sfspark.org: Jose de Castro, Cisco Spark open platform for AI: More Bots!

Recording: sfspark.org: Jose de Castro, Cisco Spark open platform for AI: More Bots!

I'm Jose pressroom with Cisco I'll start by just telling you a little bit about myself this is when I saw I was going to be like the last talk I kind of changed things up a little bit and so this is not going to be a deeply technical talk I come from the collaboration domain world of voice and video and messaging and you know what we do apply machine learning and AI and various aspects of of our platform we are now just entering a really interesting time in collaboration specifically as natural language processing in the rise of like virtual assistants are actually becoming viable options in the enterprise right and so they've really started to penetrate consumer markets you know with the advent of Syria in Cortana and all these areas and so we spent a lot of we're are spending a lot of time thinking about that but this talk is split up in a couple sides first just want to give you some of my personal insights having done a start-up exited the Cisco and then just kind of seeing this explosion of technology and advancements in technology over the last two or three years and then we'll dive into chatbots you know what they are how our customers are applying them and then you know mine medal was mentioned a second ago and give you a little bit insight into why we acquired my meld and what we plan on doing with that and I think we all go drink beer upstairs so cool so in 2009 I founded a company with a couple cofounders called troppo we effectively wanted to democratize voice video and messaging for a new class of web developer if that sounds familiar to you it's because another company had the exact same idea about two months apart from us called Twilio their Bay Area company and we we were pretty much competing head-to-head for a long part of the duration they took a slightly different route to market we were partnering with service providers like AT&T and Verizon and you know those types of guys you know they want full-bore you know to position themselves as kind of like the new Bay Area telco and very kind of anti-establishment worked out pretty well for them they had a great IPO we ended up May insight Cisco and I'm really glad that we did because our our mission really pretty much at rofo was really less about being the best platform even though obviously we strive to be but really what we wanted to do was create the best possible user experience we wanted to put collaboration and great communication tools in the hands as many users as possible because we felt that we were entering an era where people are going to be working in different and new ways right and he was just talking about the previous speaker about allowing his employees to work fully remote now and so this is this is he attributed it to the culture of his company but this even very large companies are starting to notice that in order to retain talent we have to change the way that we work and I think that's a pretty profound thing now in the middle of all of that in the middle of selling makino starting a company selling my company joining Cisco and taking on this new role I also managed to be a dad for the first time and and so that's my two and a half year old son I think that picture has taken about six months ago but I'll tell you a little story and I promise is going to actually make sense in a bit but about a year ago I woke up and my son was sick and my wife was too and so you know I just started at Cisco and we had a big project too and so I was I was really sweating and I was like man I got to make it into the office I got all these preparations to make and then something hit me like I'm sitting there in the shower and trying to figure out how I'm going to deal with the situation and I was like all right I can do this right what do I need what do I actually need to get done here well you know I need to get me some medicine for the kid but we got to feed ourselves I'm not feeding my kids sushi by the way though so that was for me but like for it you know I got I got him some chicken soup or something and you know some sushi for myself and and then obviously I need to take all my in-person calls that I am eating that I had scheduled and I need to get them over video somehow right and so luckily I work at Cisco and that part was pretty easy and something hit me while I was sitting there fretting in the showers like wait a second within like six minutes I had pulled out my phone and I had just pretty much fixed the day and it occurred to me that this just wasn't possible just a few years ago right it's like I pulled up prime now I order the medicine i order some food you know I went into my calendar app and I reinvented all my meetings to be on Cisco spark and that was that was the joke that he stole from me but yeah our product is actually called Cisco spark and has nothing to do with Apache spark it's but anyway so that got me thinking they've got me thinking a lot about the kind of trajectory that some of these technologies are on and I started getting really curious and I started doing some research on my own before I get into that though you know just just think about how amazing this is i mean just ten years ago there was there's no iphone there's no you know you don't have apps in your pocket and but we also saw the rise of something pretty profound right Amazon Web Services started around the same time really democratizing you know mass Gale compute allowing startups to start small and grow into their infrastructure needs and then in 2013 another really profound thing happens that's about six years later we saw the beginning of this hardware hacker movement all of a sudden Hardware became sexy again she had the introduction of Arduino Raspberry Pi containerization started making its way into the kind of everyday conversations and and this this sharing movement really underscored by by uber probably the biggest success but there were many companies also that we're leveraging the cloud to find like-minded people and through technology creating tremendous amounts of value by sharing resources and that's pretty incredible but if you if you see a trend here both we have three line three plots in America a trend really but if you were to trend this out one interesting thing is happening that there's an acceleration happening here you know it took six years to get from there to there it took three years to get from there to there and you know come 2016 Wow you know just in the last 18 months we've seen an explosion in cognitive services a RvR and I keep waiting for the next shoe to drop I keep waiting for like sitting like a weird bobbles like feels bubbly but it's not but no you talk to our customers and this is one thing that I have access to at Cisco that I wouldn't have as a start-up you know I have access to real enterprises like I can just you know you know I'm an executive at Cisco I can just pick up the phone and I do and I talk to you know other executives at big companies they are they wouldn't take my calls when I was troppo but now that I'm at Cisco they do and I talk to these people and I realized that no it's like they're using this [ __ ] and actually deriving value from it it's this is not just just hype and so that got me really curious about you know where is this going and then you know shortly after that this is again this is this is about a year ago then I started doing this research about eight months ago I stumbled on a really great blog post by a guy called Klaus Schwab you guys familiar with Klaus Schwab at all he's the he's the chairman and founder of the World Economic Forum well-known economist published economist brilliant guy but he's also a futurist and he he published a book I started as a blog post and then he wrote a whole book by this same name and and so what he postulates his theory is that we are at the very beginning at the very cusp of transformation and technology that's punctuated by a a fourth and distinct in a revolution that is separate and apart from the three prior and so you may have heard obviously of you know the other three revolutions here right and so the first one there about 1700 you know first Industrial Revolution we saw the introduction of steam steam power which made mechanical production and mass production started to become viable about a hundred years after that we saw electricity and in true mass production it was possible after that about a hundred years after that we saw the introduction of you know SEM conductors electronics and what is what is commonly referred to as the digital revolution right but if you're counting one two three it's really the third Industrial Revolution and many think or at least many thought before the introduction of this is the set of papers from Klaus Schwab that we were still in the third Industrial Revolution and but there is mounting evidence and if you read the book you'll see that we are at the very beginning of of a truly distinct revolution that brings together the digital and the physical and the biological into into one and he refers to them as cyber physical systems I'm not a huge fan of that term but but anyway some of the drivers 4/4 and dust revolution pretty obvious I think you know we've touched on them a lot a lot today but mostly we've seen exponential advancements in areas like machine learning robotics etc etc I think the key word here is exponential you know one of the most radical ideas in Klaus Schwab's book is that in like you know five years time we will quite literally see advancements in technology at an exponential scale that has never been the case before in the entire course of human history so it's quite possible that in five years because of these advancements you'll go to bed at a reasonable hour hopefully wake up six to eight hours later and have missed you know a dozen massive advancements in medicine a hundred massive advancements in manufacturing semiconductor semiconductor production mathematics fundamental shifts in the world of mathematics and astrobiology and and and and so in a world where technology is advancing at a truly exponential rate again this is not hyperbole right I mean it's a truly exponential it really gets you thinking about what the role of technology will be in our lives and quite frankly what the role of humans will be you know not to be an alarmist or anything like that actually I'm pretty bullish on on this stuff I don't think the you know the bots are going to take over anytime soon but it does beg the question you know where is the value all the value being created by these systems where is that that value can't come from nowhere right the value if these systems are creating value then then that value needs to come from somewhere like consumers are either paying for it or companies are paying for it but there used to be an exchange of value right it's just like energy you can't just create value out of nothing so I'll put a pin in that because I know I've said a lot of stuff here and and excuse me if I get like overexcited because this stuff is pretty exciting to me it's like I really am excited about this stuff but also put a pin in that for a second because because this got me thinking about all kinds of things and one of them was diversity in general and well you know obviously we're lacking in that in this room right now there's not a single woman in here but but but overall overall there is a positive trend there is a positive trend in diversity in general in the workplace in general and you can say in tech as well yeah exactly yes so so I was really pleased I was really pleased when I joined Cisco that Cisco has a lot of really great programs for improving diversity within the company but also externally as well I'll put some links out there at the end of my talk that you could you could go check out a couple of them are really good and they're open to the public and I really encourage that everyone look at that but but I think even more alarming than the current state of diversity in tech and in general in the workplace it's something that isn't necessarily a a back-breaking issue right now but that certainly will become one as these advancements that I discussed do accelerate and that's that's a growing inequality and and typically when people talk about inequality you think about financial inequality that's how it's normally referred to and but really what I'm what I'm really referring to when I say inequality a skills gap that is being that is being created by by these advancements in technology and I'm definitely you know this you know done a bunch of tech startups I'm definitely not like standing here in front of you and somehow like moralizing technology or trying to immortalize technology and saying that that we that we need to necessarily slow down because of night some implied negative consequence to advancements in technology but but this chart and I don't have the link on here damn III can get that out to everyone but this part this chart is part of a broad study that was done to to study the effects of computerization and automation on society and and so what this chart is showing is that so on the left hand side you have these are all white-collar jobs by the way there's these are all white-collar jobs that are being impacted by computerization or that may be impacted by computerization and and on the left side of the chart you have low skill low paid jobs and the right side of the chart you have high skill high paid jobs and and what you'll see is that there's no [ __ ] jobs in the middle we are seeing a massive erosion in what we consider to be the ability for someone to live a middle-class lifestyle right and that's not because it people love people like to blame politicians and policy because of that but it's just it's not it's just the jobs aren't there like because the value is shifting here and then it's and then the value that was being generated here is being just automated it's just not being done by humans anymore and so is this good or bad like I think that's that's probably discussion for a beer after this talk or not and so again I'm not drawing conclusions from this alone but the one thing that I decided for myself and you guys can take or leave this and have your own opinion but it's that as a as an innovator myself and as an entrepreneur I feel that we have at the very least a moral imperative to just build systems that are mindful build system that when I say systems not just software systems but like as a whole like we're building when we build software platforms where we are imposing a way of working on to the world and if we're successful then a lot of people are going to adopt those principles that we're putting forward and and so we have a moral imperative to be mindful I think okay so that's what I was my soapbox and I'm going to like officially step off of it right now and I'm going to get into kind of the meat of this talk which is how does all this apply to my domain which is collaboration and so right here you have you know beautiful plug for our products so this is Cisco spark it's badass you should download it and try it and basically it's a enterprise messaging app you can think of it like stock for example it has bots in it all kinds of cool stuff there are a lot of reasons why we think it's better than slack for example its end and encrypted so cisco is not going to be sleeping on your messages actually we can't even snoop on your messages if we wanted to but also it has beautiful HD video it's compatible with all of our awesome hardware and it's a great tool and so so my role at Cisco is I don't build these products I look after the the platform side of it so the cloud infrastructure that powers all these great products and my job is to expose that platform to third-party developers that want to do cool things with it so if you want to embed video into your app or embed encrypted super secure messaging into your app then that's great but another part of my job is a growing number of developers who don't want to build their own apps but they want to participate in the value that's being created on these types of platforms right and a lot of those a lot of that value is being created through or being manifested through these chat BOTS right I'm sure you had to have been living under a rock to not have interacted with a chat bot you know in at all right it's not it's intentionally black yeah so yes exactly so so I'm going to talk a little bit about chatbots some of the best practices that we've seen out there and what the applications of AI and machine learning we feel are have strong applicability in this domain and then I'll give you some links to resources if you want to learn more after so we've got plenty of time and then so before I dive in directly into that you know what you won't see this in any Cisco literature but when my family asked me like what I do and what types of products my company builds sometimes struggle to talk about it right you know I can use the analogy oh it's like flock to you guys and you probably know what I mean but I can't say that to like my mom or my sister but also even this is like my other kind of professional friends and and so you know I've coped with that I've come up with like these little ways of like you know relating it to people and maybe si it's like kind of like Skype or or whatever but that got me thinking a lot of also about what is really the core value of collaboration products and what I what I kind of zeroed in on is that it's really two things people pay us money people pay Cisco or by Cisco collaboration product for two reasons they want their employees to share knowledge with each other and or share knowledge with their customers so you do that you know you share your screen you talk about things and you convey information and then that information is kind of you know retained so put quite simply and and at the end of that you want to Rive at some sort of decision so whether it's like you and a customer you agree to like a sale or internally you're sharing knowledge and you're like oh wow that's a great idea we should do that and then you make a decision and ultimately like when you buy WebEx oh you buy our fancy telepresence systems or whatever that's really what you're paying for the end of the day and so really that the whole advent of chatbots is an effort to either what do want a few things one is assist humans a helping to arrive at better decisions or derive better decision-making or take a human out of the equation on one side of that right so it's a human absorbing knowledge from some automata right and but still with the exact same goal right we want to perpetuate and spread knowledge and we want to derive better decisions and so what's the chatbot most you guys probably probably know what they are but basically they they show up as regular people in your whatever chat app that you're using but they're backed by a computer program instead and so they normally have like a nickname you'll have an avatar and they all intents and purposes look like another user on the platform but when you interact with them it's really just a computer program that's taking the text that you're typing or in some cases like we support you know video BOTS you can take in your video stream and it can do interesting things with it as well and what we found is that generally chat BOTS are classified into two main areas you have ones that are what we call command and control so they're just kind of passively sitting there just waiting for a command and there are other ones that are more conversational in nature you know Siri is a good example of command and control right Siri has no memory if I say if I ask Siri what's the weather in Miami today it's very happy to tell me but if after that I say hey Siri what about Boston like what about Boston it's a nice place I guess yes but I just asked you what was the weather in Miami why can't you connect those two writes good Siri doesn't have a memory and yeah let's get the head a little bit oops not too much okay sorry about that and and so recent advancements in machine learning have made it possible to solve that problem so you actually build systems that are truly conversational in nature and that was the main motivation for Cisco recently acquiring my milk they're a SAN francisco-based company is about 30 employees really great great bunch of people and they've spent the last seven years solving that problem they've solved they have built a natural language processing engine and a set of tooling around it and a consistent methodology more importantly for building conversational a is that have near human accuracy and by near human I'm saying you know 95% plus accuracy and turns out that's a really really hard problem to solve and so we're super glad to have them on board first and foremost we plan on using their technologies to bake them into our own products so that you can walk up to a Cisco product and be like you know ok spark you know help me schedule a meeting I want to meet with Bob and Jane you know in two Tuesday's from now and it can walk you through the whole process because scheduling is a very difficult problem it probably requires multiple interactions back and forth to actually solve them so being in the business of meetings and collaboration you can kind of extrapolate from there like what are the types of domains that we'll be working on and so we'll be leveraging the my male team to do that but what has me also pretty excited is the potential to open that up allow developers to build skills on top of our AI and again this is not something new you can take Alexa and teach it new skills today it turns out that doing that in a secure way it's actually pretty challenging I'll talk a little I'll talk about that in a little bit you know I snuck a couple slides in here and since I have some time I'll go through them as well so I talked a little bit about chat BOTS one of the interesting trends around chat bots in general is this concept of chat ops he has quick show of hands anyone heard of the term chat offs yeah one two okay a couple of years well it's great so I love this I love this this definition actually it's from one of the engineers over at Atlassian and says their chat ops is a collaboration modality that connects people tools process and automation into a transparent workflow and it connects the work needed to work happening and the work done in a persistent location staffed by people bots and related and I think it's a pretty insightful way of describing this movement I'll talk about I think it's better expressed through some examples so you know people are doing chat ops so basically chat offs is you get a bunch of people into a chat room and then you add you start adding BOTS and systems that can represent themselves via a text modality as well and you're leveraging the fact that this is being done in a linear chat modality to find ways in which a work flow some sort of business activity or engineering activity can be expressed in a cleaner more logical and preferably more natural way just as if you were all sitting together but the nice thing about is because you have these systems in these BOTS there they have some nice attributes that you don't have in like an in-person conversation I'll walk through some of those but yeah obvious examples of where this may be helpful is like a document approval flow or if you're practicing agile at your companies and you do stand-ups regularly like engineering stand-ups or whatever you know having a bot that can enforce like Jayne your term two minutes yeah no Bob because you know everyone talks over at the stand up so it should be a 15-minute stand up and so it being like a 45 minute design session we have a stand up pot that makes that not even possible yeah we literally mute your mic after your two minutes and we and then we pin the video feed over to the next person whose turn it is to talk and no one has to feel like an [ __ ] because you told Bob to be quiet it's just the bot the bot did it sorry man but our meetings end on time yes yeah lots of other examples here and so why are chat ops why our chat ups on the rise what what is so interesting about them that they've been prolific or that they've been growing so much well so they're about four main benefits to the chat ops and I'll walk through them first is these messaging tools you know Cisco spark you know things like slack and Microsoft teens and a few of them they they typically have strong identity so you know if I see your avatar and I kind of click on it and I kind of see you know it's either backed by an email address or a username or whatever I can be pretty sure that that's who I'm talking to some systems have stronger secured some stronger identity than others so in Cisco spark we do a real-time key exchange each user has their own key and certificate and so we have extremely high like strong identity in Cisco sparks you can be pretty damn sure that who you think you're talking to is actually the person that you're talking to and that becomes pretty important when you're when a bot is sitting there in a chat room and it's receiving commands from someone to like you know spin up another cluster take that cluster down move this customer from like site a to site B or you know I I CEO of my company approve this document and so identity's is extremely important and thankfully most of these messaging systems have pretty good identity second one and this is something that I didn't realize until after interviewing a couple of our DevOps engineers recently was an interesting challenge so we've got some engineers in Beijing and I don't know if you guys know this or not but there's this thing called the Great Firewall of China that pretty much makes the internet suck in China and and so it's it's pretty restrictive it's AI input but worst of all it's unpredictable things that worked yesterday don't work today and so yeah well yes that doesn't it's not helping their cause but one of the interesting things that we found was our team in Beijing were having trouble connecting to the VPN the Amazon rmr VPN on Amazon's we had a V PC and the kiddin they couldn't connect and you know all the tools they needed to do their job these are DevOps engine it all the tools they needed to do their jobs they weren't accessible to them they couldn't get on the VPN so they couldn't get to all the web tools and all that that we built well guess what and this is before we join Cisco's we were still troppo didn't get to slack just fine so like they couldn't get on the VPN to do their jobs but they were sitting there bitching slack about how they couldn't get on the VPN to do their job and here's a team in a team in the u.s. is like [ __ ] man I don't know what I wanted what you want me to do I need to go to bed like the whole reason we have a team in Beijing so that we can follow the Sun and and so you know we join Cisco and we were like all right well since we were about to join system we're like alright we joined Cisco and and and and this problems going to go away nope the internet still sucks in China as it turns out and but but guess what chat offs to the rescue right what we did was we built a bunch of a lot of the tools that we had that were behind the VPN we're pretty much just Web Apps like our DevOps team they built a bunch of web apps you know in rails and and that's what they use to maintain the network right it's like oh I need to look up customer be is you know ticket comes in from support and they're like oh wow yeah that cluster is shitty or we're seeing some alarms or whatever let's move them up to this move them off to this other cluster or let's scale them out and so you know if I had a bunch of humans that were looking at screens making decisions about the operations of our company and we said man like this doesn't need to be a web page like you just put this in chat and then we can fix the Beijing problem and also like reduce one more tool that we had to like log into and remember a password for and all that and so we did that and it had this dramatic impact on our productivity not so yeah obviously it solved the problem for the Beijing team but the other thing that it gave us was accountability because you know this is like a handful of like DevOps guys writing these web tools you think they're actually like logging really nice audit logs for like everything every button that you're clicking on this no like some guy wrote it some guy had like a perl script and he was like [ __ ] his Perl script sucks and then he's like I'm going to do a web app for it so like you like stayed up all night and like built like a rails app for it and that was it like you know you press the button or whatever any other we're they were logging some of the things that we're going on but it wasn't perfect and one of the things that happens when you move into chat ops it's a chat so when you say like you know cluster bot you know take this cluster out of routing guess what that's an actual message that's in like the chat tool and then cluster bot says yeah sure I'll do that for you and it does it so like if tomorrow all [ __ ] goes breaks all hell breaks loose and like the next guy that's on Pedro Duty wakes up he's like what the hell happened last night he could all you have to do is just scroll up and be like oh then you told the bots to do this and whatever and so you know all this work that was put into like writing these audit records and audit logs we have to sift through logs or whatever now just becomes a natural language problem and and that brings me to the last one which I just touched on as well which is context right so so what happened Oh Bob took the cluster out of routing at you know 11:05 why did he do it well just scroll up a little more and there's a good chance that you're going to see you know Ben or Bob like complaining about something be like oh I just got paged like customer blah blah blah blah and then you see the command to the bottom right so anyways chat offs is an interesting thing because I think it was meant to solve one problem ended up solving many many more so that's great but and so chat bots are cool and so is chat offs but the reality is when you go and you survey the markets just about every bot that's out there is complete garbage like if they don't work really well they get stuck and that's just because I think as an industry the you know we just haven't arrived at a set of usability Pat UX patterns that are just well established on the web right and on the web you have frameworks like Twitter bootstrap and like all these like framework you know jQuery and whatever and now you know a new generation of tools like angular and react that make it they make doing the right thing they also the easy thing to do and we just haven't done that in in the bot world yeah and so that's one of the things that we're doing investing in very much at Cisco part of that was the recent acquisition that we made of mind-meld but also we're working with companies like API that AI gupchup stamp play and a number of others to do a couple things one is you know if if we want to get all the benefits that I was talking about just a second ago one thing that you want to do is you want to build bots that are conversational in nature and obviously one of the two main challenges to doing that are understanding what the users asking you to do so that's natural language processing and then acting on that in a way that is consistent but also you know auditable and and expressive and so with the mind meld technology and the team that we brought on board we're well on our way to solving the first problem and you'll what you'll see from us are a set of tools and frameworks that aren't just compatible with our own products but hopefully many other chat tools as well for making it possible to build truly conversational a is that at the very least understand the semantic meaning of what the user is looking looking to be done and then we plan on publishing a set of guidelines best practices for how to have those BOTS actually act on that information in a way that's expressive so you know I'll close on a couple thoughts one is you know I think more important than you know building great BOTS or even all the problems that these solved going back to the you know you know kind of the the inequality you know rant that I was hung earlier I think I think even more important than all of this is as we're building systems and and I think it's more acute in my domain where we're working with conversationally is that have an end user impact we want to make sure that whenever possible we're building systems that when possible extend the capabilities of human users rather than trying to replace them like it would be really easy to just build a bot so there's two types of bot that you can build right one is like you're in a chat room with five people and you add a sixth person now which is a bot and he's an equal member to the six and he's just kind of working alongside you and maybe he does some things that are faster or with higher accuracy and and or you can introduce that bot as to be kind of a helper and some of it is just UX it's just how the bot presents itself how it introduces himself when it comes into the room but also in the types of dialogues and the types of conversations that you build into the bot it's important we found that in order to gain traction for your bot for it to be perceived as a helper and not as a replacement and we're not just seeing that with you know business users but really across the board so quick wrap-up we talked a little bit about we've talked a lot a lot of things doing about 45 minutes so we talked up we talked about diversity and and and how it's getting better and how we really need to make a better effort at it cisco has a great program or I think we started it but now there's a bunch of companies involved you can go and visit it it's called multiply diversity comm and it's all about bringing a diverse level of thinking into into the enterprise not just gender diversity but the cultural diversity as well it's a really cool site basically doesn't cost any money we're not they're not asking for donations or anything but pretty much what it is is you take a pledge to say that over the next year you're going to pick one person that you think is worthy of your mentorship or of your accolades and that you're going to help that person be noticed and to move one level up in their life and whatever that level means to you it doesn't have to be like a level at your company or a different pay band or whatever but you're just going to help them level up in life over the course of 12 months and I think if each one of us and and and a lot of a lot of us that have had the privilege of being successful can just like pick one cup one person that we feel that we feel it is worthy and is trying hard and and and and deserves to be recognized for that and just kind of help them do that well we'll have you know it's not just going to make the world a better place but we're going to see better products come come out of it because diversity is just good for software in general also just be mindful of that skills gap that we talked about earlier and and we're doing all of this not out of altruistic mean I really do believe that it's good business to do all this stuff right if we if we think of you know so going back to the chasm that we that we looked at earlier right where is it right if we go back to this um you know if you just as on tour as an entrepreneur or even as someone building products for your company spend some time thinking about like these are all white-collar jobs that are being kind of like displace by automation and computerization think about yeah it's going to take time it's going to take what 10 15 years I think is by the best estimate if you go if you go and you read this report right so this isn't this isn't the case now but it will be in 15 years so there's actually good business to be done in helping these displaced users which are currently using software systems in making a transition and so if you can do that well and if you can help the companies that these employees work for today help make that transition so that they can in turn introduce automation without having without being seen as like evil you know corporate overlord that's like automating their jobs if you can build systems that are empathetic and help people make that transition you'll make a shitload of money because big companies don't want to be perceived as being you know the guys that have to lay off 10,000 people if you can if you can build software that finds new roles for those 10,000 people within the organization then that's awesome and and and you can make a you can make a lot of money doing it as well so I think pretty much done i I have I have nothing to do with this book by the way because I get asked sometimes they're like what's in it for you I just I loved it so much they changed my worldview you know in a very profound way and so I highly recommend that you guys get it on Kindle or whatever it's it's really really good and a pretty quick read yes that you mentioned like a technology is going to accelerate then at some point we may miss what we were yeah a major medical development employed for whatever or anything else how can we Clement it if you missed it how can we what implement it if you make it miss it yeah so like I'm a doctor moving his leg do something yeah yeah one party product medicine right because it's I think there's going to be one of the first areas to be impact to be impacted by this I mean think think about it today right I mean there's a think about the the amount of time that the average surgeon has to spend constantly learning new techniques they read trade journals they go to conferences why it's not just the network with their fellow surgeons right there they're hopefully they're spending that time learning constantly learning new techniques right but it's just it's simply impossible today already without this exponential growth in AI and and and machine learning applied to diagnostics and medical techniques so it's a great question I don't know I'm not in the medical field but I imagine I imagine that the field of Diagnostics will eventually be almost fully automated so you know you'll walk into an evaluation center and it will be an AI that diagnostic diagnosis you and then outcomes that diagnostic to a human surgeon that says hey they've been diagnosed with this and here is the suggested treatment maybe we live in a world 20 years from now when when like 88 times out of 10 the doctors never even heard of that technique but now they can go train up on it they know that between now and the surgery which is scheduled for a week from now or whatever they just have to go learn that technique and that we have to create systems that help doctors quickly learn these techniques get quickly certified because they probably have never heard of heard of them before right right right and you have to be able to integrate some speakers absolutely yeah and again this is a tough problem but the only thing sure you're a joke going to disappear yeah 47% the rest 47% of white-collar jobs in the u.s. maybe you know the way to think about it is that the definition of work is going to change I'll I see I'm an optimist I love I love it yeah that sounds good to me but I don't know what that means yeah yeah I mean you know there's a lot of crazy ideas being thrown around now you know Universal income and a bunch of you know interesting ideas yeah interesting ideas being thrown around I mean you know what what do you do what do you do when it all goes back to value creation right so if you know these jobs aren't evaporating because there's a lack of production required like we're not in an austerity period right now but so so why are those jobs being lost and it's because they're being automated so these these machines are creating value where humans were before and and I guess the question that a lot of people are asking themselves and I mean I'm not even going to attempt to have a position on this but the question at least is if the amount of value being created stays the same or hopefully even goes up with less human effort then shouldn't we just all benefit from that like if the bots are doing all the work now then like even this one we just like be like whoo yeah that's all I saw go you know let's do a four-day workweek or a three-day workweek or whatever let's all go and enjoy our lives right and you know we get into that kicks us into all kinds of social issues that are just tough tough conversations to have right and in this political climate I think a lot of us are just focused on you know getting through the day then having these having these tough to sit having having these tough conversations but hopefully in about three and a half years we can actually start having a reasonable conversation about you know what we do with this cool yep an ethical question so at some point the Fox event is so good that it will pass until eBay yeah good question I mean why why does it matter yeah I don't know yeah yeah it could be a bad guy but yeah there's a lot of bad real guys okay but yeah that's a fascinating question I think yeah quick thought on that is I think identity just becomes super important at that point and the internet you know the internet was created what was originally is like a DARPA initiative so we had reasonable identity then but then somehow like we started really valuing anonymity on the internet and a lot of the big ideas that were there at the founding of the Internet kind of went away when when we started embracing this need for an annuity on the Internet and so we had an opportunity 15 20 years ago to put in place strong identity on the web we chose not to we explicitly chose not to because we wanted to be anonymous now someone's really anonymous on the web so maybe maybe it is time now to rethink that and and so you know if you're interacting with some other entity on the web you know you check its circ or whatever and it's just like oh that's a bot and and that's that right I don't know I know there's a lot of works that a lot of work that that's happening in Firefox on the some experimental branches that are looking at exactly that they're using a collect of a variety of verification techniques including biometrics and computer vision to certify not just that someone's physically at the computer but that it's actually you browsing and sorry just one one back there being constrained by the information various gamblers trying to collaborate but are you seeing any progress in terms of helping individuals or groups identified proactively made aware of activities information people as a piece sitting within their enterprise Mondex actually consulted there's often another group for what you're doing yeah yeah all right is anybody try to solve that problem yeah IBM at IBM is there any IBM people here no iBM is doing some fascinating stuff here I think I think they call it expert expert expert I'll find it and I'll send it out to the group or something I'll put it on the Eventbrite or whatever but but yeah iBM has a project exactly that it's built on top of Watson and basically it ingests a corpus of information from a variety of different sources and it will just help you find an expert within your organization on a particular topic without that necessarily listing it like on their internet profile or you know whatever and and so it uses you know you know who's who's who's creating powerpoints with you know on this topic like who's talking about that stuff on the company you know to chat boards or who is posting about who is authoring wiki pages about this stuff or writing emails about it obviously you get into some pretty sticky privacy issues when you when you talk about that and you know one of the challenges we have at Cisco is because we've made the decision to end ending ripped all the data flowing through our systems we as cisco actually don't have really a lot of insight into what you're talking about but we do understand the edges of like the connections that are being made who's talking to who but we just don't know about what and so one of the techniques that we're looking to apply without you know covering too much secret sauce is is a clever use of containerization so that we can push compute units in almost real-time to our customers edge private data centers right and so if we feel that there's an interesting event that needs to that need data that needs crunch while we don't have the encryptions used to do it we can push containers out to the edge of our customers networks where we do have the encryption keys we can perform that computation derive an insight and then push that inside out and then spin down the compute resources well I think we have time for yeah we've got a couple more [Music] yeah so was it going to happen yeah good good question I get asked that a lot actually I'm in variety of different ways but so certainly certainly chatbots have not lived up to the hype if you're judging the hype if you live in this 7 by 7 you know bubbles that we live in then it certainly hasn't lived up to the hype the hike was never that big to begin with outside of this bubble right and so and so outside of you know the VC sphere and the and it's the Technosphere here in the Bay Area like you go to you know you look at people that are using these chat tools in other parts of the world and you ask them you know are are you getting value from like your service now BOTS or your JIRA BOTS or whatever they're like yeah it's great it saves me you know 20 minutes a day you know so it's not it's not earth shattering and so if we can find ways and that's what we're trying to do by applying better techniques around conversational AI being able to take previous interactions and then predict users behavior and also what questions they're going to ask nest and then suggest what the next appropriate answer would be our goal it is very humble it's just if we can turn those 20 minutes that people are saving today to an hour and we can go to our customers and say hey we build collaboration products we're already saving your employees you know this many hours a week by not having them to get in their car and travel or whatever now they can have video calls anywhere in the world and in addition to that we're saving them an hour a day by employing this conversational AI that's just massive amounts of value right an hour a day that's over ten percent a productivity game right and so I don't know that they're necessarily will be a killer app maybe there will be maybe they're well but it's a gradual improvement in productivity and that's and that users like I mean generally they like talking to these things they find it a refreshing break from their day it doesn't talk back to them they don't have to negotiate it with them they just kind of do what they're told and they're there it's like having a third arm or another appendage right and you can turn it off it will you don't have to use them yes yep you mentioned all your tools slightly different scenario we can stop there like a side of that argument that on one hand make you people more connected and more disconnected together into while you know gentlemen before you were talking about oh editing here some person here whatever it is what is making quite now but there is also the other part of it is that maybe that's somebody think we're more walking together there would be more chemo suite totally yeah so I think that is like that you can expect in all that absolutely bad yeah yeah no I'm totally with you I am I personally you know and I can't say this too loudly although I'm being I have a camera in my face but but but no I think absolutely people are more productive when they're together in person you know if I were to do another startup I would be hard-pressed to to allow remote working policy at least for a until we've we've scaled to a certain size because I think it's just really important for people to create those those in person bonds but you know it's just not practical right I mean I you know we have a working on a VR app right now at Cisco virtual reality app and like you know the team I have to work with that had the cycles to work on my project there and chindan and it's just like you know I can go you know I've got a two-year-old you know I can't just pick up and spend you know few weeks in Shenzhen right I mean I could but yes but and yeah and so these tools make it are the next best thing to being in person but I'll tell you what like I think conversational AI isn't really limited to long distances that's what's that's what's so great about it I mean we just released a product earlier this year called the spark board it's basically like a 70 inch tablet that you stick on the wall and it's you know shared collaboration space you can go up and draw on it and whiteboard and stuff like that it's also does video and but you can like project your your laptop - it's a presentation device and it's a pretty cool thing and and so with all this you know with the mind-meld acquisition and all this investment that we're making in AI we also want to make it so that you know the device itself is like an in-person helper right and it's an entity and and so you know you can walk up to it you know someone comes up with a really great idea and you can just say hey spark you know take this action or hey remind remind me tomorrow that that you know Bob needs to get me this doubt or whatever if this is an earth-shattering stuff I can do that with my Google home today but doing but what the difference is to do it contextually as you're doing work done there is tremendous amount of value in that connection like if I literally as I'm drawing a diagram and you're like oh that's it like that's the thing that we've been missing it's in that right moment in that contest like expects extemporaneously except for anything that's contemporaneously thank you and just just get make that connection right away this tremendous amount of value there and and yeah so those little things that we're looking to make better cool I think we're out of time and it's beer time I don't know if you wanted to say a few words but thank you so much for your time and you [Applause] [Music]