IT Automation Hits an Inflection Point

IT leaders are asked to do more with less, manage mushrooming complexity and steadily steer their business into the future. Increasingly they’re relying on artificial intelligence and machine learning to run their data centers and experts say there’s no turning back.

By Jason Lopez

By Jason Lopez April 1, 2021

Automation is moving into more aspects of daily life, especially for IT pros. Not long ago expectations overshot real capabilities, but increasingly artificial intelligence (AI) and digital automation technologies are making steady strides to help manage enterprise IT operations. 

“Initially there were lots of magic wand kind of expectations,” said Rahul Kelkar, Global Head of Product Management at Digitate, a maker of autonomous enterprise software. “They have now matured into use cases that deliver measurable benefits.”

In a tech Barometer podcast interview, Kelkar and Tarak Parekh, director of product management at Nutanix, explain AI and automation trends, how these technologies are evolving and the kinds of benefits they’re bringing to businesses and organizations.

“A few years ago our customers were just content with manual reporting and some scripting or some very basic form of automation,” said Parekh of hybrid and multi cloud software company, Nutanix. But new capabilities and the need to remain innovative have “increased the appetite for automation.”

Parekh attributed the acceleration of IT automation to the global COVID-19 lockdown. 

“What this pandemic has done, essentially, has sped up by a factor of 10 the digital transformation that was already underway,” he said. 

Some companies, Kelkar added, squeezed their AI and automation roadmap for the next 10 years into 2020. Restaurants and retail firms were notably aggressive in adoption. “An example that kind of comes to mind is one of our customers,” said Parekh, referring to a global retailer which has added ways to make it easier for buyers to purchase products. 

“What they've also done is they've extended automation to a large portion of their IT infrastructure and logistics,” he said. “They've automated the entire physical infrastructure. The entire IP stack that took them maybe three months to put up, through automation, now, they do that entire process in eight hours, which was just unheard of.”

RELATED

Journey to Cloud: Step 6 - Augment IT Skillsets with AI and ML

AI and automation is proving to be practical and reliable enough to power more aspects of enterprise IT systems. Kelkar sees a permanent shift across industries powered by data, where IT automation will spread like the mass adoption of cloud computing and change expectations and productivity. 

“The old way of doing things through workflows and processes is going to be passé,” Kelkar said.

 

RELATED

Database Management Automation Brings Huge ROI

Transcript (unedited):

Rahul Kelkar: This adoption is also not straightforward. You cannot just say that, "okay, tomorrow morning I am AI." It doesn't work that way. It requires significant organizational change.

Tarak Parekh: The automation and the AI are going to help me do my job better and help me deliver more value to the businesses and the partners and the clients that I get to work with.

Rahul Kelkar: That's where things start becoming interesting and it needs to be done in the context of a specific problem.

Jason Lopez: This is the Tech Barometer Podcast, I'm Jason Lopez. The complexity of running data centers and doing IT tasks is only increasing, but the promise of AI is to empower IT teams to handle many tasks, which are becoming humanly impossible. On the line to chat about this in sort of a free-form conversation is Rahul Kelkar, the Global Head of Product Management at Digitate. He's been building products and tools in automation, machine learning, and AI for the last 10 years. And joining him is Tarak Parekh, Director of Product Management at Nutanix. He's been building enterprise software for the past 20 years, and currently he leads the manageability and API charter at Nutanix. We started the chat off by talking about one of the obvious issues in IT this past year, the effects of the pandemic, especially with regards to AI and ML adoption. Tarak starts us off here.

Tarak Parekh: Our customers and enterprises were already looking to convert or transform their businesses to be more automated, to deliver more value, to be more agile. And what this pandemic has done, essentially, has sped up by a factor of 10 the digital transformation that was already underway. The way the pandemic has affected us is it's proven to be an opportunity for a lot of companies who did their entire digital transformation they were looking to do in the next, let's say five to six years, and they just fast forwarded all of that into this particular year.

Rahul Kelkar: Yeah, I think I agree with that. So I think, as you said, enterprises were sort of forced rushed into this operating model. They had no choice but to operate and ensure availability of their IT and business systems. Of course, all of this was being done by using whatever limited hands on the deck everybody had. Now we have sort of settled into a new working model where there is a systematic solution that is being put into place to handle this situation. If I may say, for the foreseeable future, it is likely to be this way.

Jason Lopez: Yeah, well, it seems like a lot of what we thought of before COVID is a distant memory to us. I want to touch just a little more on the idea of what was happening before COVID in terms of digital adoption. So, Rahul, what was going on from your point of view before the pandemic hit?

Rahul Kelkar: As Tarak was saying there was this transformation that was ongoing. Everybody was looking at the future and trying to embrace the digital world. What that transformation would have taken, if I can use the term natural evolution scale, it would have taken 10 years. What COVID did was it sort of became a catalyst to sharpen that transformation cycle by maybe a factor of 10, is how I look at it.

Jason Lopez: Wow. So what you're saying is that we're seeing in this past year, 10 years worth of roadmaps and plans that were all compressed into 2020, pretty much?

Rahul Kelkar: That's what I feel.

Jason Lopez: Wow.

Tarak Parekh: Yeah, you know, or naturally with the arrival of Amazon, actually, a lot of the existing brick and mortar businesses such as Walmart and target have had to digitized to essentially provide a way for you to place an online order and go and pick up the order in the store, for example, and then suddenly the pandemic hit. And a small thing that Target added, for example, they added the whole thing about curbside pickup, right? And that feature kind of showed up within two to three months of the pandemic hitting. And that to me was actually pretty surprising because something like that, when you're completely delivering a new channel, a new way of delivery. It would have taken maybe a year or two year to essentially integrate with their app, with their website and have that whole logistic setup, done to adapt to what the environment is. And they were able to do that in like three months of the pandemic hitting. Now that, to your point about how fast digital adoption has happened, something that would have taken maybe 12 to 18 months, just happened in three months. Just the acceleration of what they could deliver this quickly, I don't think would have happened if the digital adoption that was already under way was not there, and they had the drive to accelerate it and take advantage of the current situation to deliver these kinds of features that build more business.

Jason Lopez: And what's interesting is as much as we see this as consumers. "Oh great, We can just go pick up today, curbside pickup." What we don't see is that there's automation behind the scenes there, and it's not just because of COVID, there's kind of a trend there.

Tarak Parekh: From a trend standpoint, given the kind of business environment we live in, I don't think we have a choice but to automate and utilize various forms of AI to react to these various business events. I mean, just as an example, a few years ago, our customers were just content with manual reporting and some scripting or some very basic form of automation. But as we all know, there've been increased demands on IT. And there's also a lot of competitive pressures to stay ahead of the pack. So this has increased not just the awareness of what the IT team and our customers need to stay ahead of the pack but it's also increased the appetite for automation. And now what we are seeing in addition to the fact that there was a hurdle to show the value of what automation and AI could bring, combined with the fact that this automation kind of let the fear in our customers' minds that automation is probably going to take away my job for example. That level of fear has subsided. We are seeing more and more acceptance towards automation. People are looking to utilize data and apply automation on top of it to essentially gain more insights. What they figured out, the automation and the AI going to help me do my job better and help me deliver more value to the businesses and the partners and the clients that I get to work with.

Jason Lopez: Right. Well, I wonder if one of you could speak to that idea of this ability to understand more and more data. We hear this all the time, big data, but what are the benefits of AI and machine learning in the IT space when it comes to being able to handle this deluge of data?

Rahul Kelkar: I think this is almost if I can use the analogy of a perfect storm, right? The whole world is hyper instrumented. There is all sorts of signals coming from IT systems, [OT] systems, several other things, right? Even consumer behavior is getting logged. So there is a super instrumental hyper instrumented world, which is producing tons of data. All the systems have sort of now become more manageable. Like there is an app for everything, for example, you can easily remotely manage everything. And then of course there’s an advent in computing that are hyper-connected systems, and there is overall democratization of these algorithms in this AI machine learning space, which is a great thing. So a combination of these three creates this, if you will, a perfect storm. And that's what is actually, if you will, this tremendous adoption, whether it comes to individuals or enterprises, is what I feel.

Jason Lopez: Tarak, what do you think of this question about data? I'm not trying to be too open-ended here, but what do you think about that?

Tarak Parekh: I think the question is really good and it's very relevant to this day and age, right? Rahul, you just put it fairly well, too. The only color I'll add is using the cliche that a lot of people use, that data is the new oil, and yes, there's lots of data. The way we use it, the way we essentially have the right level of ethics incorporated in it. And the way it actually delivers consumable value for all of us is going to be key in terms of how this vast amount of data that has been collected is going to be utilized, and how far we go ahead with it.

Jason Lopez: Well, thank you for that insight Tarak, because it's always interesting what technologists think about where we're going, we're not just hurtling blindly, but that there are people thinking thoughtfully about these things behind the scenes. So let's go back to where we were with the adoption question. I just want to find out a little more about some of the real-world things, just put some color on what's really happening on the ground right now.

Rahul Kelkar: Sure, let me take the example of retail enterprises, because retail is a very critical business, as all of us have learned in the last 12 months. So it needs to be up and running even during lockdowns, to put it simply. Autonomously managing physical stores, digital stores, is super critical for all retailers. A combination of these two is what is really required in all sorts of situations. So a solution that sort of holistically enables management of geographically distributed stores by providing top-down visibility into a point-of-sales or checkout or replenishments and promotions or inventory and logistics. And of course, all the rest of the business processes. Managing corresponding applications and devices is very critical. As a result of this complexity and distributed nature of business, the retailers are very, very keen in adopting AI and automation solutions so that they can assure this business on a 24 by 7 scale, both digitally and physically.

Jason Lopez: Tarak, what's your sense of that?

Tarak Parekh: I love the way Rahul put it. But a couple of things that retail has also taken further, I'm going to touch upon the point that Rahul made around physical stores. An example that, kind of comes to mind, is one of our customers who's a big retailer and has got stores across the globe. In addition to the fact that they've looked to delivering cashless purchases or non point of sale kind of purchases, in essence delivering customer value and making it easy for customers to buy products. What they've also done is, they've extended automation to a large portion of their IT infrastructure and logistics too, which is not surprising, but it's something that we don't get to see from the outside. For every "neighborhood store" they kind of put up, what they've done is they've automated the entire physical infrastructure, networking, application deployment, things of that nature, the entire IP stack that took them maybe approximately three months to put up through automation. Now they do that entire process in eight hours, which was just unheard of. So again, I think retail has been at the forefront of automation and AI and a combination of both in their processes and products.

Jason Lopez: Right? Well, I have a question for both of you, and this comes from just having talked to a number of technologists over the past year lately. I talked to Wendy Pfeiffer, the CIO of Nutanix, professor Art Langer, who teaches at Columbia University and has written a book on the coming technologies of 5G and AI and IOT and these sorts of things. The one thing that really seems to be causing a lot of anxiety for companies is how much this is going to cost. So I'm wondering what the ROI looks like. And I wonder if you could talk about that.

Rahul Kelkar: This adoption is a journey. As you rightly said that while it will take time, this adoption is also not straightforward. You cannot just say that, "okay, tomorrow morning I am AI." It doesn't work that way. It requires significant organizational change to go along with it so that one can restructure your organization, your team structure, so that the benefits can be maximized. In the short term, the benefits are of two kinds. One is, naturally, increasing the operational efficiency, which leads to more agile operations. Of course, the second one being improved transparency overall for more consistent decision making. So these are sort of short term benefits that start accruing immediately and both of these require significant organizational changes to be made so that you can actually count those. In the mid to long-term. Again, the benefits are of two kinds. Once the enterprise starts combining insights and automation, the whole thing starts leading to more stable IT and directly more stable business systems. So more consistent operations, [low down times. And eventually this starts providing significant business assurance so that the businesses now are free to not only trust IT. And it will always be running to the level of performance that they expect. But if they run into a situation like this, where they need to transform very, very quickly, the IT will be able to respond because of the agility benefits and adapt to the newer business models, if you will, very quickly. So those are the long-term benefits. That's what I have seen.

Jason Lopez: Okay. Well, thank you for that insight. I'd like to now ask you some questions, it's going to sound like we're putting philosophers hats on, but actually not, but in the popular notions of AI, you know, you look at a movie like “2001, A Space Odyssey.” You have this sentient being in a machine and of course they need to create this character, but this is the movies, right? And this is what, what you do in order to create drama. But if this were an actual AI capable of doing much of the same things that HAL's doing, we're really talking about a form of automation. And what I'd like you to do is sort of tie back that popular notion of AI. Perhaps maybe demystify it a little bit and explain what the underpinnings are of the actual technologies that are being implemented going forward.

Rahul Kelkar: Yeah, I think it's a great question, by the way. Initially, if I may say, the idea started posing was primarily can the machines simply do it as a sort of virtual engineer. It started with basic things like, "can I automate a task?" because that's what the humans are doing and that's where the first attempts were made and then quickly it was realized that it is really not adding up because, yeah, you automated the task, but there is this whole procedure that needs to be orchestrated in order to really realize the benefits end to end. So then it's matured into this space of, "can I automate the entire procedure?" which then starts getting into this realm of, "oh the procedure needs to make decisions because it needs to handle one, two, three, four, five situations." So that's where it started becoming the maturity started going up on the automation side. On the other axis, which is the intelligence or analytics as people call it. It first started with just slicing and dicing of data, everything leaves breadcrumbs. So the first idea that was being done was, "can I aggregate these breadcrumbs or footprint that are left by different technologies and processes into a single source of truth so that then I can run my smart algorithms on it and try and derive some insights from it?" So it started from there, into sort of descriptive analytics, if you will, "can I slice and dice the information and make it more readable for a human?" Then it turned to sort of, "okay. I mean, that's great, but it is not really helping, it just providing transparency. Can it start deriving some more insights automatically?" So from descriptive, it started to get into diagnostics where it was looking at some objectives and trying to figure out, "oh, can I support that objective or not support that objective with evidence based on the data that I have available?" Then it started getting into more like a predictive and prescriptive kind of reality where can I use that data to sort of predict upcoming events because, why aren't we driving insights so that we can eliminate bad things from happening in the future so can I take it to the next level and sort of simulate and predict things? And the ultimate aim is okay, given my prediction and given the fact that I know how I have derived that prediction, can I prescribe a course of action so that it can be mitigated? So there was this sort of dual axis maturity, if you will, on which this whole journey has been done. There are people at various stages in this two dimensional grid.

Jason Lopez: Interesting. That's really fascinating. Tarak, what's your take on this?

Tarak Parekh: Just to build on that: There has been a continuous evolution around this, which I said as Rahul kind of laid out, and the beauty of it is the practical value, that we get to consume as consumers in our day-to-day lives. Right? So for example, whether we know it or not, I think every interaction that we do with a website or a device or even the food that we eat is in some form or manner powered and enabled through automation and AI, for example. So to your comment about “2001, A Space Odyssey,” you know, it's amazing how our movie creators have essentially laid out the future for us and we are kind of living it now. Although we are on the cusp of it, and I don't believe, at least that's my personal opinion, that we're going to get to a 1984 type Orwellian style future. But the good part is we're all living the practical value of this automation in AI. My son sets his alarm on Alexa to a track on Spotify, given some context and all of that just gets automatically set up. And that's just amazing. That's done through automation. We have chatbots for example, and we have all these experiences of the pain of waiting for a customer associate to show up as we kind of listened to their favorite hold music. Suddenly, now you have chatbots which are able to answer the majority of questions that you have. For example, banking business or things of that nature, all of that's powered through automation. So with that, you know, the good part is, you know, it's invisible, it's there. And the practical value is something that we consume in day-to-day life. And, so even though with the AI and automation power, our current present doesn't look at all “2001, A Space Odyssey,” but we are living it now. It's amazing how the practical value that has been delivered and how it's actually enhanced our lives in general, on a day-to-day basis.

Jason Lopez: That's really a great point, because the real impact of AI on our lives is going to be behind the scenes, allowing us to be more human by not having to interact as much with devices, because they will be invisible; really great point. And it kind of reminds me of the way Gary Kasparov in a column once demystified computers when he said that he wasn't beaten by a thinking entity, he was beaten by a toaster, basically, billions of toasters. And it goes to the point that in the end, these are tools.

Rahul Kelkar: Yes, and Jason I think I would just like to add to what you just said, right? The Gary Kasparov thing right. I think it's a great point that you need to be able to take automated actions, handle situations, and on the other axis, make decisions that are just not here and now, but probably predicted and prescribe a course of action. A combination of these two is what actually makes the system super intelligent. I think just taking decision and leaving up to somebody else to enact it is great, but not really brilliant. Whereas if you kind of combine these two, you get into this whole closed loop thing where you are able to autonomously decide, predict, prescribe, and also close that loop by leveraging the smart automation that you have in place. That's where things start becoming interesting, and it needs to be done in the context of a specific problem. And that's where it starts becoming super exciting, frankly.

Jason Lopez: Well, why don't we wrap up with a look at what you think is going to happen down the road? What are we expecting to see in the coming years?

Tarak Parekh: I think, I just wanted to specifically say that we are just heading into a more automated and AI driven world. Certainly there has been more acceptance and adoption of automation, people are less scared of losing their jobs, consumers have become more comfortable with having AI and automation kind of permeate our lanes, companies who are at the forefront of this transformation are certainly going to be and continue building upon the lead that they have. So in general, the pandemic has kind of shown us who is able to adopt this, and who's able to just strive and not just survive in this kind of environment. We also know from our customer base that we are still in the initial phases of this automation/AI-based transformation. So to your earlier comment of whether there’re sentient beings per se, we are not there yet. We are able to understand that these are tools, and this can enhance our lives greatly and can certainly add a lot of value to our daily operations and daily lives. Crawl, walk, run is generally what our customers have kind of adopted and how they're using, or incorporating automation and AI in their various activities. And so it's a great way, or one of the things that's also come to mind is that, in addition to specific activities and specific tasks and specific products and capabilities, it is also a cultural shift per se, just like how IT was a cultural shift in terms of how our customers did business, or basically how enterprises did business. Automation and AI provides that next leap in terms of how businesses are going to do business per se. So with that, you're just going to see more of it, either you adapt and try, otherwise it will be difficult for enterprises to survive.

Rahul Kelkar: I agree, I think that eventual goal of whether it is AI and machine learning and automation is leading to sort of an autonomous business and IT system. Of course this is also positively changing and it has already changed, expectations of we as individuals when it comes to dealing with these technologies. We are now taking a lot more things for granted. Somebody else is doing it for us. So a solution of this kind that auto adapts to changing workloads, changing technology, changing business scenario, requires minimum human intervention to do this. And also has a low possibility of human error is what is ultimately the aim from an enterprise point of view. And this will effectively lead to a closed loop autonomous enterprise which is going to have all of these attributes. Just to draw a parallel, now this the way the world has adopted cloud as a technology, it almost feels criminal if you cannot swipe your credit card and consumer service on demand. So that kind of permanent shift is going to happen once these technologies become mainstream in the enterprise. The old way of doing things through workflows and processes is going to be passé. If I can use that term, is what I feel.

Jason Lopez: Rahul Kelkar is the Global Head of Product Management at Digitate. Tarak Parekh is the Director of Product Management at Nutanix, where he leads the manageability and API charter at the company. This is the Tech Barometer Podcast, I'm Jason Lopez. Tech Barometer is a production of The Forecast. We invite you to listen to other podcasts or read tech stories at theforecastbynutanix.com.

Jason Lopez is executive producer of Tech Barometer, the podcast outlet for The Forecast. He’s the founder of Connected Social Media. Previously, he was executive producer at PodTech and a reporter at NPR.

© 2021 Nutanix, Inc. All rights reserved. For additional legal information, please go here.

Related Articles