The Self-Healing Network | Nitin Kumar at Appnomic

Business systems are becoming as complex as the human body—but what if, like the human body, these systems could self-heal?

Nitin Kumar, of Appnomic, claims his product can do just that, and even prevent problems from happening in the first place.

In this episode of UpTech Report, Nitin describes how his product, called “Heal,” uses artificial intelligence to predict, diagnose, and prevent enterprise level problems autonomously.

How exactly does it work? Nitin offers some real-world examples.

More information:

Nitin Kumar is a seasoned C-level operating executive and management consultant in the TMT sector. His 20+ year professional story has unfolded in the Silicon Valley, the epicenter of innovation, disruption and opportunities. 

Currently the CEO of Appnomic, he has led organizations ranging from start-ups through multi-billion-dollar companies in Regional, National and Global roles. Notable exec roles include HP, Deloitte, PwC, FTI etc. Held P/L (>$1b), led teams (>1000 people) and experienced 90 countries.

Nitin has played a variety of hands-on executive roles ranging from Start-up CEO, Product Management, Chief Growth Officer, Chief Transformation Officer, M&A Integration/Separation Leader, BU Leader and Management Consulting Partner focused on top-line growth across corporate and private equity portfolio companies.

His functional expertise includes Sales, Marketing, Products, Service, Pricing, and Customer Experience driving top-line growth. Nitin has received multiple global awards for innovation and is an internationally recognized thought leader in the areas of business model design, transformation, and M&A.

DISCLAIMER: Below is an AI generated transcript. There could be a few typos but it should be at least 90% accurate. Watch video or listen to the podcast for the full experience!

Nitin Kumar 0:00
Ever since IT professionals have existed and operations have been tasked for identifying a problem quickly and fixing it even faster. Our proposition is that what if the problem didn’t happen at all?

Alexander Ferguson 0:16
Business systems are becoming as complex as the human body. But what if like the human body, these systems could self heal. Nitin Kumar of Appnomic, claims his product can do just that, and even prevent problems from happening in the first place. In this episode of UpTech Report, Nitin describes how his product called heal uses artificial intelligence to predict, diagnose and prevent enterprise level problems autonomously. How exactly does it work? Nitin offers some real world examples. I’m excited to hear more about Appnomic and the journey that it’s been on and for you as well, how are you innovating and growing? To start us off? Help me understand what is economic? If you had to describe it in five seconds? What would you say? So economic

Nitin Kumar 1:02
as the word itself is self explanatory, right? It means automic applications. That’s how app anomic came together. So autonomous applications,

Alexander Ferguson 1:11
simple as that atomic applications. So then this company be started already back in 2008 2009, as a service company, and 2016 it pivoting to more of a product software offering? Correct?

Nitin Kumar 1:26
Correct. And its current configuration. We’ve been around for about three years.

Alexander Ferguson 1:30
And you came on last year, really now being able to ramp this up and move it forward. This is not the first business you’ve led before our division. That’s correct.

Nitin Kumar 1:39
That’s right. I’ve done a number of different divisions and led a number of different intrapreneurs slash entrepreneurial opportunities through my career. So definitely not my first and hopefully not my last.

Alexander Ferguson 1:52
So now you’re here in economic and the opportunity that this organization exists for what is the problem that you see that you’re looking to solve?

Nitin Kumar 2:02
So it’s a very simple proposition that we have, in today’s world, we are migrating from the era of user generated content into machine generated content, data and transfer of data is at galactic scales, okay. And because of that the applications, the human resources, the volume and velocity of data are going to unprecedented growth and the ability to run operations in an autonomous manner is going to be a challenge. So economics value proposition is that we detect, predict and prevent problems, performance problems, particularly before they even happen.

Alexander Ferguson 2:40
I love that simple explanation for specifically, the industry that you’re focused on after is what that you could

Nitin Kumar 2:48
help. So this is a question I get more often than not. And who’s our customer? I’d say everybody, because we’re in the digital era, right? Every product is going digital, every business model is going digital. Having said that, our sweet spot is really industries with high transaction volumes. And with high availability needs, like the four nines, right? Think about banking, think about e commerce, think about telecommunications, think about your Uber and Lyft, where millions of transactions are passing through. And in real time,

Alexander Ferguson 3:16
if you don’t have a whole lot of data, that means that one person could probably just look at it. Okay, I get it. But if you have millions of data points and a large organization where there is no real way to properly manage it, that’s where a system like yours can come in and play a role.

Unknown Speaker 3:31
Yeah, absolutely. And the

Alexander Ferguson 3:34
one thing I remember reading where you would stay to somewhere else, I mean, the future we’re going to is there going to be that much more data points coming in? There’s going to be a lot of IoT devices and etc. So being able to manage that is only going to increase the difficulty. Is that correct? Would you believe? Would you say that? Yeah, absolutely.

Nitin Kumar 3:53
Right. And today, it’s not that you’re running your software on the cloud alone, you’re running it on premises, you’re running it on the edge. Sometimes you don’t even know where your software is running. You have API’s integrated into a whole bunch of different things, right? So for ever since IT professionals have existed and operations have been tasked for identifying a problem quickly and fixing it even faster. Our proposition is that what if the problem didn’t happen at all?

Alexander Ferguson 4:17
So the self healing, let’s dig into that give me a use case, an example of a problem that occurs in a real world type of company that this would solve your your solution.

Nitin Kumar 4:28
Okay, so let’s take a very simple example. Assume it’s Black Friday and your traffic is on an exponential spike. Okay, you have millions of dollars of overprovision hardware just to handle that spike. Now, we can do two things. One is that we can assess the workload and say, Hey, like Alex go and shop because he’s entering the last digit of your credit card, right? Hold this Nitin Kumar guy back because he came in to do a password reset or browse, hold him back by a second. So that alleviates all of the stress on the hardware you can actually compress hardware capacity. Second, if there is a need that workload is going to land on a certain virtual system, we can kick off an autonomous script that will create a new virtual instance and make the workload really land on that particular side. Right. So help feeling self healing happens in two different ways or three different ways.

Alexander Ferguson 5:19
So both prioritizing by user as well as, yeah, by and then workload. So this concept of, okay, let’s spin up a new virtual server to to accommodate the workload, somebody has could already be doing that. But your software can do that without the need of another person monitoring and actually

Nitin Kumar 5:38
mean, that is the proposition, right. And when the workloads hit a certain critical mass, when with the IoT devices coming in, it’s no longer going to be possible for humans to monitor every bit and byte at every given nanosecond. Right. So we’re now entering that era, where automation is a given autonomy position is the new thing.

Alexander Ferguson 5:57
I love that. I love that your simple statement of that? How many then? Is there a threshold that you find when you’re working with companies, like if you’re going over this number of transactions or data points or whatever, then it makes sense to have our type of solution, this type of solution? You know, as I

Nitin Kumar 6:14
said, high transaction volumes. And high workload is generally the sweet spot whether you have extremely high availability, right? I mean, we may not be necessarily suitable for a supply chain system sitting inside an older type of an older type of an economy company where the volumes are high. But the velocity is just not there.

Alexander Ferguson 6:35
Can you dig into a bit more of how the technology works? How is able to detect these types of problems and then solve them?

Nitin Kumar 6:43
Yeah, so let’s take a look at that, right. So in 90% of the cases, or the software that claims to be predictive today, is actually looking at an insert as an issue or a problem after it has happened, right. And they kind of generally pick the signals from either logs and alerts or lower level incidents that are logged. Now, I’m not saying that’s entirely wrong, that’s definitely required to enrich the data set. But the issue or the problem is informed by a myriad of incidents. And these incidents, find their root causes or Genesis in the events in the technology, landscape. And when I say events, there’s probably hundreds and 1000s of them occurring in real time, right, your disk input output, your network bandwidth, your application logs, your server communication protocols, your thread velocity, so we’re watching them in real time, and a machine learning algorithm is going to be looking at them in real time and saying, Oh, this can become a problem on Monday, this can lead to a outage at 2pm. This could lead to a performance degradation at night, and things like that. So you’re actually using that to predict, yes, we grab the alerts, data, logs data and the incident data. But that’s only to enrich our machine learning algorithm, not as a primary input parameter. This

Alexander Ferguson 8:01
company’s not been around since yesterday, it’s you’ve had the data points for a while, would already back from 2008. When I started with these algorithms, that’s when they were initially being developed, or is it more recent that you’ve been putting this together?

Nitin Kumar 8:14
So it’s been a journey, right? Any product is a living being, and it’s a journey. So when the problem really was identified, and the genesis of the company started, machine learning wasn’t even called machine learning. So our executives at that point in time invented a term called application behavior learning. There were no libraries today to import. So everything is generally hand coded. And it’s fairly robust. And that’s why we have some patterns in this area as well. And then a number of change of directions have happened, we went into what is called application performance management, we went in to play around with AI ops. But then when we really looked at our portfolio of patterns and technology and what we have in terms of our assets, it was really a big pivot that we made into self healing. And that pivot was very interesting because we looked at it very holistically we did a capital infusion. But as a financial pivot we brought in some new executives on a talent pivot we changed our brand from a universal catch all product collapse one into now the overarching umbrella call he’ll with pointed propositions here for the cloud he’ll for on prem he’ll for the edge, here for SAP, and very shortly here for serverless, which is a new paradigm coming into the IT operations world.

Alexander Ferguson 9:28
How many data points and stuff are you now managing or customers that you’re able to work with? Right that you’re working with right now that’s coming in.

Nitin Kumar 9:36
We have close to about 50 customers which are growing in number. And the data points that we look at, which is we’re looking at the events, which is about 70,000 KPIs we monitored by minutes. So it’s not a non trivial exercise to kind of watch all that and also impart machine learning also look at it in real time, and enrich the algorithm and teach the software how to learn

Alexander Ferguson 10:00
Do you license or plan a license? The technology is really just that new direct customer relationship that you’re going to have?

Nitin Kumar 10:08
Well, when we said licensing, we don’t license any part of the technology. Now, prior generations of the product had license revenue or licensing, but we have kind of put an end to that for subscription only model now. And we generally track our metric with arr.

Alexander Ferguson 10:24
Gotcha. Do you plan or have current partnerships with other types of platforms or technologies that are also do monitoring or hooking into existing platforms? So

Nitin Kumar 10:33
there’s, there’s, there’s a couple of different things here, right. So we can ingest data from existing APM tools or monitoring tools or infrastructure tools and make them self healing. We also have our agents are leading in Proposition isn’t necessarily that we can go and replace all these tools or stuff like that, right. So we can do that we have our own agents, but our proposition is to really make themselves feeling. So we have an API is integrated into a number of different API’s and software vendors into banking software, into telecommunications software, into service desk tools, and, and a myriad of other alerting tools. So in today’s world, the ecosystem is very important, we recognize that and we absolutely want to expand our ecosystem and continue down our growth path.

Alexander Ferguson 11:19
As smart to say, you don’t have to work with us, you have to use only our stuff. But the fact that you will plug in with other existing

Nitin Kumar 11:27
world, it’s all about how well you play with the others, not how quickly you displace the others. Competition is cooperation today. And we might we believe in monetizing the ecosystem, as opposed to building a big monolithic, enterprise solution that can be catch. All right, it quickly becomes unscalable in today’s world.

Alexander Ferguson 11:45
I appreciate that feedback. So looking forward, then this final question of where do you see the company in five years from now.

Nitin Kumar 11:53
So five years is too long. In today’s world, right? In Silicon Valley, we get disrupted out of the face of the earth in five months, but we have a plan for the future. Right? We are definitely scaling up more customers, we’re looking at new geographies, you’re signing up new partnerships, we’re obviously have big ambitions to sort of exponentially grow our footprint. At the same time. From a technology standpoint, we have serverless computing on our roadmap. We have a number of beta customers, and we’re rolling quickly and that decentralized computing with Blockchain is a reality. So we’re closely looking at that as well. Right. And obviously, with all these 5g and all these technologies coming in, we have a myriad of opportunities that open up for the company to look at dimensions and horizons, which we previously couldn’t look at because of connectivity bandwidth, or technology limitations, those may go away as a few of these things at a critical mass and unfold into becoming more mainstream.

Alexander Ferguson 12:50
Knowing that you can’t necessarily predict five years and technology, but having a plan for you say 5g blockchain and others to be able to incorporate it and utilize that as well as growing your own footprint. I appreciate the vision of then the direction that you’re going, what’s a good place for people to go to to learn and what’s a good first step that they can take.

Nitin Kumar 13:11
I think there’s a couple of three different resources right, if you go to our website, there’s a very high level video on self healing. We have a number of blogs that are written. If you go to app, which is our website, you will find the resources section where we have a number of white papers as well. And lastly, you can look at some of our customer testimonials on the website and see which use cases they use and how they learn to use artificial intelligence and embark on a journey from automation to autonomous nation.

Alexander Ferguson 13:43
Be sure to check out part two of our conversation with nitin in which he offers some key insights in the thinking of innovation holistically.



YouTube | LinkedIn | Twitter| Podcast

Technology Isn’t Enough | Stephen Bulfer at Streamloan

Letting Go of the Old | Nitin Kumar at Appnomic