Microsoft Ignite: Ask the Expert - Abnormal Security

Microsoft Ignite 2020 Session

Ask the Expert: Bringing Next-Generation AI Innovation to the Enterprise

Listen to a live Q&A session from a previous Microsoft Ignite session (which you can view here). This esteemed panel discusses how Microsoft is partnering with Abnormal Security to bring AI to enterprise email security.

Scroll down for a full transcript

Speakers

Evan Reiser

CEO & Co-Founder of Abnormal Security

Saam Motamedi

General Partner at Greylock Partners

Jeff Ma

VP of Microsoft for Startups

Why Abnormal

Learn More About the Abnormal Difference

Transcript

Ted Roduner:

Hello, everyone. Welcome to the “Bringing AI Innovation to the Enterprise” – Ask the Expert session. I’m Ted Roduner with the Microsoft for Startups team, and I will be moderating today. I’m joined by Jeff Ma who leads the global startup team. We’re thrilled to have Saam Motamedi, General Partner at Greylock Partners, as well as Evan Reiser, CEO and founder of AI security startup Abnormal Security. This is an Ask the Expert session. The operative word here is “ask”. This session is designed for you. Please ask questions. We’re talking about a timely topic, AI innovation in the enterprise beyond the buzzwords. What does that actually mean? So please ask questions up, upvote questions, and we’ll curate these for the panelists to tackle during the chat. Finally, so, you know, the session will be recorded and we ask that you abide by the Microsoft code of conduct. Please be civil, be human. And please, please take part in the session. With that, I’ll hand it over to Jeff.

Jeff Ma:

Thanks, Ted. You said the operative word was “ask.” I think the operative word is “experts” here because we clearly have two amazing experts to cover this topic. For those of you guys who did not see the prerecorded session that the three of us did, with Dean who is the CISO of Fox, we’ve been talking a lot about how Microsoft can help bring innovation, sort of next generation AI into the enterprise by helping it to be easier. Helping startups make it easier, helping them to work with enterprises because traditionally it is pretty hard for startups to work with enterprises. So Saam, as a VC partner at Greylock and someone that has worked with many startups, what are the challenges that you think that startups have in working with enterprises, specifically startups that are pushing the envelope around technologies like AI?

Saam Motamedi:

Yeah. Thanks, Jeff. It’s good to be with you guys again, and it’s a great question. So I’d start by saying that every week at Greylock, we have entrepreneurs come to our offices, pitching new company concepts using AI. And I think we, as much of the world are very excited about the potential of AI specifically to reinvent business workflows across the enterprise. That said we don’t call ourselves AI investors. We invest in companies that solve end-customer problems. And if they’re using AI, machine learning data as a part of the product advantage to drive a superior solution, that’s awesome. And we can get very excited about that. The challenge is many companies struggle with connecting their use of AI and technology to an actual customer problem, and then delivering a solution that can scale and solve the customer’s problem with very, very high quality. And so then to your question around like, “Hey, what are some of the challenges involved?”, I bucket them into two pieces. One is on the technology side and the second is on the go-to-market side. So on the technology side, I think the challenge is how do you as a new company pick a problem and then actually deliver a solution that uses real AI for that problem quickly and to customers scalably. And I think part of what goes into your decision around who to partner with, and part of why we’ve been excited about what you guys are working on at Microsoft and Azure is the suite of services that you can offer startups to leverage and much more quickly get to market, cutting edge AI products.

Saam Motamedi:

So that’s on the technology side, now on the customer side, the challenges if you think about …worldwide, there’s something like 20,000 companies that do north of a billion dollars a year in revenue, you know, medium and large enterprises, this segment controls something like 85 to 90% of IT spend. So if you’re starting a new company today, if you can target and sell into that segment, there’s enough runway to actually build a really big business. Then the question is, how do you actually get into that segment, penetrate, build credibility with those customers and partner with them. And I think that’s the challenge. And again, like part of why we’re so excited about Microsoft for startups is Microsoft and Azure have relationships with a very, very large percentage of these customers. And so, you know, if I’m running a new company, I’m trying to break into that segment of the market, being able to partner with someone like Microsoft and leverage those relationships to get my solution to market. That makes a world of difference.

Jeff Ma:

So Evan, I’ll kind of ask you similar question from your point of view as an entrepreneur. You know, a repeat entrepreneur. What challenges have you seen selling into the enterprise as a startup and what are the problems that a company like Microsoft can solve for you?

Evan Reiser:

Yeah. First of all, thanks for having me. I think there’s kind of two things. And I’ll maybe kind of echo what Saam said around both technology and business. On the technology side, building AI solutions, where AI is kind of the core ingredient that makes it work are very difficult. And so that a lot of small companies that don’t have thousand person engineering teams struggle to build some of the more fundamental building blocks, to get their solution up and running in a way that’s enterprise ready. There’s a lot of work that’s involved and we’ve had great success building on some of the core building blocks of cloud infrastructure to help us go do that.

Evan Reiser:

And the other kind of big problem that a lot of startups have is trying to reach customers in a way that customers understand, so it’s hard to get in front of the biggest companies in the world. And that’s cause there’s so many startups and they’re all saying things that sound very similar. And so, you know, being able to partner with someone that’s really credible and use that tool to meet more customers and help understand why, it’s not just like some other machine learning startup, I think that’s been a very powerful tool that’s helped us and I imagine will be a great approach for other entrepreneurs as well.

Jeff Ma:

When you use the term enterprise-ready, what does that mean to you? And again, like how can a partner like Microsoft help you become enterprise-ready? Can we, or is that something that like startups have to do on their own?

Evan Reiser:

Well, I think there’s two things. One is, we’re a cybersecurity company, but this probably applies to just any enterprise company. Every enterprise is extremely nervous about, and rightfully nervous about data privacy and security. So I think making sure you’re building on a really, secure and trustworthy platform like Azure, that goes a long way for customers. The second one is enterprises have more complex use cases and there’s multiple users and you just need to have more functionality that’s really focused on solving your business problem. So as a startup, it’s hard to spread yourself too thin. So if you can build on top of solutions and services that gets you most of the way there, things like Azure or Azure AI cognitive services, that allows your team to focus more on the specific business problems. So rather than having a very kind of thin solution that doesn’t have the enterprise features, if you can kind of build on top of great infrastructure, it allows you to your team to focus more and more on things that customers really care about.

Jeff Ma:

And obviously Microsoft’s current commitment to security makes it easier for you guys to solve the specific use cases that you want to go after, where you think that your technology is well suited to go do that.

New Speaker:

Saam, when you think about that term enterprise-ready, what does it mean to you? Because I’ve heard it a couple of times in the last couple of days, as we’ve talked through a lot of this stuff. And I think it is an interesting concept that the idea of a startup, which is so counter to enterprises becoming enterprise-ready.

Saam Motamedi:

Yeah. It’s an important concept. And I second everything Evan said, I think that was well summarized. You know, you’d be surprised how quickly a startup can become enterprise-ready if they focus on building and selling to enterprise from day one. And then they also surround themselves with partners who can enable them to do that. So when we think about enterprise readiness, we very much measure it both on product readiness and then the go-to-market maturity of the organization to actually be able to successfully acquire and service enterprise customers. So on product readiness, when you’re selling to the enterprise, versus let’s say an SMB segment, there’s a whole elevated set of expectations around things like quality, stability, success. And also integrations. And enterprises have complex IT environments and if you want to sell into these environments, you have to cooperate with what technologies the customer is already using.

Saam Motamedi:

And so I think, again, that’s why taking advantage of services like those offered by Azure and building against those where you get scalability built-in and where you get quality built-in, it can just dramatically accelerate the time you get to enterprise product readiness. And I’m very optimistic. I think it’s never been better to start a company oriented towards selling to the enterprise because you have all of this foundation in place to go build on top of. So that’s on the product side. On the go-to-market side, look, there’s a whole DNA around selling to enterprise, right? When you, as you build out your sales and marketing team and organization, bringing in folks who understand this type of customer’s expectation and have the ability to build these relationships, selling to these relationships, and then importantly, service these relationships, it’s absolutely critical. And so both of those things have to be in place before you can sort of successfully and repeatably sell into the enterprise.

Evan Reiser:

Jeff, if I could just add one more thing. I just want to highlight the point that Saam made, where if you kind of plan for that from day one as a startup, you can do it very quickly. And I say this with humility, but our first Fortune 500 customer, we signed nine months after we started the company. And the only reason we’re able to do that is for two reasons. One is we knew that’s where we wanted to go. So we’re able to invest in all the things that Saam talked about: security, privacy, scalability, all from day one, including customer success. The other thing, which is more on the technology side. And I saw a question come in about how long does it take to deploy an Abnormal solution?

Evan Reiser:

Well, the other thing to get to enterprises quickly is you need to build on a platform that allows you to to integrate quickly and for customers to see value quickly, that’s particularly important for the enterprise segment. And so, you know, because we built our product on top of the Microsoft platform, our customers are able to integrate in one-click, all through Microsoft APIs. We’re able to go back retrospectively and then show value to that customer within seven days. So, because we’re taking advantage of all these cloud native tools, both Microsoft Graph API and Azure, that allows us to deploy the product quickly. The customer can see value quickly, and that helps accelerate the kind of these enterprise sales cycles. And that’s something that just wasn’t possible five years ago. And so I think a lot of entrepreneurs are scared about selling the enterprise, but it sounds long and complex, and maybe they’re not ready, but I would love to just assure other companies. And I think we’re an example, if you focus on that and you take advantage of some of these great enterprise-ready technologies, there’s no reason you can’t land the Fortune 500 customer in your first year of being in business.

Jeff Ma:

So Evan, when you talk about planning for that. Clearly there must have been a customer discovery process that you went through. What kind of customer discovery process did you go through to understand what are the bells and whistles that you needed to sell into the enterprise? When I think customer discovery, I think about, Oh, you go to Starbucks and you interview like five people or whatever, and that you do the traditional, like Clayton Christianson kind of thing of asking a bunch of uncomfortable questions to someone. Did you guys go through that with enterprise type customers? Because I would assume that that’s hard because to get the CTO of Walmart or something to talk to you about what they’re doing and what they need probably sounds much more difficult than to get some Joe Schmo from the Creamery in San Francisco.

Evan Reiser:

Yeah. So, the short answer is, yes, we took that approach and we talked to, I think we probably talked to over 50 customers before we even wrote a line of code. So we’ve been always been very, very customer focused. It is very difficult. Because there’s 5,000 startups that also want to talk to the CTO of Walmart. And so that, I think that is a big challenge for enterprise startups. Every startup will benefit from the more time they can spend with customers, the more they can understand the problems and pains that’s going to help guide their product development and their company building to build an enterprise solution. I think that is a big challenge for startups.

Evan Reiser:

And I think that if I go start another company in the future, I think probably one of the most important decisions I want to be thinking about is, Hey, what is my plan to get in front of those top 2000 enterprise companies. Even when you think about cloud infrastructure, one of the benefits of using a platform like Azure, it’s not just a bunch of technology services that some of the things I’ve been really happy with working with Microsoft is some of the business benefits. It’s the ability for customers to purchase our solution directly from Microsoft through the Azure Marketplace, the ability to have this Azure Cosell program so that we can go meet these customers together and learn what the problems are. Those are some of the important ingredients that help enterprise founders build better products that are more effective for their customers.

Jeff Ma:

And that’s the reason I asked about the customer discovery process. Is it’s one thing we’re thinking about doing at Microsoft for Startups is because we have such great access to these incredible enterprise customers. Can we leverage them to help you get to your first MVP by allowing you to get access to them? Where we vet the people they talk to so that they don’t have to talk to 2000 startups. They can spend time with the right ones that are part of the Microsoft for Startups program. Ted, did you have questions that have come in through the chat?

Ted Roduner:

Indeed, I do. We have a question for you, Jeff. “All of the major cloud vendors have startup programs. What are the top reasons to partner with Microsoft versus the other guys?”

Jeff Ma:

Wait, there are other cloud vendors? I didn’t know that! That’s obviously true. I mean, our competitors all have programs that tend to center around credits. And ultimately I think of credits very much like airline miles. It shouldn’t be the reason that you choose to fly an airline and it probably isn’t in the long term. So what we’re trying to do at Microsoft for startups is become much more than a credit program specifically right now, where we know we can provide tremendous value. And you’ve kind of heard this expression over and over again, is helping startups go-to-market and helping startups sell into enterprise and helping startups get more customers in the B2B or business to enterprise state. So we have very scalable programs where companies like Abnormal can come in and Cosell with us, meaning we have 40,000 salespeople that will sell Evan’s product and will be able to retire quota against it.

Jeff Ma:

Abnormal is able to publish in our Marketplace, which allows our customers to transact directly through Microsoft, meaning they can avoid procurement and they can do their deals on Microsoft paper. We also have great resellers who are partners of us that can resell the Abnormal solution to many people. The CIOs still often buy through a reseller. So having a reseller in place that’s trusted that vets solutions is very, very helpful for startups and Microsoft obviously has these great relationships already with resellers. So it just eases one friction point for startups to work with us. We will be moving more and more into earlier points of the life cycle for startups, helping startups achieve their minimal viable product quicker or get to product market fit quicker, all these different sort of buzzword expressions that are important in the stages of a startup. But right now, our focus really is around this idea of business acceleration because it’s where we know we can provide tremendous value to startups. And as we’ve seen with some of the early startups that we’ve been working with Abnormal as an example, observed.ai, as another example, we’ve been able to really accelerate business growth.

Evan Reiser:

And just one quick thing to add on to that, Jeff, when I go talk to a Fortune 500 CIO. And he, or she says, “well, why should I pick you guys?” There’s lots of good technology and product reasons. But if I can go and say, well, part of the reason is because you can call up your Microsoft sales rep, you can learn more about the product, you can buy directly from Microsoft. Oh, and by the way, if you do, you’ll get Azure consumption credits, which help you reduce your commitment. That is very meaningful to a lot of enterprises. That’s not going to be the only thing. Ultimately you need to solve customer problems and have a great product they love, but that is a differentiator that does matter a lot of the time.

Ted Roduner:

Great. And we’ve got, we’ve got another question. This one is for you Saam. Someone asks, “how has the role of a VC [evolved] in working with startups and companies like Microsoft?” So I guess this sort of triangle, of VC interacting with corporates.

Saam Motamedi:

Yeah, it’s a great question. So, you know, when we invest in partner with a company like we have with Evan and Abnormal, we get very, very involved and we want to help them build the largest possible company. And so, you know, a lot of these decisions are actually made at a board level. So for example, Evan, you can speak to this much better than I can, but I, as we’re having this conversation, I’m reminded of many board conversations we had, some number of quarters ago around what should the cloud platform partner for Abnormal be, if you take a very long term view and for a modern SaaS software company, a decision like that is a very, very important decision. And so I think now that, Microsoft in specific has been able to differentiate against some of the other cloud providers. There’s more conversation happening at the boardroom and with investors having a perspective as well around what are the dimensions to consider in this decision and what’s the right decision for a company to make. And so I think that is one primary way. It’s changing fundamentally as an investor, like we want to help the company do what’s going to help it deliver the best product to as many customers as quickly as possible. I’d say from my perspective, just to also tie this to the last question that was asked, Jeff, you talked about credits, and on the technology side, I think the cloud’s differentiated from one another and have different pros and cons. To me the thing that Evan raised internally at Abnormal, that we were sort of blown away by and I do think is very, very new and differentiated is everything happening on the go to market side. And, you know, I’d say like, as an investor and board member, the ability to have a company accelerate its go-to-market into large enterprise by leveraging their cloud partner… I’ve never seen that and that can be sort of completely game changing for a business, particularly at an early business, that’s trying to build credibility and connectivity into the enterprise as quickly as possible.

Ted Roduner:

And this is a follow on Saam, “can you talk about the process that you went through at Greylock when you decided to invest in Abnormal Security?”

Saam Motamedi:

Yeah, absolutely. And it was, it was very much a mutual decision to partner together. But you know, fundamentally at Greylock, we’ve been investing for 55 years and over the last 15 to 20 years really focused on enterprise and consumer software. And we make very, very few investments per year. We are looking for companies that not only have a chance to grow revenues and scale and go public, but actually become enduring franchises. And a decade from now be the leading public independent companies in markets that are relevant and growing. And the reality is they’re very, very few companies actually have a chance of pulling that off. And so when we, when we make investment decisions like we did with Evan and Abnormal, at the early stages, we’re really looking at three dimensions.

Saam Motamedi:

One is the quality of the founding team. The second is the market opportunity. And the third is the team’s ability to deliver a long-term differentiated product into that market. And so in this case, I won’t flatter Evan too much, but we had the privilege of… Evan, Sanjay and the team they’ve built at Abnormal is one of the absolute top teams at the intersection of machine learning and enterprise software, and they had previously built and sold companies and then scaled larger public companies, machine learning platforms and the skillset, and going back to Jeff’s question on enterprise readiness, the familiarity on how to build these systems for enterprise readiness, that is extremely scarce. And so we were really excited about working with them and then they intersected that with a market opportunity that we think is pretty phenomenal where all of us now working from our homes, all communication is digital, email is the most important vector of that digital communication and the integrity and security of that communication is more important than it’s ever been.

Saam Motamedi:

And if you think about like you take a step back and you ask, Hey, like what’s happened over the last 15 years and email and digital communication security broadly, really in my view, two things have happened. One is enterprises are moving to the cloud and primarily moving to Office 365. And then the second is, attackers have gotten more sophisticated and have shifted their focus from, let’s say more basic spam and phishing attacks to much more advanced business email fraud and spear-phishing. And so I think Evan and Sanjay’s thesis, only two, I guess, a little over two years ago was: “Hey, can we reimagine this market, built for the cloud, both for Office 365, natively leveraging the APIs that we can use to deliver the solution. And then can we differentiate by using cutting edge machine learning techniques to actually solve these much more advanced attacks? And if we do that, can we become the market leader in email security?” And, you know, so the combination of that, the team and then our confidence that the team would deliver a superior product actually made it a very easy decision. That’s what went into it.

Evan Reiser:

Saam, can I add one more thing too, just because I think it’d be relevant for other other entrepreneurs. I think another big thing had nothing to do with us or Greylock. But it’s just the general shift to the cloud. So as probably many of you guys know, like the grand majority of enterprises have gone and deployed Microsoft 365. And so if you think about our industry, email security, 10 years ago you had to buy this box, you had to put it in your closet, you had to hook it up with your mail flow and now, we’re able to go deploy a product all through Microsoft APIs, a customer one-click and deploy Azure application behind the scenes and then get 90 days worth of results in a couple of hours. So, a lot of opportunities now for enterprise, especially those are the leveraging AI, it’s easier than it’s ever been in the history of civilization to go deploy a high-scale AI application very quickly for enterprises with relatively little work. You know, probably a hundred X less work than would be required 10 years ago because of these amazing platforms that enable us.

Evan Reiser:

And just to connect the dots: I think that applies to like lots of other markets. And I’m sure that, maybe even broadly for just IT in general, there’s lots of applications that are only so good because they were on-premises 10 years ago. And now, especially with some of the Microsoft platforms, whether it’s Microsoft 365 or Dynamics 365 or Azure, it’s gonna enable a whole new class of applications. And there’s, there’s probably dozens of billion dollar companies that will be created on that platform. And, hopefully entrepreneurs are going to go figure out how to solve these problems and build those companies.

Ted Roduner:

Great, Evan. To follow that, I have a two part-er for you. The first question is: “is Abnormal Security available across industries? “

Evan Reiser:

Yeah. Short answer: Yup. Everyone uses email. Everyone has email security issues and we don’t discriminate.

Ted Roduner:

Perfect. This is sort of …. Matthew asks a question: “dealing with security problems with AI is tricky. Since there are classes of problems that are targeting a system in a way that was not in the mind of the designers, i.e., an exploit. Can your AI deal with such unexpected problems or develop specifically what is the exploit versus some flux in that is abnormally or anything outside of defined metrics? Can your AI deal with concepts instead of metrics?”

Evan Reiser:

Yeah. I mean, that’s actually a really great question. And this has been a struggle in cybersecurity for a long time, a lot of cybersecurity technologies, think about antivirus or even anti-spam, the systems are built to recognize bad things, memorize those bad things, and then try to see when they come in again, they recognize them and stop them. And so that’s historically been a very effective approach for dealing with these different issues. And AI is very good at, you know, pattern recognition and memorization. The challenge that we see with a lot of modern day cyber security attacks, and there’s countless examples that you guys may see in the news, is the very successful ones are very targeted, very personalized. And they take advantage of these exploits, where it’s not something has been seen before.

Evan Reiser:

And so as guys know, it’s hard to build software that anticipates these unknown things. And so the approach that we take, and this is probably generalizable to many different products and many different industries is rather than trying to like memorize all the bad stuff we know about. We try to actually understand, what does normal behavior look like and how do organizations normally act. So when Ted and Jeff shared email with each other, what’s that normal conversation. Now, when you make a payment to your bank, what is the normal way you do that? And so, we don’t actually know the new attack that nobody’s seen before, but we can begin to understand and memorize, how do businesses work and how do people do things? And we can detect anomalies against that. So the core of our technology was understand the normal behavior and looking for these abnormal events that are likely to be attacks. So the, the short answer is no, well, we can’t predict what the next exploit is going to be, but we can tell you when something very strange happens, that’s very likely to be an exploit and that’s the basis for our approach to email security.

Ted Roduner:

Great. We’ve got a lot of questions specifically for you, Evan. Someone asks, “is there a particular cloud service that you have your eye in the Microsoft cloud that you think will help you take your solution to the next level?”

Evan Reiser:

Yeah, I mean, there’s a couple. Some of the new things that we saw yesterday were quite impressive, although they’re not all related. I think the ones that I’m personally most excited by though I’m not the expert in our company on this. Some of the cognitive services are very interesting, I think specifically around natural language processing and computer vision. In the last question, part of the question was, “Hey, can the system understand the difference between like metrics and concepts?” And so by applying more of these human perception technologies, such as computer vision, that will allow these machine learning systems to get smarter and make better judgements on behalf of security and IT teams.

Speaker 3:

And that is the premise of AI. Suck up all this data and make smart judgments. To save people [time], to basically do work for people. And so I’m particularly excited about just like the direction that those products are going. I’d love to see what else we can do with some of those kind of new services. And just to give you a concrete example: things like form identification – that’s actually very important. If you’re trying to detect a fake invoice or fraudulent invoice, to be able to understand, “Hey, here’s a form, pull out the different components. Oh, this thing here is not just black pixels, that’s actually a routing number. And that routing number looks nothing like anything we’ve seen before.” I think more and more of these cognitive services will allow AI companies to uplevel the sophistication of how these systems are making decisions. And generally that’s going to power a bunch of new AI applications for enterprises.

Ted Roduner:

That’s great. I think we have time for one more question. Jeff, this is coming your way. There’s a question. “How do you evaluate companies for the Microsoft for Startups program? What sort of companies and startups are you looking to work with? “

Jeff Ma:

Yeah, right now we, and I think generally we’ll always look for startups that we really feel like we can help. Startups that align with the industries that we believe we have incredible value in things like healthcare and life science, manufacturing, retail, financial services. We’re interested in working with open source companies also because we believe that there’s just tremendous value in nurturing those types of companies. And ultimately we’re looking for companies that are selling into the businesses and really selling into the enterprise. Because again, that’s where we think our value proposition aligns with what they really need. So, ultimately we also want companies that we think have a chance to become very successful. Just like investors, we have limited resources from our team. It’s a smaller team than some of our competitors. And ultimately we want to make sure that we allocate our resources towards the companies that we think have the greatest chance of success. So ultimately we believe like, Hey, let’s focus on companies right now where they’re well aligned with Microsoft as well. And they’re well positioned in the life stage where we can help them. Eventually we’d like to obviously help every company out there and help every entrepreneur. That’s our goal. That’s Microsoft and Satya’s goal is to empower everyone to do more, but where we’re going to focus right now is around where we think we can help the most.

Ted Roduner:

Great. I think that’s time. I’d like to thank everybody for participating. I’d like to thank you guys for taking part. And with that, we’ll, we’ll say goodbye.

Want to learn more?

Schedule a personalized product demo to see:

  • Threat analytics, insights and reporting
  • Automated Triage, Investigation and response tools
  • Platform integrations into SIEM, SOAR
  • …and more
Automated Triage, Investigation and response tools

Want to learn more?

Schedule a personalized product demo to see:

  • Threat analytics, insights and reporting
  • Automated Triage, Investigation and response tools
  • Platform integrations into SIEM, SOAR
  • …and more