? They're waiting to see what other use cases can people do. And a lot of them are just answering, you know, the board requests them to say, what are you doing with, uh, chat g pt?
And they write a white paper, which is kind of regurgitation of what other companies are [00:17:00] doing. Like, I mean, every use case is that we can do a customer service, we can build this plan. What the board now wants is actual projects that they can work in place in some, some practical experiments to do. So, um, you know, I'm actually advising the CXOs and, uh, CEOs more or less to say, go ask them what they are doing today.
Don't ask 'em for a white paper or what they'll do tomorrow, because if they're not doing something today, even if it's wrong, right? If they're doing some silly experiments now, those are the [00:17:30] CIOs, CPOs, and people you should bet on. Uh, not the folks who give you a white paper on it because we don't know, uh, I mean, Listen, I, we all worked in ai.
I, I just, uh, somebody reminded me, my wife invited me. We both did the thesis on AI 30 years ago. And, uh, you know, so AI has been around and, and, but we couldn't make simple things. We couldn't do natural language processing. We tell the computer to differentiate between Larry Bird and, and Larry the Bird. I mean, it got confused.
Uh, [00:18:00] we wrote the worst programs and we struggled. We actually got divorced even before we got married because of the project in e I. So, um, the issue now is though I'm, and, and we always thought that neural network course was kind of a bogus thing, right? I mean, we, all of the stuff felt like unless you have logic, how can you program some computer to do it?
What is scary for me is this seems actually seems to work. We feed it a bunch of data, and this thing seems to have a mind of its own, if you will, right? I mean, the, the [00:18:30] academics are gonna freak out, but to me it feels like, what is the logic? How did you figure it out? I just tell you a bunch of data. So the best thing to do for CIOs is to not be so pragmatic and wait till this wave goes, uh, and do some experiments.
Set up a structure where you can take some chances. Uh, you know, I, we call it credo typing. We used to do that in, in D Mware, m i and Zoom. A lot of folks are doing it. Try something, see where it goes, and then, [00:19:00] you know, come up with those use cases that makes relevance to your business rather than, Doing it into like an E R P pro.
This is not an E R P project. This is, you need to change your culture structure to be more extraordinary. That's one. The second is I'm thinking our CIOs, the right people to lead it anyway. They may be a few people, two CIOs and CPUs who have the charter and the vision to lead it, but I think your business leaders are gonna take it oath.
The marketing folks are gonna [00:19:30] say, why am I having a call center? Set up overseas doing accent training to save a few dollars. Why not can do it intelligent. Right. And, and I don't know if you guys saw the viral video from Tesla, uh, on their call centers, where, you know, the roof guard is calling, it looks like a normal person asking you about their experience and what cars they're buying and kind of leading them, ending up with, uh, scheduling a test drive.
Right. And uh, That is done completely. And, and it [00:20:00] looks to me like this person doesn't know he is talking a robot, right? I mean, so that, those are the kind of things we should expect and should make them faile failures quickly and then move on. So that, that's, that's the challenge is how many of the CIOs and CTOs are taking the risks and, and b, they have to stop some of the products or e r p products and, and, you know, sales products and other kind of products.
They have to, they have to stop because, You don't know where the future is headed. There's no point trying to do something big, uh, [00:20:30] enterprise project right now and missing out the media. So ba
[00:20:35] Vijay Ganesan: if you tell a, a data leader specifically that. You have to start somewhere. Uh, there's often questions around data quality.
We don't have our data quality story and governance and all that stuff in place. And once we have that, then we'll start these new initiatives. What would you say to that, those folks?
[00:20:52] Bask Iyer: You know, I can be very blunt. I mean, that is gonna be, uh, a little bit more polished, so wait for his answer as well. But my thing is, [00:21:00] every company you go, you know, you hire somebody and say, Hey, we're not able to get.
Quality data on customer. We don't know which salesman is selling where, and they start out with our massive data quality is terrible. So they go into a process of let's clean a product. Master customer master you art becomes a hundred year product. And then they clean that project and then the person meets the next, uh, data officer comes in and they start the same project.
You know, let's clean up the data. Sales people or business people are dying, but [00:21:30] they are making decisions with, with spreadsheets and whatever information they could grid or download certain information, uh, they, they can't link that, right? So the first thing I would do is do some, this is not a multi-year product.
And second is do not throw every term that comes out as a science project. So it gives me data reporting, you know, I need a sales report, who's my top salesman, then became a data management project. And then it became a data science project, then a AI ML [00:22:00] product, and now it's gonna be a generative AI product.
And the request from business is still, I need to report to know who, what best sales people are, or where can I sell? Where can I put, so I, I'm worried that we will now start rebranding our data writers as AI ML people, or generative AI ML people. And the focus should be on. Work with the data you have.
There are some nice tools to, to clean them. I can work with [00:22:30] garbage data, but don't you know, wait for this thing to be so perfect and then start your project because the companies are losing patients with that condition of data in the LAR company. Now, for the real scientific well thought of logical answer and politically correct answer, let me turn it over to Avinash.
[00:22:48] Awinash Sinha: Yeah, I, I'm agreeing with you in the principle, uh, bas if you look at, it's true that there's no perfect state for data. Also, if you look at it, I think the practical approach that you're talking [00:23:00] about, trying to make best out of what we have today and incrementally improve, I have a great proponent I've seen in industry do things.
If you look at B two C data and B two B data, there's a distinct, uh, characteristics, differences between these two in terms of the data quality. Particularly for B two B data enterprise businesses. If you look at it, the higher order of, uh, the, the stack they are in terms of the size of the organizations and things like that, you can enrich that [00:23:30] data much better with a third party, uh, companies that exist in industry today.
But when we go lower down the order, uh, the enrichment, et cetera, is, is, is less, less, uh, effective. Uh, whereas from a truly, from a business sense perspective, if you look at an enterprise market, uh, the top, most of the company make most of the money from the top two segment, top one segment, top thousand, top 500 accounts, right?
And there, the, while there is a whole ocean [00:24:00] cleanup is one problem. Big problem. We can dissect the problem from keeping business lens in mind. I think that's a very practical approach. Prototype you discussed in other contexts is, applies here too, right? Uh, third party data. You can do machine learning, every data cleaning through machine learning, like identifying the names, which are, looks like you're matching with transactions.
Now, there is a technology about, uh, graph database, which are, you may be familiar with that also very deeply. You can look at the, which, [00:24:30] which exact customer we interacted more and gives more weight onto. Those customers and ranking and ordering and cleansing and duping and all that stuff. So the point here is that connecting technology to a business outcome, keeping an eye on that and taking an incremental approach, and there is no perfect estate, no company of certain size and bigger can say that despite all this investments.
So I'm pretty much aligned with you. And then the, the, depending on industry budget words that's [00:25:00] going on at the moment, and the roles get the same set of skillset. The smartest set of people will move from a one job class to another job class, and all that is stuff, and those will be the leaders in that.
So we'll look for it. The key thing is, I call it like outcome enablers, identifying top talent where outcome enablers will deploy whatever the right technology is. And those are the people who do well in their career, make impact in the company, and then move up in the career ladder. Great. I love the Tom
[00:25:29] Vijay Ganesan: outcome
[00:25:29] Awinash Sinha: [00:25:30] enablers.
We should advertise as as job outcome enabler.
[00:25:35] Bask Iyer: Yeah, that's a great direct one. Or do something. My job description is, please do so. Do something.
[00:25:43] Awinash Sinha: I like Nike logo. Right. Just do it. You know, I, I think
[00:25:48] Bask Iyer: the law of averages says there are. There are only so many AI specialists in the world. I mean, every one of us are working on ai. I'm gonna say a thousand people who are [00:26:00] really doing bad groundbreaking work in ai. And the big companies and startup companies such as yourselves are gonna grab them, right?
So if you look at majority of the companies, it'll be almost impossible to hire. Really groundbreaking, uh, ai. Um, but what they, you can hire is the outcome enabler. People who look at, have intelligence to say, yep, I know this is what is happening in different companies. Let me have the partnership with the big Blos, the [00:26:30] Globals, um, am Amazon, et cetera, and also the emerging companies, and do a few experiments that makes relevance to my business.
If you wait for mid spring other companies to become, you know, a multi-billion dollar company with millions of customers and so on. You are not gonna get the tension of the company that you can get right now, and you can, you can kind of develop the product. You can, you can, you know, I worked very successfully, even with Zoom, it looks like a giant company, [00:27:00] but there have been times when we were able to quite feedback to the company, take chances with a company and develop with a, morph it into a product that makes sense to the enterprise.
And so, so one of the things I'm telling people is there's so many startups, so many companies, early stage companies, Who have really clever people that large companies may not be able to hire for, for various reasons, not to stop patient. They're dying to work with you. So just do that. I mean, you don't have to figure it out.
Go and give the problem statement to them and [00:27:30] say, Hey, can you, if you have a data problem, go give it to a few people and see if they can figure it out, and they let the best person win.
[00:27:36] Vijay Ganesan: So, Avina, switching gears a little bit, let's. Talk about the modern data stack, big data specifically we're seeing the emergence of cloud data warehouses and data lakes.
They're becoming the central repository of data. And you know, this idea of single source of truth for all enterprise data that's been talked about for years and years, but it's probably becoming more of a reality now with, uh, with these modern cloud data [00:28:00] warehouses. Um, and so obviously a lot of advantages in terms of.
Security, consistency, manageability, single source of truth, no copies, you know, privacy and all that kind of stuff. What, what are you seeing in this trend towards warehouse centricity and what would you advise data leaders and how should
[00:28:15] Awinash Sinha: they be thinking about this approach? That's true. Vi I actually, uh, like if you look at any new new generation company about cloud companies, they're straight going to this architecture.
They're, they're, they're putting their data warehousing. Sort of lake on [00:28:30] the one of the hyperscalers. Now there's a lot of advantage for this. Uh, if you look at industry, take a step back 10, 15, 20 years or so. Uh, first thing was migration from on-prem to cloud was happening. All transactional applications or science applications, any company, any founders like yourself, you, you, you're gonna create an idea.
You're gonna straight go to cloud, one of the hyperscalers, and you're gonna build on that. Same thing, CIOs, CTOs, if they're coming up with a new project or new new asset they wanna build, they'll go on cloud. [00:29:00] However, the legacy companies, um, are still in the journey of migrating from on-prem to cloud, even on transactional applications.
But the advantage on data warehousing, being on cloud is on many manyfold, any large companies. If you're not on a cloud, then a scaling becomes a problem In data warehousing, long running job. Problem sales leader asking where are we landed? Are we landed on target, we committed to street or not? And then finishing all the jobs, uh, who are [00:29:30] down into we, and looking into whether in 24 hours can we finish all the jobs, whether for sales leader or for finance leaders and things like that.
Those are things of the past in the cloud. So in the past people, what they did to solve that scalability problem, they fragmented the data, they created the data marts, and then you have a different problem, latency problem because these are all these different islands now. Now we talk about AI ml, if all the data is together, not only is the latency is gone, you can write machine learning model and more [00:30:00] parameters and input, more accurate analysis you can do versus if you can have a very little less set of dataset in your own data mods.
So it has a tremendous value. In that sense, and I see that as more and more cloud adoption happens on, for data warehousing, either on directly onto hyperscalers, any of the top hyper top three hyperscalers have their own certain core analytics solutions. But then there is, market also has some couple of really [00:30:30] large, on top of that stack, uh, custom built purpose built data warehousing without naming any particular.
Partners in that space and we work with all of them. Uh, at Zoom. This trend is powerful and I see this will continue to happen. It'll foster also innovations. If you think of you have to execute a lot of AI ML model in a on-prem solutions. So what would've happened, the CIOs and CTOs have to review the security and you have to ship a copy of data to that startup.
Now, a startup [00:31:00] can plug and play in the same way. You just need to give access. It's logical separations, right? On the same. So power of innovation, uh, gonna be quite high. And where we are with all AI ml we talked about in the front of this call, that will work better on, on data warehousing, on being the cloud.
Now, couple of things, I I project forward, I see conversions happening. So when, when you have the, and we [00:31:30] already experimenting to.
So for us, we have hyperscalers where our transactional applications, a few of them are hosted now. We have both one of the hyperscalers, two of the hyperscalers and also custom built or purpose built the warehousing cloud. And then we are thinking that, hey, these were not two different island. You remember in the past, you transact here and you report there.
If you look at consumer applications, [00:32:00] where is transaction? Where is analytics are intertwined? It's a smart devices, a smart apps and all why enterprises are not smart. They're still talking on gonna transact here and then you're gonna report over there, right? The cloud native architecture can bridge this gap.
So this means I should be able to do that together and we are experimenting on this is totally doable. So great promises.[00:32:30]
I mean, I can keep on talking about the benefits, right? So we are, data marketplace is evolving because whoever the business partners we do business with, their data is also on the same, same cloud or things. So we can share the data against each other without copying the data, right? So that has another advantage.
And my forward looking view on these things is that eventually it'll take time in industry. Everybody initially will go on, uh, on uh, generative ai trying to do it their own, but market [00:33:00] consolidation will happen and some sort of market model marketplace will evolve. But that's little far down. Plus is the data marketplace.
And then few years down the road, uh, machine learning or l l m model, marketplace will evolve where you keep your data. My IP is just on the model and we can operate across this. That's very
[00:33:21] Vijay Ganesan: interesting. We've talked about data marketplaces quite a bit, and application marketplaces, you're talking about like L l M marketplaces, model marketplaces.
That's. [00:33:30] That's very interesting. You talked about experience, how customer experience is so core, uh, you know, customer delight that is so core to success of Zoom and
[00:33:39] Awinash Sinha: Apple. I mean, obviously
[00:33:41] Vijay Ganesan: to deliver on that kind of rate experience, you have to have a lot of analytics. You have to have a really, really good pulse on every interaction that you have with the customer, both in product and out of product.
And so data analytics is a powerful weapon for. Particularly for
[00:33:56] Awinash Sinha: product led companies where you have to
[00:33:59] Vijay Ganesan: have a really, really deep [00:34:00] understanding of usage behavior to optimize metrics around conversion, engagement, retention and things like that. And,
[00:34:06] Awinash Sinha: and I'm sure Zoom's doing
[00:34:07] Vijay Ganesan: a lot in this area. What's your advice for analytics leaders in product led companies and what should they do to achieve the same level of customer delight?
Customer experience that,
[00:34:21] Awinash Sinha: uh, zoom has been able to achieve? I think Vij, you touched a little bit, uh, around, uh, customer experience and we're a subscription business. And, uh, in [00:34:30] subscription business you have to really closely follow the lifecycle of customers. And it's a infinity loop, right? Even existing customers, you may be looking at adoption, you may be looking at churn, you.
We can attach, create more value, make them aware about the new products and offering coming together and internally for the sales department, marketing department, or customer support department, providing the insight, both coming from product basically to this cloud [00:35:00] architecture. You have an, we have an ability to look at product telemetry data, as well as business transaction data and intersecting them insights.
When we intersect this two data, join these two data and do a cohort analysis. Cohort analysis at product level, at segment level, particularly for large enterprise cus companies our size and bigger. We'll have customer segmentation we'll also, we'll have some flavor so in geo's, [00:35:30] dimensions or other dimensions once we connect the data around product.
And then customer lifecycle from business transactions. Be slice and D from many different perspective, and that's where the insights comes, but it goes one step further. What the new architecture allows is sometimes those insights doesn't have to recite just in a reports. It could be fed through Aled approach into transacting application.
In the past, past reporting [00:36:00] of bi traditional BI analytics, you can't even update a single record from your, you can only read it's a. Ghana has wrote this. Now these are bi-directional. You can take action in the platform where the action is supposed to be taken. And that's where I was talking about the convergence piece.
Particularly. This convergence is needed in the, in the demand gen to cash lifecycle, uh, for the businesses. That's where the money is. That's where most of the value creation for the internally is [00:36:30] externally is from a customer experience perspective, whether I'm marketing customers. Whether we are sales, post sales experiences being delivered in product experiences being delivered, right?
All this capturing in a very digitized way. Very
[00:36:45] Vijay Ganesan: interesting what you said about the real insight is the intersection of product data and business data, right? And then the ability
[00:36:53] Awinash Sinha: to feed that back is, is very critical.
[00:36:56] Bask Iyer: So I think with subscription model that real time becomes [00:37:00] critical. There's no time for copying data, processing data, fixing the, I mean, you should do that, whether you do it all at at real time speed, more or less.
Uh, and then also, you know, you're migrating from tools. Now, you're not dependent, I think I said that is you're not locked into one tool. Sometimes in enterprises, you're locked into one data warehouse or tool, because it's impossible. In a system that is running to migrate to another tool, it's just impossible because by the time you do the migration and move the data, the data has been updated in the production system and [00:37:30] you don't have time.
So a lot of people just sit with legacy systems because they cannot migrate. Right. So I think, I think use real time processing, doing it cleverly with clever architecture, not copying data, all this stuff. And there are a lot of interesting companies that are doing all this and a ton of companies doing in consumer space.
Right? I, I bet Netflix is following me and saying, Hey, this guy is not clicking on this content for the last week. He's going to, to pry it. Maybe it's time to refresh the content or show something [00:38:00] else for him, because I'm gonna drop the subscription right, and come back two words later. It's not like, it's not a, not a fight, it's just that I don't see anything you, so let me get outta here and come back when you have something interesting to show.
So the, I think the ability to track all those in real time subscription model become extremely critical. Um, so I think, I think that is the, that is the urgency
[00:38:20] Awinash Sinha: for data. Hey,
I'm personally inspired by looking at some of the innovative [00:38:30] thing in consumer space and then trying to bring to enterprise. Enterprises are slow moving things and all that stuff, but disruptive work happened in consumer space. Even there, like I talked earlier about convergence of transactional and analytics has already done in any single sticky consumer applications or interfaces.
It's already done. Enterprises will take several years to get there where you have a seamless access to this and insights and action together. Right? But that's one other thing is to [00:39:00] look at where to look for these things. I
[00:39:02] Bask Iyer: know how we, we talked about this. There's, there's no such thing as B two B in my opinion, because it's always B two C because the consumer, I imagine, I mean, I am now looking at all the bad systems I deployed now as a my own company.
Uh, I'm looking at. Working with all those large corporations with their back office building and invoicing and processing, and I'm thinking how, how bad these are, right? So, so I'm looking at modern systems by consumer systems, but I'm looking at [00:39:30] these systems, logging onto their legacy, whatever, e r p, whatever systems to just get paid or invoice.
I'm thinking I'm frustrated, right? Because I'm a consumer, even though I'm a B two B person. How long does it take to upload an annual invoice? What is it? What is going on here? Right? So I think, I think the less we consider ourselves, all of us are consumer business. And you look at it that way because a consumer is using it and the consumer is using consumer tools every day.
And there is, and and the younger ones, the next generation of folks. I'm gonna [00:40:00] be confused, you know, it's like I'm using an iPhone to do all my work, work by, go to a company and they gimme a blackberry or something like that. Right? It's like you, you can't, it just, it's all, it's all experience. The Netflix, the Gameses, the Tesla, all these guys have set a standard for experience and you cannot give any excuses.
You can't say, I'm not A, B, B, C, you know, I don't have resources. I don't have the funding, I don't have the service. Guess what? That's what they're all expecting an iPhone like experience, but they go use your
[00:40:28] Awinash Sinha: pocket. It's very [00:40:30] true. The decision makers are now coming from a generations. Of the book, but flip the swipe.
I think
[00:40:46] Thomas Dong: what you guys have demonstrated here is that you're CIOs who are very connected into the business requirements, what the business users need, and, and so I thought maybe we could end this conversation and discussion with you each sharing a call to action [00:41:00] for data leaders and let's leverage your 40 years plus of combined experience as CIOs and, and leave our audience with, uh, some actionable, uh, insights they can take, uh, into their own business.
Let's start with you, Basque.
[00:41:14] Bask Iyer: Let me start off, give a few things. Uh, first is, um, focus on on business outcomes. I think Avinash said that is, and, and deliver something. You know, it's always good to take a few singles and you get a home run. So don't start with, let me [00:41:30] do the ai ml chat, g p t, whatever. If it, it'll happen, do it quietly.
Do the experiment, but start with delivery some value today. Uh, like in 30 days. Then start doing some experimentation on all these new systems today. And it is subscription models, so your risk are very low. If you don't like the company, you pay whatever you can get out next month. So it's not like how we used to do with huge capital budgets and so on.
So that is what, don't be too pragmatic to do, too dogmatic. [00:42:00] You know, don't, don't be, um, don't, don't be afraid. Take a few risks. Uh, the second thing I would say is, um, perhaps. It's time to break up this it and data it decent. We can keep talking about decentralizing data. So the data architects data chief should look at more like vendors rather than command and control.
You know, it's like you gimme all the data and any data comes to me and I control the warehouse. Seems to be one approach that has the, just enable the freedom so that the sales folks can do [00:42:30] what they want to do. Don't try to solve all the problems for them, just fundamentally create some bridges for them that they could do their jobs easily.
And, you know, take care of information, security, access, whatever, but just decentralize that. Have decentralized data chiefs. Every department needs a data expert. So I think it's to look at the structure, uh, and, and divide that. And the third thing I would say is, uh, the organization boundaries are getting into the way a lot of enterprises, and it's largely because we, we did too many C-suites, suite jobs.
You know, [00:43:00] previously there was one person you go to get your IT or technology. Used to be a C I O or c t or we used, I mean, I used to be called the head of IG or Geek or something. It wasn't even the title. The minute you create this officer position, it creates between a cso, a ct, o a, chief Data Officer, chief, c I O, and you're fighting internal battles and the businesses are just gonna go do whatever the heck they want.
So perhaps structurally we have to kind of try to come bring it together. I mean, you don't have five CFOs reporting [00:43:30] on finances of the company. You don't have 10 HR people. You don't want a c e O, you don't want vi to be talking to eight people to figure out what to do with this strategy on, on data. So I think it's perhaps trying to, um, look at it either collaboratively, work together or ally work together because a lot of companies, that becomes a problematic and political environment.
That's fascinating. Alright, Ash.
[00:43:55] Awinash Sinha: On the same question, I share some of the viewpoint, uh, that bass said and I mentioned [00:44:00] upfront. And I very personally, well, uh, big believer into outcome led technology solve. So what problem are we trying to solve? What are the company level top objectives and goals are? And then figure out what's the right technology is to go and solve.
You should not be on a small experimental basis. We can see that, hey, where can we use lmm, uh, for solving problems or experimenting? Big betts are always business outcome to technology driven versus technology to business outcome [00:44:30] driven. So that that is always keeping in mind, and not just for CDOs or CIOs, but cascading that mindset down in the organization is very important.
If you wanna build an organization that is outcome led, it shouldn't be just with a leader's job. Every engineer's. Product manager should have a line of thinking between what they're doing, how the top level outcome enabling. That's internally in management terms is also about org alignment. [00:45:00] Right?
That's one thing. Um, second thing is, uh, The federated approach, uh, kind of a different way of what was said about that. Freeing up the data. There's industry term about freeing up the data. You look at the KPIs, if you look at most of the large industry, whether the sales or marketing and all, there is a standard set of KPIs that everybody tracks should be tracking.
There are some unique, uh, KPIs for that company because each large successful company will have some [00:45:30] unique value prop. What is the unique thing with that company? The problem is everybody starts from a scratch. If you have to run the sales, there is standard metrics for sales, a standard metrics for finance, or a standard ministry for retention in hr, legal, the companies reinvent the wheel a lot.
We becomes bottom up reporting that, Hey, I need this report, this field, and things like that. Why don't we align on the corporate objective and start from what are the right KPIs? Our leaders are being measured on how we measuring the health of the business. [00:46:00] And then go down from this like inverted thinking.
Right now we have and it's the bottom up, the report building, the demand coming. Hey now executive expert. Hey, this doesn't number match. We start top down. That way it'll be much faster and we don't have in many, look at any large company. We'll have a report, proliferations. Because of the bottom of thinking there.
Sometimes top down thinkings makes more sense. Uh, in this case, a standardizing on K P I I think is a low hanging fruit in industry, [00:46:30] and then only 10 to 20% of the variations you can do that based on the unique value prop that your company's doing are unique customer situations that you have. Um, these two things I will say.
And the third piece is, uh, data savviness. Uh, is everybody needs. These tests go forward at this. As ai ML becomes, uh, mainstream, even for business decision maker, ideally, a good business decision makers are balances between the qualitative judgment call and quantitative [00:47:00] data set to balance this out.
Right? And, uh, equivalent of this in data is coming as a machine learning on one side, and qualitative input is through the generative ai and these two will will converge. So those are the couple of advice. Outcome driven, inverted KPI led, LED driven analytics, and, uh, analytics driving the business, bottom ups like reporting, driving the bottom ups, and innovation, culture, [00:47:30] prototyping, et cetera.
Past,
[00:47:34] Vijay Ganesan: I think the biggest thing for me was of two things. One, uh, This idea of customer experience, you know, obsessing about customer experience, uh, whether it's Zoom or Apple, some of these super successful brands have been successful, primarily because they've obsessed about experience. Every little detail of the application, the, the device, there's been obsessiveness.
And I think that's increasingly [00:48:00] important because like what these guys were saying, you know, the same kind of expectation is now set in enterprise two. What people are used to in consumer products. There's a whole generation that, that, that's expecting that in every product, whether it's consumer or enterprise.
And so, so that's puts a really high bar and everybody should be thinking of. Experience the same way as Apple does or Zoom
[00:48:23] Thomas Dong: does. Yeah, and I think that mindset of finding innovation and kind of crossing over from B two C to B two [00:48:30] B, you know, the convergence is, is, is obviously there. My other takeaway was I.
This idea that, um, you know, you can never solve data quality. You're always gonna be chasing data quality. And in this subscription economy that kind of makes it very real, right? The risk is low to try new technologies and, you know, we didn't like as a C I o we for, you know, 20 plus years, you know, it was a very different world 20 years ago when you're buying on-prem software.
It was a, a major [00:49:00] investment. You're making it. Big bet on a vendor and you're building entire teams and organizations, uh, around that. Uh, but in the subscription economy with SaaS, uh, technologies, you know, a a startup betting on a startup isn't a massive risk, and that's where you're gonna get your, uh, kind of, kind of the, your biggest opportunities, uh, for a step function in your business.
And I thought that was a very valuable advice, um, to CIOs who are. Looking at new technologies, especially as new initiatives [00:49:30] arise. I think generative AI is, you know, perfect example of that right now. Like everybody's gotta make an investment in there. Everybody in who's in it is a startup. Uh, but it's very low risk to, to stand up new initiatives around.
[00:49:44] Vijay Ganesan: Perfect is the enemy of good. Right. You know, get started somewhere. Don't work. Wait for the perfect solution. And, and the cloud architectures make it much easier that you can plop in different analytics tools that work off your central data warehouse and the risk is low. You can give it a try, and that it's not a very heavy lift like some of the larger enterprise projects. So, so that was great insight. The other other thing around insights and analytics, what. The key insights come from the intersection of product data and business data, and that's obviously very dear to what we talk about at Net springing. It was great to see that. Reinforced by him that you know, if you, if you, that the, the high value insights do not come from just the product data. It comes when you intersect that product data with business data, and that's where you can make business impactful analytics work.
[00:50:38] Thomas Dong: Thanks so much for joining us today, Bask and Awinash. It was a real pleasure chatting with you both. That concludes today's show. Thank you for joining us, and feel free to reach out to Vijay or I on LinkedIn or Twitter with any questions or suggested topics for the future. So until next time, goodbye.