(bright orchestral music) >> Hello everyone, and welcome to theCUBE Conversations here in Palo Alto Studios. For theCUBE, I'm John Furrier, the co-founder of SiliconANGLE Media inc. My next guest is Bruce Arthur, who's the Vice President of engineering at Banter.ai. Good friend, we've known each other for years, VP of engineering, developer, formerly at Apple. >> Yes. >> Worked on all the big products; the iPad-- had the the tin foil on your windows back in the day during Steve Jobs' awesome run there. Welcome to theCUBE. >> Thank you, it's good to be here. >> Yeah, great, you've got a ton of experience and I want to get your perspective as a developer, VP of engineering, entrepreneur, you're doing a startup around AI. Let's have a little banter. >> Sure. >> Banter.ai is a little bit a chat bot, but the rage is DevOps. Software really models change, infrastructure as code, cloud computing. Really a renaissance of software development going on right now. >> It is, it's changing a lot. >> What's your view on this? >> Well, so, years and years ago you would work really hard on your software. You would package it up in a box and you'd send it over the wall and you hope it works. And that seems very quaint now because now you write your software, you deploy it the first day, and you change it six times that day, and you're A/B-testing it, you're driving it forward, it's so much more interactive. It does require a different skillset. It also doesn't, how do I say this carefully? It used to be very easy to be craft, to have high craft and make a very polished product, but you didn't know if it was going to work. Today you know if it's going to work, but you often don't get to making sure it's high quality, high craft, high value. >> John: So, the iteration >> Exactly, the iteration runs so fast, which is highly valuable, but you sort of just a little bit of you miss the is this really something I am proud of and I can really work with it because you know, now the product definition can change so quickly, which is awesome but it is a big change. >> And that artisan crafting thing is interesting, but now some are saying that the UX side is interesting because, if you get the back end working, and you're iterating, you can still bring that artisan flavor back. We heard that cloud computing vendors like Amazon, and I was just in China for Alibaba, they're trying to bring this whole design artisan culture back. Your thoughts on the whole artisan craft in software, because now you have two stages, you have deploy, iterate, and then ultimately polish. >> Right, so, I think it's interesting, it used to be, engineering is so expensive and time-consuming. You have to design it upfront and you make one version of it and you're done. That has changed now that engineering has gotten easier. You have better tools, we have better things, you can make six versions and that used to be, so back in the day at Apple, you would make six versions, five of which Steve would hate and throw out, and eventually they would get better and better and better and then you would have something you're proud of. Now those are just exposed. Now everybody sees those, it's a very different process. So you, I think, the idea that you. Engineering used to be this scarce resource. It's becoming easier now to have many versions and have more engineers working on stuff, so now it is much more can I have three design teams, can they compete, can they make all good ideas, and then who's going to be the editor? Who evaluates them and decides I like this from this one, I like that, and now let's put this together to make the right product. >> So, at Apple, you mentioned Steve would reject, well, that's well-documented. >> Sure. >> It's publicly out there that he would like, really look at the design-side. Was it Waterfall-based, was it Agile, Scrum, did you guys, was it like, do you lay it all out in front of him and he points at it? What were some of the work flows like with Steve Jobs? >> So, when he was really excited about something he would want to meet with them every week. He'd want to see progress every week. He'd give lots of feedback every week, there'd be new ideas. It was very Steve-focused. I think the more constructive side of it was the design teams were always thinking about What can we build, how do we put it in front of him, and I remember there was a great quote from a designer that said. It's not that Steve designs great things, it's that you show him three things, and if you throw him three bad things, he'll pick the least bad. If you show him three great things, he'll pick the most great, But it's not, it was more about the, you've got to iterate in the process, you've got to try ideas, you take ideas from different people and some of them, like, they sound like a great idea. When we talk, it sounds really good. You build it, and you're like, that's just not, that's just not right. So, you want, how do I say this? You don't want to lock yourself in up front. You want to imagine them, you want to build them, you want to try 'em. >> And that's, I mean, I've gotten to know the family over the years, too, through some of the Palo Alto interactions, and that's the kind of misperception of Steve Jobs, was that he was the guy. He enabled people, he had that ethos that-- >> He was the editor, it's an old school journalism metaphor, which is, he had ideas, he wanted, but he also, he ran the team. He wanted to have people bring their ideas and come in. And then he decided, this is good, this is not. That's better, you can do better, let's try this. Or, sometimes, this whole thing stinks. It's just not going anywhere. So, like, it was much more of that. Now it's applied to software, and he was a marketing genius, about sort of knowing what people were going to go for, but there was a little bit of a myth for it, that there's one man designing everything. That is a very saleable marketing story. >> The mythical man. (laughs) >> Well, it's powerful, but no, there's a lot of people, and getting the best work of all those people. >> I mean, he's said on some of the great videos I've watched on YouTube over the years, Hire the best people, only work with the best, and they'll bring good stuff to the table. Now, I want to bring that kind of metaphor, one step further for this great learning lesson, again it's all well-documented on YouTube. Plenty of Steve videos there, but now when you go to DevOps, you mention the whole quality thing and you got to ship fast, iterate, you know there's a lot of moving fast break stuff as Zuckerberg would say, of Facebook, although he's edited his tune to say move fast and be reliable. (laughing) Welcome to the enterprise, welcome to software and operations. This is now a scale game at the enterprise side 'cause, you know, you start seeing open source software grow so much now, where a lot of the intellectual property might be only 10% of software. >> Right. >> You might be using other pieces. You're packaging it so that when you get it to the market, how do bring that culture? How do you get that innovation of, Okay, I'm iterating fast, how do I maintain the quality. What are some of your thoughts on that? Because you've got machine learning out there, you've got these cool things happening. >> Yup. So, you want, how do I say this? You just, you really need to leave time to schedule it. It needs to be in your list. There's a lot of figuring out what are we going to build and you have to try things, iterate things, see if they resonate with consumers. See if they resonate with people who want to pay. See if they resonate with investors. You have to figure than out fast, but then you have to know that, okay, this is a good prototype. Now I have to make it work better because the first version wouldn't scale well, now it has to scale, now it has to work right for people, now you have to have a review of: here's the bugs, here's the things that are not working. Why does this chatbot stop responding sometimes? What is causing that? Now, the great story is, with good DevOps, you actually have a system that's very good at finding and tracking those problems. In the old world, so the old world with the shrink-wrap software, you'd throw it over the fence. If it misbehaves, you will never know. Today you know. You've got alerts, you've got pagers going off, you've got logs, >> It's instrumented big-time. >> Yeah, exactly, you can find that stuff. So, since you can actually make, you can make very high-quality software because you have so much more data about what's going on with it, it's nice. And actually, chatbot software has this fascinating little side effect, with, because it's all chats and it's all text, there are no irreproducible bugs. You can go back and look at exactly what happened. I have a recording, I know exactly what happened, I know exactly what came in, I know what came out, and then I know that this failure happened. So, it's very reproducible, sort of, it's nice you can, it doesn't always work this way, but it's very easy to track down problems. >> It's event-based, it's really easy to manage. >> Exactly, and it's just text. You can just read it. It's not like I have to debug hacks, it's just these things were said and this thing died. >> No core dumps. (laughs) >> No, there's nothing that requires sophisticated analysis, well the code is one thing, but like, the sequence of events is very human-readable, very understandable. >> Alright, so let's talk about the younger generation. So, we've been around the block, you and I. We've talked, certainly many times around town, about the shifts, and we love these new waves. A lot of great waves coming in, we've seen many waves. What's going on, in your mind, with the younger generation? Because this is a, some exciting things happening. Decentralized internet. >> Bruce: Yup. >> There's blockchain, getting all the attention. Outside of the hype, Alpha VCs, Alpha engineers, Alpha entrepreneurs are really honing in on blockchain because they see the potential. >> Sure. >> Early people are seeing it. Then you've got cloud, obviously unlimited compute potentially, the new, you know, kind of agile market. All these young guys, they never shipped, actually never loaded Linux on a server. (laughing) So, like, what are you seeing for the younger guys? And what do you see as someone who's experienced, looking down at the next, you know, 20 year run we see. >> So, I think what I see that's most exciting is that we now have people solving very non-technical problems with technology. I think it used to be, you could build a computer, you could write code, but then, like, your space was limited to the computer in front of you. Like, I can do input and outputs. I can put things on the screen, I can make a video game, but it's in this box. Now everyone's thinking of much bigger, Solving bigger problems. >> John: Yeah, healthcare, we're seeing verticals. >> Yeah, healthcare's a massive one. You can, operation things, shipping products. I mean, who would've thought Amazon was going to be delivering things, basically. I mean, they're using technology to solve the physical delivery of objects. That is, the space of what people are tackling is massive. It' no longer just about silicon and programming, it's sort of, any problem out there, there's someone trying to apply technology, which is awesome and I think that's because these people these youngsters, they're digital natives. >> Yeah. >> They've come to expect that, of course video conferencing works, of course all these other items work. That I just need to figure out how to solve problems with them, and I'm hopeful we're going to see more human-sized problems solved. I think, you know, we have, technology has maybe exacerbated a few things and dislocated, cost a lot of people jobs. Disconnected some people from other sort of stabilizing forces, >> Fake news. (laughs) >> Fake news, you know, we need-- >> John: It's consequences, side effects. >> I hope we get people solving those problems because fake news should now be hard to solve. They'll figure it out, I think, but, like, the idea is, we need to, technology does have a bit of a responsibility to solve, fix some of the crap that it broke. Actually, there's things that need, old structures, journalism is an old profession. >> Yeah. >> And it used to actually have all these wonderful benefits, but when the classified business went down the tubes, it took all that stuff down. >> Yeah. >> And there needs to be a venue for that. There needs to be new outlets for people to sort of do research, look things up, and hold people to account. >> Yeah, and hopefully some of our tools we'll be >> I hope so. >> pulling out at Silicon Angle you'll be seeing some new stuff. Let's talk about, like just in general, some of the fashionable coolness around engineering. Machine learning, AI obviously tops the list. Something that's not as sexy, or as innovative things. >> Sure. >> Because you have machines and industrial manufacturing plant equipment to people's devices. Obviously you worked at Apple, so you understand that piece, with the watch and everything. >> Yup, >> So you've got, that's an internet, we're things, people are things too. So, machines and people are at the edge of the network. So, you've got this new kind of concept. What gets you excited? Talk about how you feel about those trends. >> So, there's a ton going on there. I think what's amazing is the idea that all these sensors and switches and all the remote pieces can start to have smarts on them. I think the downside of that is some of the early IoT stuff, you know, has a whole open SSL stack in it. And, you know, that can be out of date, and when you have security problems with that now your light switch has access to your tax returns and that's not really what you want. So, I think there's definitely, there's a world coming, I think, at a technical level, we need to make operating systems and tools and networking protocols that aren't general purpose because general purpose tools are hackable. >> John: Yeah. >> I need to have a sensor and a switch that know how to talk to each other, and that's it. They can't rewrite code, they can't rewrite their firmware, they can't, like, I want to be able to know that, you have a nice office here, if somebody came in and tried to hack your switches, would you ever know? And the answer's like, you'd have no idea, but when you have things that are on your network and that serve you, if they're a general, if they're a little general purpose computing device, they're a mess. Like, you know, a switch is simple. A microphone, a microphone is simple. There's an output from it, it needs, I think we, >> So differentiated software for device. >> Well, let's get back to old school. You studied operating systems back in the day. >> Yeah. >> A process can do whatever the hell it wants. It can read from memory, it can write to disk, it can talk to all these buses. It's a very, it can do, it's very general purpose. I don't want that in my switch. I want my switch to be sort of, much more of these old little micro-controller. >> Bounded. >> Yeah, it's in a little box. I mean, so the phone and the Mac have something called Sandbox, which sort of says, you get a smaller view of the world. You get a little piece of the disk, you can't see everything else, and those are parts of it, but I think you need even more. You need, sort of, this really, I don't want a general purpose thing, I want a very specific thing that says I'm allowed to do this and I'm allowed to talk to that server; I don't have access to the internet. I've got access to that server. >> You mentioned operating systems. I mean, obviously I grew up in the computer science genre of the '80s and you did as well. That was a revolution around Unix. >> Yes. >> And then Berkeley, BSD, and all that stuff that happened around the systems world, operating systems, was really the pioneers in computing at that time. It's interesting with cloud, it's almost a throwback now to systems thinking. >> Bruce: It's true, yeah. >> You know, people looking at, and you're discussing it. >> Bruce: Yeah, Yeah. >> It's a systems problem. >> Yeah, it is. >> It's just not in a box. >> Right, and I think we witnessed the, let's get everyone a general purpose computer and see what they can do. And that was amazing, but now you're like I don't want everything to be a general I want very specific, I want very little thing, dedicated things that do this really well. I don't want my thermostat actually tracking when I'm in the house. You know, I want it to know, eh, maybe there's someone in the house, but I don't want it to know it's me. I don't want it reporting to Google what's going on. I want it to track my temperature and manage that. >> Our Wikibon team calls the term Unigrid, I call it hypergrid because essentially it's grid computer; there's no differentiation between on-premise and cloud. >> Right. >> It's one pool of resource of compute and things processes. >> It is, although I think, and that's interesting, you want that, but again you want it, how do I say this? I get a little nervous when all of my data goes to some cloud that I can't control. Like, I would love if, I'll put it this way. If I have a camera in my house, and imagine I put security cameras up, I want that to sort of see what's going on, I don't want it to publish the video to anywhere that's out of my control. If it publishes a summary that says, oh, like, someone came to your door, I'm like, okay, that's a good, reasonable thing to know and I would want to get that. So, Palo Alto recently added, there's traffic cameras that are looking at traffic, and they record video, but everyone's very nervous about that fact. They don't want to be recorded on video. So, the camera, this is actually really good, the camera only reports number of cars, number of bikes, number of pedestrians, just raw numbers. So you're pushing the processing down to the end and you only get these very anonymous statistics out of it and that's the right model. I've got a device, it can do a lot of sophisticated processing, but it gives nice summary data that is very public, I don't think anyone's really >> There's a privacy issue there that they've factored into the design? >> Yes, exactly. It's privacy and it's also the appropriateness of the data, you don't want, yeah, people don't want a camera watching them when they go by, but they're happy and they're like, oh, yeah, that street has a big increase in traffic, And there's a lot of, there were accidents here and there's people running red lights. That's valuable knowledge, not the fact that it's you in your Tesla and you almost hit me. No. (laughs) >> Yeah, or he's speeding, slow down. >> Exactly, yeah, or actually if you recorded speeders the fact that there's a lot of speeding is very interesting. Who's doing it, okay, people get upset if that's recorded. >> Yeah, I'm glad that Palo Alto is solving their traffic problem, Palo Alto problems, as we say. In general, security's been a huge issue. We were talking before we came on, about just the security nightmare. >> Bruce: Yes. >> A lot of companies are out there scratching their heads. There's so much of digital transformation happening, that's the buzzword in the industry. What does that mean from your standpoint? Because engineers are now moving to the front lines. Developers, engineering, because now there's a visibility to not just the software, it's an end goal. They call it outcome. Do you talk to customers a lot around, through your entrepreneurial venture, around trying to back requirements into product and yet deliver value? Do you get any insight from the field of kind of problems, you know, businesses are generally tryna solve with tech? >> So, that's interesting, I think when we try to start tech companies, we usually have ideas and then we go test that premise on customers. Perhaps I'm not as adaptable as I should be. We're not actually going to customers and asking them what they want. We're asking them if this is the kind of thing that would solve their problems. And usually they're happy to talk to us. The tough one, then, is then are they going to become paying customers, there's talking and there's paying, and they're different lines. >> I mean, certainly is validation. >> Exactly, that's when you really know that they care. It is, it's a tough question. I think there's always, there's a category of entrepreneur that's always very knowledgable about a small number of customers and they solve their problems, and those people are successful and they're often, They often are more services-based, but they're solving problems because they know people. They know a lot of people, they know what their paying point are. >> Alright, so here's the real question I want to know is, have you been back to Apple in the new building? >> Have I been to, I have not been in the spaceship. (laughing) I have not been in the spaceship yet. I actually understand that in order to have the event there, they actually had to stop work on the rest of the building because the construction process makes everything so dirty; and they did not want everyone to see dirty windows, so they actually halted the construction, they scrubbed down the trees, they had the event, and now it's, but now it's back. >> Now it's back to, >> So, I'll get there at some point. >> Bruce Arthur it the Vice President of Banter.ai, entrepreneur, formerly of Apple, good friend, Final question for you, just what are you excited about these days and as you look out at the tooling and the computer science and the societal impact that is seen with cloud and all these technologies, and open source, what do you, what are you excited about? >> I'm most excited, I think we actually have now enough computing resources and enough tools at hand that we can actually go back and tackle some harder computer science problems. I think there's things that used to be so big that you're like, well, that's just not, That's too much data, we could never solve that. That's too much, that would take, you know, that would take a hundred computers a hundred years to figure out. Those are problems now that are becoming very tractable, and I think it's been the rise of, yeah, it starts with Google, but some other companies that sort of really made these very large problems are now tractable, and they're now solvable. >> And open source, your opinion on open source these days? >> Open source is great. >> Who doesn't love more code? (laughs) >> Well, I should back this up, Open source is the fastest way to share and to make progress. There are times where you need what's called proprietary, but in other words valuable, when you need valuable engineers to work on something and, you know, not knowing the providence or where something comes from is a little sticky, I think there's going to be space for both. I think open source is big, but there's going to be-- >> If you have a core competency, you really want to code it. >> Exactly, you want to write that up and you-- >> You can still participate in the communities. >> Right, and I think open source is also, it's awesome when it's following. If there's something else in front, it follows very fast, it does a very good job. It's very thorough, sometimes it doesn't know where to go and it sort of meanders, and that's when other people have advantages. >> Collective intelligence. >> Exactly. >> Bruce, thanks for coming on. I really appreciate it, good to see you. This is a Cube Conversation here in the Palo Alto studio, I'm John Furrier, thanks for watching. (light electronic music)