CEO and Founder of Enterprise Browser startup Island, Mike Fey, talks about entrepreneurship, innovation, and the future of web browsing. Paul explores Mike's experiences working with venture capitalists like CyberStarts and Sequoia, and startup-advising CISOs, getting early customer feedback during the ideation, seed, and early growth stages. Mike describes the origin story behind Insland and Enterprise Browsers.
Mike and Paul discuss AI, ChatGPT, and what new applications we may see AI used for. Mike explains the issue with miseducating neural networks, and how AI will change building technology, along with its dangers. Mike also riffs on a myriad of technology topics from ChatGPT to quantum computing, Web3, robotic process automation (RPA), and more.
Check out Island.io to learn more about their enterprise browser, or reach them on Twitter @island_io.
Mike Fey can be found on LinkedIn.com/in/michaelfey.
Send feedback to host Paul Shomo on Twitter @ShomoBits or connect on LinkedIn.com/in/paulshomo.
CEO and Founder of Enterprise Browser startup Island, Mike Fey, talks about entrepreneurship, innovation, and the future of web browsing. Paul explores Mike's experiences working with venture capitalists like CyberStarts and Sequoia, and startup-advising CISOs, getting early customer feedback during the ideation, seed, and early growth stages. Mike describes the origin story behind Insland and Enterprise Browsers.
Mike and Paul discuss AI, ChatGPT, and what new applications we may see AI used for. Mike explains the issue with miseducating neural networks, and how AI will change building technology, along with its dangers. Mike also riffs on a myriad of technology topics from ChatGPT to quantum computing, Web3, robotic process automation (RPA), and more.
Check out Island.io to learn more about their enterprise browser, or reach them on Twitter @island_io.
Mike Fey can be found on LinkedIn.com/in/michaelfey.
Send feedback to host Paul Shomo on Twitter @ShomoBits or connect on LinkedIn.com/in/paulshomo.
Actually the only reason I do this show and write about the startup space because while the industry is full of imitators talking to founders, startup advising cisos, this is the most densely packed part of original genius in cybersecurity. Could you talk a little bit about your origin story dealing with you know the venture capital analyst, the advising cisos? Yeah. So let's start with I'm such a believer when you just said. Before Dan and I did anything. I was actually Dan and I and Brian Kenyon. We called more than 50 what we thought were industry leaders. And they happened to be cisos and CIOs as some of the most important companies on the planet. And I left the same voicemail or text for each of them. Listen, I'm thinking about starting a company. I want to go over what I'm doing and see if I'm off base. What was amazing? Every one of them took the call. The same people that we really hard to sell to. We're happy to appreciate listen. This individual is going to go off and start a company. I'm happy to give them advice before they go and direct their career down a path that might fail. The genealogy of cybersecurity is a new kind of podcast. Here we'll interview notable entrepreneurs, startup advising cisos, venture capitalists, and more. Our topic, the problems of cybersecurity, new attack surfaces, and innovation across the startup world. Welcome. I'm your cybersecurity analyst, Paul shomo. So today we're going to bring back Mike Faye, who is the CEO and founder of island that was our episode three pretty popular episode on the enterprise browser space and of course island in particular and Mike you know, he was the head of semantic and blue code before that as well as what he's doing now with island. And so some of these entrepreneurs and founders I'm going to have back on to talk about just general entrepreneurship, working with venture capitalists and the general topic of innovation and where things are going and Mike was kind enough to give us some time for that. So that's what we're going to listen to right about now. Very interesting stuff that this whole area of the virtual assistants, I mean, chat GPT is obviously caused a big stir. I mean, AI, it's hard to tell but it does appear the AI maybe just stepped onto an exponential evolutionary curve. So I wanted to ask you how different do you think browsing is going to be in the future? Well, I think it's I think it will change dramatically. I think it already is. You know, I think there are a couple of big things headed our way that are going to change our situation epically. AI is one you know. You think about just searching. The theory is with AI that you ask a question and get an answer. Not back a list of paid for search results plus the answer you're in somewhere in that journey, right? So that whole advertising model gets disrupted. There's a lot of free productivity stuff that have been given to us because of that advertising model. So who's paying for that moving forward you know? How does that play? But then when we think about that AI, we also have to think about how data gets up. Let's take a step back and just take something as easy as Grammarly. You know, my writing is horrible. I have no skills in that. And anybody that wants to help me write better, I appreciate, but I sign contract after contract that I will not let customer data move along. And if I assume that everything I write, every bit of customer data interact with gets uploaded to a cloud to get scrubbed. I may or may not be breaking the contracts I sign. So you see people block those extensions. And now you've seen that with OpenAI. And I don't want to be careful. I don't want to talk about any particular company, but when you're asking a AI engine to do things for you, to look at data, it doesn't forget what it saw. It's learning from you as well. So if you have a bank, actively engaged with that, it makes sense someone else can ask questions to learn about what you actively did. And start to kind of understand that and start to get a picture of that. So what we upload to that, there's almost going to be another set of gateway technology that has to show up to govern what it is we're willing to send into a public cloud. Which may spawn private clouds that are more open. Who knows? But I also think it's not going to be one AI engine you know. AI at its purist is like having a bunch of smart Friends. And you know who to call up for an answer. Right. And right now, we call one person. And we get the best shot. But after you have more friends, they start this guy really knows the restaurants. And this one's really on movies. This one knows cars. And you kind of call your Friends and you get the best advice. That's where AI is going to end. It appears to me is this kind of arbitrage of dedicated information. How we send data to that? What we ask for that? What's confidential? Well, it's not is all very interesting part of the journey. That's one of the big changes. The second big change is quantum compute. That one will be big in the enterprise. Think of it like this. The minute quantum compute is available online for reasonable price. That means it's available to hackers for a reasonable price. And this is available today. That means I can break the encryption stream of even the most important data. So every self respecting bank and the like at some point will have to say, you know, we're going to up our encryption algorithm. The mathematical jump to that is far superior than what the hardware organizations are ready to do. So this whole breaking inspect that is core to network security dies almost instantaneously like that because they're performance and expense. So the idea backhauling traffic in a quantum compute world starts to become really interesting and really difficult. Then you throw on peer to peer and Web3.0, and you just, what are we really doing here? So those two in concert start to become it. And then you throw on the privacy laws. All three are going to make it really interesting on how you navigate this beautiful world that's being built. And having your main operating system uncontrolled on instrumented ungoverned is not how you're going to see your way through that process. I just recorded an episode with HiddenLayer. One of the innovation sandbox finalists this year. And they protect machine learning systems from attack, miter has a new framework, it's beyond minor attack. It's called MITRE Atlas and steals basically covers attacks on machine learning systems. And you kind of touched on this. One of the weird things about adversarial AI attacking is that it can learn it's basically learning what your proprietary secrets are, things that you don't want people to know. I mean, that's really what deep learning is. If there's a vendor and you're using the product that does deep learning, it's learning your dataset and about you and it's taking that intelligence back to a central location. You know, there are certainly things the AI vendors can do to help. They can try to mask the requests and not learn and build filters and safe zones and all of that stuff. And I think they will. But inherently, if you continue to think of AI like you do an individual, like you're doing intelligence, you can't sit me down and ask me a series of thought provoking questions. And then expect me to forget them on the next person walks up. And that's really at the heart of it that the best algorithms will be always learning. That would be what makes them amazing. So to think that we're the only ones that get to leverage that, when we make the request, is probably flawed logic, but I will say this. I think what is possible, where it's going is such an early stage, we will be shocked at where it gets to. I don't believe it's inherently evil and it's going to ruin the world by any stretch. I think it is that exponential next productivity game. And honestly, we're due, this idea of search on peck, you know, trust, I just want answers. We just want information in our fingertips. It's all we've ever wanted. I think we can get there. We still have to be creative humans and do great stuff. You mentioned a phrase there that I think is really important. You said turn off learning. So I could see a vendor that's delivering some type of AI solution where I say, okay, you can learn off of me because I'm getting it for free, but I want you to turn off learning and they might say, when I had to pay for it. I don't want you to learn from me. And I think that I think that makes perfect sense, but when you tell an AI system to go review all mortgages written in the public record for a zip code. It did it. A data's been resolved that's been looked at you know, can it really forget? I don't know. And when I worked on neural nets, one of the challenges we had is when we trained it with data, we put their own data in, getting it out was very hard. And you know when you build these learning algorithms, the idea of intense molecular control of what you learn and don't learn is tricky. Now, luckily, there's lots of brilliant people out there that I think have chosen this as their future field. So maybe they'll figure it out. But I do think you're wise to assume when you interact today. That the cloud of anonymity is not where it used to be. Then there's also just the expense of it. You know, how these things charge. Who's using it? When are they calling it? What for? You know, we saw this on Amazon showed up. Every remote admin could call different processes and stuff in your Bill from Amazon would be wildly different. Every quarter based on what a developer chose to use and not use you know. Like are they using the database we have or do they call one of the ones inside of Amazon? Are they using this or that? And so you can see the same thing occurring is my call center worker using an AI tool to do their job when I wasn't planning for that. But they're using my instance. There's also just this control and governance to expense that will have to be there. None of this stuff is free. What such a wildly different world you mentioned feeding bad data to a neural network while you train it? It used to be you had a problem where you amassed too much code and you had a legacy code base that was like there was this inertia to it that caused your product to slowly die. But now you bring up a point that the race is really training your large language models, training your neural network, building that internal intelligence, not building up more features and lines of code. And you brought up a good point about what could be a decay there is miseducating your neural network over time. Well, honestly, I mean, we haven't we had this problem you know. On April 1st, this week, virtually for me, all my social media was pointless. I couldn't tell what was real and what was it? Michael Jordan was coming back to play again. Like all this crazy stuff. And you know whether you want to get in the whole trusted news or the like. What sources do you trust what sources don't you trust? Where do you go for your data? What are you learning on? And that's why I do think you'll get any specific algorithms that are trained off data that we may appreciate. You could imagine if I was trying to get financial AR legal AI for that matter. I want to understand if I'm a large law firm, and I'm using that. Where did you train it on? Did you train on some person with an opinion or some person with a degree? Some person that has tried cases or somebody that watches them on TV. Where are you getting this opinion set? Because when I search for important data, that's one of the hardest things to really believe is, what is this site? And is it trustable? Well, it's mining those same sites for information. So I think that's a whole nother problem, differentiation that these AI engines are going to give, which is their pedigree. What do they allow the data to come from? Sources, I think, is a good way to summarize that. I think that is a will be a huge thing in the future is, okay, you learn this, where did you learn it from? I want to see. I mean, that's what we as humans do. We cite sources yeah. It's an important part of the, like you said, the pedigree of knowledge. You know, it's funny though, in your daily decision making, how many things do you just know and you don't know why you know? Well, one of the things that's interesting about setting sources too is citations obviously has been a big part of academic publishing. Where did that great idea or that theory come from? And then you can kind of track back in time who got the most citations and they became the most celebrated intellectuals. And so if AI starts to basically note where things originally came from, suddenly you get to start to see that there's a portion of humans that are original genius and the rest are just imitators. AI can tell us which ones it's copying off of. It's a funny idea. You know, I think what it allowed for is a real burst of creativity because the gap to execution will drop. You know, it'll be about the bright idea and less about the execution of it when you can get such good advice on how to complete the last mile effort. But who knows? We're definitely early on and I know when it comes to the enterprise browser, integrating in smart AI. Basic AI to start. Auto populating fields where they need to be done. Just watching a user and go, you know, you'll be copying and pasting this data back and forth for the whole day. Can I just start assuming you want this? I remember the first time I saw that in Excel. And I didn't have to drag a formula just. I had Monday, Tuesday, and drug over and it figured out, oh, he means the rest of the week. I'm sure every friend I have. Imagine if you're sitting in a call center or in a production area or distribution environment, and you were doing the same darn thing for the last three months, and all of a sudden just starts filling in for you. You're going to feel better. You're going to feel it's improvement. You're going to make less mistakes. The productivity will increase so many ways to use it before we head towards you know what if it gets the cure for cancer wrong you know? There's so many wonderful easy things to do. And I think we might lose track of that right now because we're all thinking of the grand plan, but let's just get my DoorDash order right. Let's start there. Well, I focus a lot on original genius. Actually, the only reason I do this show and write about the start of a basis because while the industry is full of imitators talking to founders, startup advising cisos, this is the most densely packed part of original genius in cybersecurity. Could you talk a little bit about your origin story dealing with you know the venture capital analyst, the advising cisos? Yeah. So let's start with I'm such a believer when you just said. Before Dan and I did anything. I was actually Dan and I and Brian kenyon. We called more than 50 what we thought were industry leaders. And they happened to be cisos and CIOs and some of the most important companies on the planet. And I left the same voicemail or text for each of them. Listen, I'm thinking about starting a company. I want to go over what I'm doing and see if I'm off base. What was amazing? Every one of them took the call. The same people that we really hard to sell to. We're happy to appreciate listen. This individual is going to go off and start a company. I'm happy to give them advice before they go and direct their career down a path that might fail. Every one of them took it and they took that call quickly. They didn't push me out weeks. They took it that week. And so you saw these bright minds wanted to help. They wanted to see innovation. They wanted to lean in on that. And so we talked to more than 50 of them when we went through it. And I mean, we're talking hours of conversation and individuals. And then they calls up later and say, I thought about it some more. Someone that they'd call on their friends. Hey, listen, I discussed with my friends. Here's what you need to think about. I mean, it not only was just it was so great to see humankind work that way, that these very busy people to get hounded by every startup on the planet to buy their stuff. We're 100% willing to game theory with you. They didn't want to see another startup go down a bad path, build a product that didn't need to be built. They wanted to be in on that. So when it was time to fundraise, I kid, most companies start and they go hire a couple. They get a small amount of funding to pilot something. We felt like we knew we knew it was needed. We were not confused. We knew exactly what had to be built. We had talked to all the right people that we game theory did out enough. So that's why I went for 20 million out of the gate. I wasn't debating what had to be built. And we knew how far we had to take it. It wasn't to prove out could we build a browser that did a couple of things. No, that was dumb. We had to go to product and we'd sit in front of these great people and they go, yes, you got it. So we went with a very heavy funding route. And with a 200 plus $1 billion TAM, the VCs were excited that somebody was tackling a disruptive big area, not just a second coming of a given tech. The other thing that you know, to be honest, that really helped is they know that everybody involved had experience operating. We were coming from a space we knew well. We didn't need to do it. We didn't decide to become entrepreneurs. We solve a problem, and we wanted to build a company for it. And they love that. They want to experience people that do it you know. A lot of founders are very young. We obviously weren't or later in our career. We'd seen a lot and given what we were tackling, we were the right shape, size, you know, experience level for what we were tackling. If you come from an alternate space, tangential cybersecurity, you didn't really help enterprises solve their problems for 20 years. You're going to build their own features of the gate. You're not solving a complexity. You're going to focus on zero days in the browser and hiccups in the browser. You're going to try to tell people how to be more secure in the browser. In reality, there's so many tools for that. There's so many different ways to do that. And you can sit there and debate that all day. Instead, we thought, what are they failing at? What is their nightmare? What could we solve? What is a big enough solve that they think about this idea that they actually even consider it? And that's where we started. So you know I will tell you our first year of development despite all our customer calls, which we had the 50 to start with and then CyberStarts brought in another hundred customers in Sequoia brought in more insight brought in more in all this feedback we got to get. We didn't change our build order for the first probably 18 months of building the product. When you can let a development work, go build, heads down you know, you're not whipsawing them around. They can get pretty good product out the door. And that's what we're able to do. But back to your starting point, there are so many geniuses in our industry that are operating and hid away from innovation because they're busy. They're busy trying to save the world, right? I mean, some of the people I called, I remember one called. I had to call me back and said, listen, we're dealing with the Chinese right now. They're attacking our organization. Can I call you tomorrow? I thought to myself, what a wonderful individual. If the Chinese were attacking me, I might talk to you next month you know. But all you did tomorrow, it's just they've learned to live in this phenomenal space and having access to them is one of the treasures. And by the way, we kept most of them as advisers to this process. Most couldn't be paid for advisers, most just want to help out as they can. And their insight and direction of where to take it, where to take it next. It's just absolutely priceless. And you mentioned Bob, I mean, he's one of them. I met Bob in this process.