Insights from an interview with Sean Mcdermott and Phil Tee on the AIOps Evolution Podcast
Sean McDermott, CEO of Windward Consulting sits down with Phil Tee, CEO and Co-Founder of Moogsoft to discuss the role of Artificial Intelligence for IT Operations or AIOps. The discussion analyzes the current state of machine learning, the changes automation has on IT Operations, and the future of AIOps.
Current state of machine learning
AIOps is statistical probabilistic reasoning rolled into algothrims. Currently, machine learning is probability and statistics combined with the art of computer science. There is statistical reasoning present in AIOps in order to find answers and problems through sophisticated algorithms versus hard coding. The key difference between AIOps and traditional computer science is that what you build with machine learning is a meta-algorithm that learns from data and what rules you would like to apply to that data set.
Automation across IT Operations
The amount of data that needs to be processed compared to the existing labor force is accelerating faster than budgets for talent. This creates significant opportunities for IT teams to look at how they can leverage machine learning to automate mundane tasks and make sense of the terabytes of data coming into and out of the organization.
Machine learning is the primary tool being leveraged by leading software vendors who tout Artificial Intelligence. Machine learning can play a significant role for IT Operations teams to predict and even resolve outages before they occur. Further, machine learning can process data and find patterns humans can’t.
IT teams that step in front of the movement to machine learning and artificial intelligence will turn IT into a competitive advantage for their business. The conversation of how IT adds value to the overall business equation can become a thing of the past if IT leaders can align their AIOps strategy to the business value chain.
The Future of AIOps for IT Operations
The primary goal of IT teams in regards to AIOps is to deliver services in the most efficient way possible and determine how to close the data gap. IT leaders want ALWAYS-ON availability, to decrease downtime and prevent outages. Human capital is an extremely valuable and expensive component to the business. Companies want to free up their highest performing team members to work on innovation, not completing repetitive tasks that could ultimately be automated.
Problem identification and resolution can also be automated with AIOps in a “way that is not possible with the rules-based system” Tee said. The large amount of data that AI provides allows for more problems to be solved before they turn into major issues. AI can essentially improve customer outcomes with fewer working hours in an automated and repeatable way.
As artificial intelligence continues to progress and advance, IT leaders must adjust and modify IT Operations in order to keep up with their competition. Machine learning continues to evolve and businesses must prepare their infrastructure to provide the best service to their customers. AIOps has the potential to bring more effective and efficient services to customer and employee interactions through automation and identification of high-value human touch points. The future of AIOps looks bright for IT.
Want to dive deeper? Watch the full episode below, listen here or view the transcript below to catch the highlights.
Show Notes:
Phil Tee, CEO and Co-Founder of Moogsoft
https://www.linkedin.com/in/philtee/
Transcript:
Speaker 1 (00:01):
SEAN: Welcome to the AI ops evolution podcast. This series features visionary IT leaders who are paving the way for the next evolution of the IT industry. Discover the truth about where we are in the adoption of artificial intelligence for IT operations and actionable tips that you can implement to become a more effective IT change agent. In this episode, we are joined by Phil Tee. Phil has been placing bets and winning in the IT space. Since the early nineties, he co-founded Moogsoft© and invented Netcool. By 2012, Phil had become frustrated by the lack of innovation and systems management and founded Moogsoft, a pioneer and leading provider of AI ops solutions that help IT teams work faster and smarter. Now let’s join your host, Sean McDermott, a mission-driven serial entrepreneur, IT engineer, and AI ops visionary for this exciting discussion. Welcome, everybody, to the AI Revolution Postcast today
(01:03):
SEAN: My guest is Phil T., Phil and I have known each other for gosh, 25 years. So, we’ve been around the block a few times. So, Phil is the co-founder of a company called Mike Amused, the founder of Riversoft. Both of those companies went public. One went on the NASDAQ, the other one on the London stock exchange, I think prospectively. You’ve done a bunch of things since then. So why don’t you tell us a little bit about yourself? Give yourself a little bit of time here to talk about who you are and, and … oh, by the way, I learned something about you today. You have a bachelor’s in physics. I knew that you’re a math guy, but I didn’t realize physics. And I have a lot of appreciation for that.
Speaker 2 (01:52):
PHIL: Because I’m an art guy who came out with an electrical engineering degree and I still want to have that happen. And physics was the bane of my existence in engineering school.
SEAN: But you also have a Ph.D. in philosophy, so I did not know I was supposed to call you Dr. Phil, because you’ve never told me that all these years. You keep these things relative to yourself.
PHIL: The PhDs are informatics; information, science, mathematics, and computer science. You know, physics is a funny old thing. It was my first love. It’ll be my last love, in this life. I kind of viewed as, it’s either something which you are in love with or something you are in hate with. And if the resolution is keen enough on the video here, you will see I visit a library, stacked with all kinds of fun stuff from quantum field theory, through differential geometry, and everything else. But for my professional life, it’s the hate side for me because it probably brought down my overall GPA about four-tenths of a point. It was the bane of my existence.
Speaker 3 (03:09):
SEAN: Well, you know, I love helping people through physics, and if it makes you feel any better, probably one of the most famous theoretical physicists alive today is a chap called Edward Witten, who is very famous for String Theory. And actually, his first degree was in History. So it is entirely possible to go all the way from the Liberal Arts to theoretical physics and actually end up with the field of metal. Which is a pretty good outcome.
Speaker 2 (03:42):
So, fill in the gaps of what your Riversoft micro means for the audience here.
Speaker 3 (03:49):
PHIL: Sure. So, the professional life was really an extension of all of the fun stuff early on in the sense that, everything that I’ve been involved with, Micromuse and Riversoft and beyond, has really been about trying to answer this very simple question: People own a lot of digital infrastructure. They’re basing their businesses quite often on the availability of that digital infrastructure. And there’s a small number of people with a limited amount of information whose job it is to keep all of that available. And so, naturally, software tools evolved to assist with that. And for me, coming from my background, it’s almost a parallel, to the whole mission of theoretical science, which is based on a very small amount of data. What can I predict about the reality of the university’s case for that digital infrastructure? So with my community, it was cool because it was rules-based with Riversoft. It was topology discovery to a root cause analysis and all the while we were evolving the algorithmic approaches that we have, to do that. And, and of course, Moogsoft, in this incarnation, has really taken AI and machine learning and attacked the same problem in the context of modern infrastructures.
Speaker 2 (05:07):
PHIL: Which is going to be more and more important because the amount of data coming out now is exponential. When we were doing monitoring networks in the 2000s, compared to now, it’s this exponential factor of more data every day, every minute coming off and it’s humanly impossible to process it.
SEAN: So, let’s talk AI and machine learning. I think a lot of people just say, “Oh, AI ops” and “he dropped it around.” I’m not really sure a lot of people truly understand the math and what’s actually going on here. So, in layman’s terms, especially for someone like me, who doesn’t understand physics, describe what, what machine learning really is.
Speaker 3 (05:56):
PHIL: So it’s kind of something that nothing, if I’m going to be really honest. I think a lot of people get very afraid of the terminology because it sounds so different and there’s a lot of psychology around. You know, AI and the robots will eat us, some Terminator and all the rest. So I must come out and say, yeah, we’re all in a simulation. By the way, if we’re in a simulation, somebody needs to dig, poke the current one. But quite literally, it’s got a virus. But the fundamental characteristic of machine learning versus maybe traditional, software and computer science is that what you build with machine learning is kind of a meta-algorithm, an algorithm that learns from data, with rules, if you like, to apply to that data.
Speaker 3 (06:50):
PHIL: So, traditionally, if you were to go back 40, 50 years to the origins of AI, there were essentially two camps. One camp was the rules-based crew. So, for example, when you’re trying to understand an English language sentence, you will pull down the language to a set of functional rules around grammar: subject verb object, I, before E except after C, all that kind of stuff. And you would use those rules to make sense of the data that you have. So you take a system, you mobilize it into rules, you apply the rules to the system. The other camp was statistical. And the idea there is, okay, if I start with nothing, and I just give you lots of examples of sentences, you can work out by the statistical distribution of verbs versus adverse words, and then defining in English, and then work backward to an effective understanding of the grammar of the language.
Speaker 3 (07:55):
PHIL: The problem was, back then those statistical approaches were very compute-intensive, and the rule-based approaches were transformed in the computers that were available. So one got put to one side, the other one kind of took off machine learning; that statistical approach, that business of saying, for example, just give me lots of examples of cats on our workout, and the features of the image of a cat. The overall features mean that it’s a cat, you know, with those sharp teeth, and really that’s the fundamental difference. So what you find in AI, is that machine learning is really a lot of statistical probabilistic reasoning rolled into algorithms to spot the thing that you’re looking for versus a lot of hard-coded rules.
Speaker 1 (08:45):
AD: We’d like to take this opportunity to tell you a little bit more about our sponsor Windward Consulting Group. For More than 20 years Winward has been helping IT leaders run complex global networks and data centers to deliver impactful business services. Windward is the full package with industry-leading solutions, architects, and integrators, with certifications, partnerships, and deep expertise to integrate leading ITSs and AI ops tools with leading technologies like service now and Splunk Inc from technology implementation to integration to manage services. Winward has your needs covered with Winward you get strategic thinking and real-world results. Do you want to see if Windward’s ATM is a good fit for your needs? Go to get.winward.com/aiopspodcast to book your discovery meeting today.
Speaker 3 (09:36):
SEAN: Where do you see machine learning going? People kind of think that machine learning is new, but where do you see it going in the next three to five years? Because computational power is increasing. You know, Moore’s law… So where do you see AI kind of going from there?
PHIL: So in some senses, it kind of bears in mind a couple of quotes around, you know, the relationship to mathematics and natural science. There was a famous physical chemist, Peter Acars, who once said, “If it’s possibly mathematics, it becomes a theory in science.” You might say that, if you look at what we currently do with machine learning, it’s probability and statistics bleeding its way into the art and the craft of computer science, but there’s a huge body of more complex mathematics that I think is beginning to approach.
Speaker 3 (10:47):
PHIL: For example, there’s a relatively obscure area today called topological data analysis. It uses techniques of something called homology and [inaudible], which is differential geometry, to try and work out the shape of data. And so if you can imagine plotting data points according to how catalyst they are, how many eyes they’ve got, and you kind of get a point in space where every example of a cat versus a tiger versus a dog versus a giraffe. You could use that to work out the shape of a surface that fits in and then take an arbitrary data set and compare it to that shape and use very powerful theories of mathematics to work out whether or not you’ve caught it. Another example of the training data that you can have that’s just one technique that is, like I say, relatively obscure today that I think is going to be mainstreamed tomorrow.
Speaker 3 (11:47):
PHIL: And it’s very compute-intensive or extraordinarily powerful, and perhaps an advance on neural networks. Another problem with neural networks is they’re pretty predictable. Ironically, given this early on people were worried about determinism. And by that I mean if you give it the trading data and then it’s very good at telling you when something is another example of what you’ve trained the network for. What it’s extraordinarily poor at is novelty. It’s not very good at giving you, say, “You told me to look for a link dancer, you know, world. And now I’ve got a neighbor loss neighbor restore. Is that a kind or a link-up-link-down type thing, which is a network science course? It is why you need to tell the neural network path, but actually, a traditional neural network would suck it up.
Speaker 3 (12:40):
PHIL: But these new generative neural networks that people are beginning to think through and design algorithms for are specifically designed for that target thing and there is active research on that. And the list goes on and on as to where I can see it going, but it’s not going to the headline AI in the title, right. Artificial intelligence. That’s an extraordinarily loaded term. And I would probably say one thing about AI that’s perhaps unpopular in some circles: How can you build an artificial intelligence if you can’t describe what intelligence is? And so I think the thinking computer we all are imagining is somewhere else. It’s not impossible to imagine that it will happen. And I think the advances are going to be much tighter, much more human, like database reasoning that is quicker to train and more powerful in terms of its predictive capability.
Speaker 2 (13:51):
SEAN: So, it’s really interesting that you said all that, right? Because I’ve been having conversations with a lot of people about the term AI too, and it’s a misnomer, I think, in what we’re doing, because really what we’re doing is ML.. But ML ops doesn’t sound as cool as AI, right? So, Gardner, came up with a term or reclaim. But now what’s interesting is you’re starting to hear people say, well, AI in this context really means augmented intelligence. Okay. Alright. But that’s for another podcast and we’ll do that. So we’ve gone through the theoretical and we’ve gone through the understanding of this. I appreciate that. I think my people need to really understand the basis of all this.
Speaker 2 (14:40):
SEAN: What are companies trying to tackle today with AIops? You guys are on the forefront of this. You guys have been connected, what, five years ago… When Moog was just starting out and I don’t think you really even had an office in San Francisco yet, and now you’ve got a big office in San Francisco. So five years have gone by, and steam is starting to come up. I was at the Gardner Symposium show, fall of last year, 2019, and AI ops is everywhere now. So what are companies really trying to tackle today with AI ops? It’s getting noisy, so help us understand what companies are already trying to do.
Speaker 3 (15:30):
PHIL: So I think fundamentally it boils down to automation, um, the drink bottle. So if you really think about what a typical customer in the “before” phases, when we meet them, their complaints have a number of things in common, such as availability is not where we want it to be. There may be a desire to reduce downtime or Richie soundstages by significant amounts, but whenever they’ve attempted to do that, they really have been able to get there by throwing bodies at the problem. And, of course, whether you’re in economic boom or economic decline, human resources is an extremely valuable and expensive commodity, so you don’t want your best people doing repetitive and menial tasks. So if you think about the operations use case, typically what people are looking for us to help them reduce that amount of menial, straightforward kind of tasks such as ticket catch and dispatch, so that somebody has acknowledged it, but then there’s all the repetitive, frontend, problem management incident management.
Speaker 3 (16:55):
PHIL: We can automate with AI in a way that is not possible using the rules-based system because of the amount of time it takes to essentially construct the rules. The need to find that the underlying infrastructure has changed. And, of course, when you do that, and it’s due to another beautiful characteristic of machine learning and AI, which is the more data the better. And so we see more by definition when we wire ourselves up with the customer; we’re actually able to catch more things, catch more problems before they escalate into a true service-impacting issue. So, to summarize, fundamentally, customers are typically using this to improve. We want to see availability improvements, and we want to do that with fewer working hours.
Speaker 3 (17:48):
PHIL: So, reduce my tickets, reduce my inbound alert load. And we want to do that as far as possible in an automated and repeatable way. There are probably plenty of other things I can kind of point to, like the almost collateral effect of implemented AI ops is that we capture a lot of the tribal knowledge. So there’s the ability for continuous improvement where we see repeating issues that can be remediated and sharing of knowledge. So, the workforce comes up to speed more quickly, but principally the benefit is availability and workload. Good.
Speaker 2 (18:27):
SEAN: So I’d just love to get your opinion: How do you respond to people saying, like, “Oh, you’re trying to automate me out of a job.”
PHIL: Right.
SEAN: We’ve been hearing this for 20 years. When I started real ops in early two-thousands, mid two thousands, we were all about automation. It’s all we ever heard. People say it never happened. So what’s your response to that right now…
Speaker 3 (18:57):
PHIL: You know, obviously in a contemporary setting with data and with some extraordinarily trying times economically, all of a sudden with the COVID and we have been used to very strong economic growth, which has been going on really fundamentally for a hundred years in the United States, and the developed world — and in the developed world is even more. The reason why that’s relevant is there’s always more work. The amount of work is increasing faster than the labor force is increasing. So, actually, I don’t see being automated to a life of leisure any time soon. Yes, there will be different work. I think some very surprising jobs will go, you know, the legal profession and the medical profession, the front end, maybe automated by AI, as much as your network operator or your SRE or whatever the role may be. But I think that this will create employment. I think it will create better employment for the individuals concerned. And it will change how we think about the working careers that we have.
Speaker 2 (20:17):
SEAN: Well, I know it’ll create new jobs too. It’ll create jobs that we can’t even know about today. You think about the cloud… And people say, Oh, moving to the cloud is you’re shifting all the work to Amazon, and we’re going to lose jobs, but now you have all kinds of these real cloud governance roles that didn’t exist, three or four years ago.
Speaker 3 (20:39):
PHIL: Well, so there’s a writer called James Burke who is responsible for connections the day the universe changed quite well. From some work we did for 10 or 15 years to get together now, he describes it like this: When you have a massive change in the world–the Industrial Revolution, for example, is a classic one. The invention of printing is another one. and the black death. We’re living through one right now with the technology issue revolution. What fundamentally characterizes them is if you take a grandchild and his/her great grandparents, the great grandparents will likely not understand what it is that the great-grandchild does for a living. That’s true. Think about it. If you were born in the 1880s and your great-grandchild goes out to work in the 1940s, you pretty much know what they’re going to do.
Speaker 3 (21:35):
PHIL: They’re probably working in a factory; they might be an accountant or a lawyer. They might have to go fight an afar landlord, or that was pretty common currency. In the 1880s, maybe things were a little bit different, a little bit more modern, but then when you think about somebody born in 1960, and you think about somebody going to work in 2010, in 1960 the question would be, what is search engine optimization? How would you explain that to your great grandparents and the 1960s? Somebody who had a rotary phone doesn’t know what to do. They pick up the receiver. They don’t know what that thing is that you turn your finger with. They’re saying, “ What is this crazy thing in my hand”, right? We’re all aware of, you know, the parent’s smartphones, and they’re more powerful than the sum total of computing on planet earth in 1960. We created something called the internet that hadn’t been invented that would blow the minds of people. And, we throw these things around like that, effectively that disposable. Maybe that’s going to change; maybe, they are going to become disposable.
Speaker 1 (22:50):
Introducing RedMonocle, the only commercially available software solution that provides IT executives with real-time AI-driven analysis of their IT investments and associated risk factors, RedMonocle simplifies tools, discovery tools, rationalization, and risk analysis. So you can achieve always-on service availability, mitigate the security risks of technology acceleration and optimize it. Spend RedMonocle is a single source of record for it. Tools, costs, and features powered by a Knowledge Base of over 1500 commercially available tools run. “What if” scenarios to see the full impact of technology change and identify gaps that may result in security, breaches, and downtime, and you’ll get an expert portfolio analysis from your dedicated portfolio advisor, RedMonocle’s CIO, his best friend, who turns your IT tool portfolio into a strategic advantage. Want to take a test drive? Go to RedMonocle.com/aiopspodcast to book your demo today.
Speaker 2 (23:52):
PHIL: So you’ve got customers saying, how do I deliver services better? How do I close the gap? Because I think that’s what we’re really talking about. It’s the data gap, right? And people aren’t going away. The job’s going to go away, and infrastructure is getting more complicated. The data coming in from the structure to understand it is getting more complex. We’re trying to fill the gap…
SEAN: So you’re saying, AI ops is something we should be looking at. Great. So what do you see as some of the challenges for customers? I mean, you’re on the front end of this, right? You’ve been doing this for five years. What do you see as major challenges for people adopting this technology and operationalizing it?
Speaker 3 (24:40):
PHIL: It’s an old trope. There’s the three Ps: patience and people, processes and technology kind of thing. And two P’s in materials and it’s still true. If I look at customers that we engage with we certainly bring the technology to the table, but if people or processes are not brought along with it doesn’t get the benefits for them that are really there for them to address. And fundamentally what it means is, look at the training, bring people with you, meet people, and understand that this is a change that is necessary and that this change will benefit them. So it’s kind of back to your original point: this is not going to automate my job away.
Speaker 3 (25:30):
PHIL: If we’re going to compete with X, Y, Z company down the road, we’ve got to adopt all this new digital infrastructure, and if we don’t make the change we’re toast. So it’s very important that you solve the vision and the mission, and then enable the people with the skills to resolve the technology. And some of that means overcoming this psychology of resistance to AI as a tool that’s being used. And then the second piece is, always encourage being progressive with process change. You’ve lived this one better than I do. The harder you go at the front end with process change, the more resistance you get. It’s almost like Newton’s third law.
Speaker 3 (26:21):
PHIL: Equal opposites, a reaction. So you’ve got to let people adopt at a pace that is comfortable to them. Very early on, a mistake that we made– we talk about the situational workflow, which (to strip away the nonsense terminology, what we really mean by that is a collaborative ticket rather than a siloed ITIL, V3 four) –we raised a database ticket. But it wasn’t the database. It got moved over to compute with the collaborative instance and everybody’s involved at the get-go. And we decided who the right team is to be on every instance as we deal with it. We’re very early on. We made the mistake of trying to, frankly, run that down people’s throats, which is, give up your ITIL V3, do it this way.
Speaker 3 (27:05):
PHIL: It’s far far better. Now we were both simultaneously right and wrong. It’s definitely better. You definitely get a huge benefit from it at the backend, but you’ve got to get people to be able to adopt it. And actually what they want to do is they want to just get comfortable with the benefits of the better instant detection, the better mean time to detect, the better mean time knowledge. And then, as they get comfortable with that, you nudge them to the collaborative incident. You raise one ticket for every alert, but you also raise a ticket. For instance, that may be covering six or seven different alerts, and you encourage collaboration. And what we found is it kind of gathers pace under exam momentum, and you end up where you want to get a little bit quicker. My prediction with that is, I suspect that in two, three years’ time, as people adopt tools like Slack and Zoom, it’ll probably become less of a resistance for customers, but fundamentally it’ll go at the customer’s pace, solve the dream internally, involve the community around operations SRE, make sure they’re enabled and make sure they’re not trying to bite off too much early on with the process transformation.
Speaker 2 (28:28):
SEAN: I think what’s going on with AI is that you got so much discussion in the market today about AI and replacing. You’ve got Andrew Yang talking about AI, self-driving trucks replacing 3 million truck drivers. You’ve got Elon Musk talking about that the greatest threat to mankind is AI. That will all play out at some point. But I think with all that noise going on, I think people are looking at AI really. They’re scared, right? Now there’s a term that they can associate with the fact that they’re being replaced. Where before it was just, well, talking about technology. Now there’s this term and everyone’s talking about it and it’s all contextual.
Speaker 2 (29:24):
PHIL: One of the things that we advocate for a lot, exactly what you just said, is really having a strong leadership vision, articulating that, and making sure that people understand the value of these changes, and then really focusing on the operationalizing of it. Meaning: How people are going to use this stuff every single day. And how’s it going to change their world? And that gets back to what you said, training and doing it at their pace. So we’re big, as consultants who are responsible for implementing this stuff, we’re really big on the operationalizing of this and the psychology behind that. And we actually are rolling out new solutions around operational change management. And you can read those.
SEAN: So moving on. Where do you see this all going next?.
Speaker 3 (30:32):
PHIL: The daily exercise of a growth ambition of Moogsoft … if you look across the whole operations tool…
Speaker 2 (30:49):
SEAN: o, obviously, you’ve founded a few companies, you’re a highly successful CEO. Do you still “geek out” on this stuff?
Speaker 3 (30:58):
PHIL: Are you kidding? Yeah, absolutely. You know, I love my job. I love running the company. I love leading a mix of people. We have a fantastic team, Greg honors, and I love my customers. All of these people you know, know for a very long time. So the community in which we sit is fantastic. But I am at my happiest when I’m getting with the technology. In fact, you can visit research.microsoft.com and see that we’re super active in the research community, around AI and machine learning. And I love the fact that we get to be seen simultaneously as thought leaders there as well as key-minded businesspeople around the job of building new salt.
Speaker 3 (31:55):
PHIL: I love seeing the outcomes. It’s great to have an idea. It’s fantastic to sort of noodle around a concept, but seeing it in action is even better. We started a program in 2014 to try and build a new way of thinking about how you do root cause analysis with networks apology. And we did some pure research. We published some papers. We came up with Cinco Vertex Entropy, and I now get to go visit customers who are using Vertex Entropy to improve their availability outcomes. I mean, it’s perfection, I don’t know a better life than that. And as I look forward I want to see us take more of the research that we’ve done and make it real for customers. But it’s a business.
Speaker 3 (32:51):
PHIL: If you like the kind of the entrepreneurial thrill of what we’re trying to do, look across the whole of the operations market. There are billions, maybe tens of billions of dollars globally invested in technology. That’s as old as my career is long. Maybe somewhat immodestly largely due to the fact that I read quite a bit of it. Back in the day in the 1990s, you couldn’t understand mobile internet. Decades before virtualization became a thing, a server was something that if you dropped it on your foot in 1990, it sent you to the emergency room. Today, you can’t drop a server because it floats in imaginary space. It’s a virtual idea. So I think that there is an opportunity to completely generationally refresh that entire toolchain.
Speaker 3 (33:51):
PHIL: And I know what’s going to replace it: it’s going to be much more cloud resident, much more AI-enabled, much more agile, and much more modern, frankly. And I think Moogsoft offered the challenge to us that we could invent that world, and we could be the driving force behind that generational refresh. And that’s what gets me out of bed in the morning. That’s what gets me excited. Because the outcome for our customers in the market is necessary. If this global digital transformation is going to … take the facts and I’ll maybe make one slightly contemporaneous comment about what’s going on in the world with COVID. If you think of a world where everybody’s wearing one of those– a wearable computer — how much easier would COVID have been able to deal with? If you can monitor everybody’s health simultaneously 100 percent of the time, and know when you need to intervene to help somebody to stop, maybe an epidemic spread, or the rest of that kind of stuff, that’s a digitally transformed world. And that’s what we’re trying to enable.
Speaker 2 (35:05):
PHIL: When I started real ops in 2004, when we were going around raising money, the people who have been truly invested said we were dealing with these issues in the nineties. They haven’t been solved yet. And the reality is that we’re still dealing with a lot of the same stuff that we dealt with in the early 2000s and 1990s. To me, bringing machine learning into the equation now is really one of the most exciting things as an operations guy. I implemented a lot of software you built, right? I built knocks. And I went from event management to root cause analysis to topology and all in this kind of long way.
Speaker 2 (36:00):
PHIL: So I’m an operations guy and bringing AI into this now, and it is really exciting to see. Because it’s not just about AI monitoring and processing massive amounts of data, but to get to those resolutions, it’s also about bringing AI into my passion; a lot of animation and what we did at real ops, 15 years ago. If we had technology like this today … We were writing rules, essentially workflow-based rules. It was a process engine on top of an integration layer. We would be so far ahead bringing this technology into automated remediation and things like that. So that’s what excites me. And I’m glad to see that’s what you’re thinking too.
SEAN: So, two things I want to wrap up with. One, is I obviously want to give you a moment to give a plug to talk about MOOC software that you guys are doing. You’ve been doing that and doing a great job. I have a different question. I’ve started a number of companies. I’m now on my third software company.
Speaker 3 (37:22):
It’s a rehab for that population.
Speaker 2 (37:24):
There is a risk. And I keep saying, “I’ll never do it again.” And then I do it again. But you’ve done this too, right? And you’ve been massively successful. What makes a great product? Speaker 3 (37:40):
PHIL: You have to be at the intersection of needing capability. And I think that that really is the fundamental driving thing now. I can make all sorts of comments about, be in the user’s shoes, build it from the user’s perspective, make sure that it’s adoptable and consumable. But fundamentally great products are the intersection between need and capability. And I’ll perhaps argue by example, Salesforce is an incredibly successful business, amazingly adopted technology. But you don’t have to throw a rock very far to hit somebody who hates the product. And, actually, in some senses, it’s kind of successful because of the way in which they’ve dealt with building a community around fixing the stuff that’s missing. It’s clunky, old-fashioned UI, and difficult to manage.
Speaker 3 (38:44):
PHIL: It’s got no reasonable API to work with like a modern product would have. But very beautifully at the outset of Salesforce, the no-software thing, the ability to have an enterprise-grade CRM effectively, which you didn’t own. The software and the capability of being able to deliver that as a service met very beautifully, and created a megalith, a company, and probably one of the most successful software products of all time. You would say the same thing about DAS. You can say the same thing about the relational database. I’ll be honest, you could say the same thing about my communities and most of the mix for sure. Are we the best Moogsoft we could ever be? No, otherwise I would have sacked my engineering force a long time ago. For every cent of profitability that I could make, we’re still trying to be the best Moocsoft we could ever be. And we’ll continue to do that. I think the need and the capability met in a very significant way at the very beginning.
Speaker 2 (39:56):
PHIL: And that’s why building software companies is so hard. Because sometimes the capability is years before the need. And if you’re a visionary, you know the need is there, but you’ve got to convince people of that need. And We’ve gone through a couple of evolutions of that with my companies over time. Or you’ve got this idea and then you go out and you’re like, why don’t people see this? And you’ve got an education process And sometimes it takes a few years for the two to intersect.
Speaker 3 (40:37):
PHIL: Are you familiar with the phrase “patient impatience”?. You sometimes have to believe and be patient with the market. It catches up with an insight that you’re trying to turn into a company, but you also have to be impatient as well. Otherwise, you’re gonna miss them, you’ll be too relaxed. You’ve got to be continually on your guard with it.
SEAN: But you know, here’s the four-letter word that nobody likes to have met. When an entrepreneur has had some success, they say, my biggest ingredient was luck. I was very, very, very lucky to be at the right time and the right place for my communities to have fantastic colleagues and backers and customers and partners all along the way. I’m pretty sure there were people who were quite as talented as I am, twice as dedicated, and twice as hard-working as I am that didn’t make it.
Speaker 2 (41:39):
PHIL: Absolutely. I say that all the time. Luck is hugely important. But you have to work hard to put yourself in a position to take advantage of when it happens. When we sold real ops in 2007, when we started that company, I thought we had about a three to five-year window to hit that hard. But we got bought out at two years and nine months. All of our competitors were acquired within a six to nine-month period, and that whole thing was over. Never to be recreated again. But we just happened to be there at the right place at the right time. Like you said, there’s the intersection of need and capability at the right time.
Speaker 3 (42:31):
SEAN: So, o, they get so Moogsoft. We’re the pioneer in AI ops. The company is eight years old, with 150 customers out of the Fortune 2000 worldwide, growing at a massive clip, a knowledge thought leader in this space. And I can point to outcomes at pretty much every large Wall Street bank, large insurance companies, all manner of different enterprise businesses where we’ve seen reductions in such as Algera tickets by as much as 90% improvements in MTTR. And MTT a M T T D being the primary measures of operational efficiency, improvements of 40 to 75% in those. We also have a new, entry-level version of the products going GA at the end of May, and ops express, which is a fully multi-tenant SAS offering. And it’s free for the first 31 days of use. Please sign up for a trial. Excellent.
Speaker 2 (43:41):
SEAN: Good. Well, my friend, we will be seeing each other more. It’s been quite a ride over the last 20, 25 years. I think I met you guys in ‘97. So crazy, the same age as my company. A good group of people over there, you guys are great. Good luck. It was great talking to you, and thanks for taking an hour out of your time in this environment. I know that there’s a lot of other things you could be doing, trying to figure out how to deal with this crisis we got going on. But I appreciate all your time.
Speaker 3 (44:27):
PHIL: It’s a real pleasure. Great talk, Sean. And hopefully, when things calm down a little bit, let’s get a beer somewhere.
Speaker 1 (44:36):
Thank you for joining us in paving the way for the evolution of AI ops and the next generation of it. Innovation. If you are inspired to dive deeper into the movement, check out all the resources available@aiopsevolution.com.
