Cognifest NYC 2017: Marc Teerlink, Capitalize on AI: Moving from Startup to Scale up

Cognifest NYC 2017: Marc Teerlink, Capitalize on AI: Moving from Startup to Scale up

Recording: Cognifest NYC 2017: Marc Teerlink, Capitalize on AI: Moving from Startup to Scale up

[Music] laughs they've done this on the fifth floor where we actually have a full instance of Watson the problem is if you want to meet Watson in person March 17 is the first option it's very popular and for the people in the back we have one two I just dropped my backpack there in the corner away three four five six seven chairs here you are welcome to stay at the buffet but we do have seats for you so let me tell you a little bit about this building so this building is a combination of IBM and some startups that are flown for the building so we have part of our cloud business here we have IBM Watson here this is clearly also the Watson headquarter and being of Dutch origin I am incredible proud that we actually are on original Dutch soil you know the moment the streets have a name here this is actually Dutch soil and according to the fact that we never actually signed the papers we technically could still claim it back I'm sure that number 45 would be very compliant for that um a little bit about my job I am the chief business strategist that means I work with a series of phenomenal teams that we have that are that I manage like a portfolio so each of them is like a little start-up themselves they play the general manager the marketing to go to Margaret they have a CTO and they create a new offering and as it continuously up or out of those offerings if they will actually make it to even an experimental or to a global availability service now we do that because another part of my teams actually set the business strategy so we work with IBM Watson CTO and we work with our research and saying what do we already have ready enough our research that we dare to touch it and it won't explode in our hands then the second part is that we say where do we see there is a market and we focus a lot as IBM on the enterprise market so where do we see a market that has the need for certain capabilities and that market could be developers that could be enterprising could be IT departments that could be software companies but that needed capability to make a difference different that with or without AI with or without Watson in our case and then we combine those together we start talking with a series of client if we do a lot of outside in thinking so this is really scary but we actually listen to our potential clients and then we still have an opinion now one of the people you if how many of you are gonna be the rest of the week going to the events like tomorrow and Wednesday in first day okay so rheya boy are you yeah then the back gentleman at blue shirt Ramey Jarrah is a chief technical officer it's a CTO for our offering compare and comply I'm gonna come back in it later but it's literally about semantic understanding and comparing two things without drea we still would be figuring out how to do it great thing is he's gonna do a little thing tomorrow rare could you want to give it a one-liner in your spiel let's rock and roll how within five years from now any decision any individual of us will make there needs to be informed will be a I powered will be AI infused will be AI doing the information forming part we might not realize it but we actually don't realize a lot of the technology we're using today either now this is a very pretty bold statement to make but when we see the uptake between Watson playing jeopardy in 2011 the last four years after or the four years after that we were going from incubation to actually making the platform as we have it here and the adoption of the platform by both enterprises and app developers many of them keeping under the rap we're today and you know this Simon Simon's gonna be our next speaker I'm looking forward to that part but you notice this from many platforms first take the AI in it and they won't say they have it because they first want to know for sure if it works good enough and actually makes a difference right the last thing you want to say I have something really cool but it won't work for the coming two years right it's a great marketing message so five years from now now the question is how are you today going to make actions to capitalize on that so let me give you a very simple thing people don't want moonshots with artificial intelligence people don't want the next disruptive kind of things yes there's an Ali Baba and there's an uber but they are one out of very very few that actually make it in that numbers what people actually want is something really really simple take one thing you really know well take one step out it had heavy lifting and put some AI on it people really want it's even if you look at things like uber or lyft hailing a taxi in New York the only thing it takes over that I actually have the freaking thing is gonna stop for me and if it's actually on its way to me and the second part I can jump out without paying for the bill that was all the differentiation was in it so I rather talk about intelligent augmentation I live in Washington DC so I'm pretty much comfortable with artificial intelligence but the point is but a point but a point is it's not gonna happen the coming years that artificial intelligence is fully gonna automate it take over heavy lifting so let me give you an example in the US alone if you are diagnosed with cancer god forbid 20% of the people diagnosed with cancer in the US are eligible for clinical trial less than 5% get matched because it takes a medical doctor and his or her team hundreds 1,680 hours to go to all the parameters genetic dispositions trials that have happened before the distance the fact that you have supportive brings you the physical distance if you can handle losing your hair and out on all those things and they get 15 minutes from the insurance company so you can't blame a medical professional that they don't have 260 and in 80 hours per conversation to do this but what we do with what's on here Memorial sloan-kettering all we do is we take that one step out that heavy lifting that going through all the details and say here it is now this is a transparency I'm gonna tell you what I have and how I got to it and use the medical doctor still have to do what you did with your residents how did you come to that conclusion but studies did you read if I take this one out which you still have done that so transparency we're gonna come back on that is one of the three core principles that I think is gonna bring an incredible value how did this heavy lifting advice that supported me actually come to this conclusion or a recommendation now important is how do does a I learn so let me give an example where go to the Internet today there's plenty of artificial intelligence coat tools platform solutions that do for instance visual recognition and they all have a library with some classifiers in that off the shelf the problem is now to find a platform that actually allows me to teach it something only for me now let's say I have a I want to open down in the building there's a little space open and Ray and I have fed up with all this AI stuff we are going to start our own smoothie shop but we want to make sure that the bananas that are kind of come in every morning that we actually check them at the supplier if they have the exact moment of ripeness that we need gonna need for our magic sauce so I'm gonna train this visual recognition say hey I know snotty 7% you recognize bananas do you recognize this yes it says it's a banana like okay 92% but so I'm gonna feed it some pictures of examples of how I want things to be this is how you recognize a ripe banana here's a series of pictures and within like minutes it says hey this is 94% of banana 63% ripe not good enough but way most human beings are 66 67 percent accurate in recognizing visual objects you know the jokes about the muffins and chihuahuas right and that's so I'm gonna now spend one a half-hour 99 0 minutes and I'm gonna highlight on all these banana pictures what makes it exactly perfect for me and now I have trained my classifier now this is big difference we're gonna come back and add in what since situation it's yours it's not the world's to jours you can choose to share it so I hear you think so I'm gonna make money with bananas so let's take a difference who view users Shazam maybe if that song that you don't know it okay so I don't know how it is for you but I have this on a regular basis like I see this magazine say where can I get this piece of clothing right and it's a very interesting thing but other people from net approach a yoke state they spent a hackathon for the weekend they had multiple teams and one of their teams in a weekend build an app that says let's take a series of pictures of celebrities and see what we have in our catalog and say do you have something like that do you have something in this style you don't have to have exact this thing or death thing but can you dress me like that that's really a serious catalog and search engine booster right so that that's the kind of thing that I'm trying to talk about it's not gonna be big if you really want to do something today I take the process that you already have you're selling clothing over the web they say can i power this with something more valuable so Watson is not a single entity in spite of the therapy sessions with Carrie Fisher and the discussion with Bob Dylan which were awesome to be it let's be clear Watson is a platform a series of artificial narrow intelligence more than 30 components now when you stand outside and you go to the Starbucks you're not gonna talk with the barista about problems with your heart unless you want them to call 9-1-1 now right you're not gonna talk basically with any person here in this office about the stock market unless you really think he or she is a stock market broker you never look definitely not gonna talk back see with your doctor about the quality of the coffee or the amount of cups that you drink you'd rather do it with your barista so the whole point is we are very very comfortable that human beings have a narrow specialty but we expect artificial intelligence to be the star trek level like computer how many people are in the room and then the computer answer is like there's 72 but only for intelligence you can figure out which one's right so Watson is a platform that has grouped its things in understanding reasoning learning or interactions based on different signals six-oh can be visual signal signals can be text sickness can be video segments can be photos so let's we just talked about the learning let's take it one step further my daughter studies here at NYU and a few weeks ago she landed on LaGuardia and she spent 72 minutes on the tarmac that's very normal I understand so she's text or she tweets out hey LaGuardia only 72 minutes tonight great on the tarmac and the boat from LaGuardia ray please thank you so the point is how do you get sarcasm how do you get the emotion out of it right so let me give you a real example who of you does not like Star Trek here because that's the moment to go to the bathroom okay good imagine [Music] I had the incredible pleasure this year a DC comic um to launch basically the story of the PlayStation 4 bridge crew together with Watson you guys are awesome but having for your audience beats everything must be very clear so in starting brisk you you play ID with individuals that can be beside you in the room there can be in other locations or you play with AI characters now just imagine for 10 seconds you're Kirk or Picard or Janeway or whoever you want to be and you literally just don't use the button reduce your voice like ensign warp 9 number 1 all faces forward and all these voice commands and all your sarcasm like there wasn't really a good hit was it and the game characters that are computerized understand you cool it's some available on Amazon that way more interesting for those of you that are developers we put the code on github so for example a web user connected with like an HTC HTC or a Sony Playstation speech-to-text and conversation conversation uses natural language understanding and therefore words into context when I say the Paris Hilton for the weekend on Google I do not mean date with the socialite but I mean I actually really want to take my wife away for a weekend right unfortunately I get 54 million hits for the socialite so language into context understanding speech to text and bringing it these are two services only that make this game suddenly so much more cooler and it is a cool game so let's go let's go to something like I like I already made a joke like I'm global and when people ask what do you live in I mostly say airplane throw out seat the promise I come and those of you to travel and Simon you must have the same you come to a hotel you come late in the evening the VIP row is the only road that is totally blocked up because everybody nowadays is iffy IP at the hotels and you finally go to the 21st floor to your room yet like the Wi-Fi password they did not give me the Wi-Fi password you pick up the phone of course nobody picks up at the front desk you go down with the elevators five minutes again in the VIP line you get the Wi-Fi password you go back to your room I think what time does breakfast start now there's a start-up in in LA go moment they started one half years ago with ivy Ivy is a conscious you just use texting on your phone a bunch of the high-end hotels in the US and the UK are using it let me just show this [Music] I hate miss eros [Music] [Applause] [Music] forty-eight thousand rooms are down IV the two top hotels of Caesars in Las Vegas actually won the Platinum customer service award Caesars never won any customer service award ever you would expect them to say we have a I we're gonna get rid of the from there there's people right but the point is if you if you stood in front desk in a hotel the last 20 years you dropped off more you got more TripAdvisor response more help people with the Wi-Fi you know that people can live longer without water then they can live with our Wi-Fi password today scientifically proven so the whole point is these people get much more heavy lifting that had nothing to do with the reason why they chose hospitality in the first place so now what they literally say is hey IV is actually bringing the humanity back to the human hotels that have been using IV Cornell just did a study had eight percent more employee interaction with customers free deformer percent uplift in positive reviews of these properties at Trip Advisor and 95 percent of 2017 s-class that graduated Cornell wants to work with a hotel with this technology it doesn't take jobs away it would be incredible stupid to look to the bottom line instead of the top line because it creates and phenomenal value because you want people to the location is the reason people come to your hotel the service is the reason they come back so let's go a little bit stretch it more out so we now sell conversation right the point is it's more than conversation this is taking millions and millions of messages and filtering down base of context so another startup is influential dot Co influential focuses on eliminating the waste and media by so if you are large Browns you make toothbrushes or razor blades or mascaras you understand that traditional advertising is gonna help me anymore I mean it coming two years the amount of ad blockers in iOS and Android are going to take the loss of the spontaneously advertising off how do I get influence how does millennial get influence youtubers bloggers texting social media so what they wanted to do is literally go to the social media and say how do I find for this week this month for this product the best influences for my products [Music] so these are two companies that didn't actually even exist two years ago the trace serious amount of money from investors and it's something really simple they're not taking over a whole ad agency they just say hey let's help at agencies and companies that work with that so now now we have gone to understanding let's go to reasoning the people who will benefit the most are coming three to five years from artificial intelligence or actually dos that are in a knowledge profession JP Morgan Chase Manhattan has 19,000 people that went for seven years to college law school passed the bar to become a lawyer and they do financial regulatory and after seven years the reward is that they can sit behind an app that basically says read the new documentation for the regulatory spots what's actually relevant put it in one column then read the Terms and Conditions from this product that you're responsible for put it in the second column and then free classify what this is about security privacy whatever what a phenomenal use of seven years of brain training right it's also by the way regulatory compliance is the white collar job with a high suicide rate in the States today because you can't do it right you can only do it wrong you can never be 100% compliant so what if what have you used a technology actually what if not just lawyers not just compliance but things about reading an EMR and actually then looking to the claim form about reading the description you have on a website for your package I mean ice and packages to Europe all the time and actually get the right custom code so it isn't four freaking four weeks in customs you know because I don't know about you but if you ever send something to Europe and it's a book don't say commercial value say it's a gift it's a commodity because otherwise it's gonna be for four weeks in customs and you miss that birthday and your mother's gonna be really upset in my case so one of the predictions we have we think that we in five years lawyers no lawyer will do their job without AI anymore so one of the things that is really going to be important so we talked about paraphrase right the beginning was paraphrase is how we want jeopardy give me one thousand versions to come to the same answer right and that's what conversation and paraphrase basically was then we go to the other part the Memorial sloan-kettering and the medical inference so I have like hundreds of documents at ASV be SCC zdz and slowly I make the chain of connections over these concepts but to semantically compare with each other to say are these two phrases in two contracts in two documents the same and so you realize how sexy this is an Oxford comma difference it's the difference between the disclaimer and an exclusion and a Ferdie million dollar lawsuit so it's actually quite interesting what if your company solely can be a hundred percent compliance what if your compliance people actually can say maybe we should bring 193 pages for the new checking app back to 50 and get the rest of the stuff out what if you actually could add value to simplicity so we don't think that artificial intelligence gonna replace humans we think it's gonna be augmented intelligence a lot of impact is going to be on knowledge workers and it's gonna be man and machine the one liner I want you to take away from this let the robots do the heavy lifting let the robots do the processing and let the humans do the thinking now I understand people say that's gonna be hard with technology and who few drives a car okay who've you talks to their GPS who have you cursor to the GPS who have you has given it yeah you so much you are lying I know that oh he was giving you a GPS a name who've you has followed the GPS blindly and walk driven down the steps of the parking garage because it said make a sharp right now this happens in where I used to live in London basically once a week in a parking garage that I used to frequent once a week the GPS says make a ride make a sharp right you've reached your destination they now have these anti-tank columns in front of these steps to make sure that passengers don't get by cars so let the human really go to thank you now for those of you that X here are in this and want to make some money on this lesson number one do you know where your training data is gonna come from do you really know with a brilliant idea how are you gonna train that do you have examples of the bananas you have people that know everything about bananas or whatever your example is lesson number two how are you gonna monetize that training what is gonna be your revenue model if you do not have a revenue model you do not have training data then you have a brilliant concept you're not gonna make money you're gonna help someone else get rich so people always focus out this is the cost saving I'm gonna make and I think again this is not the approach to take you have to look to the benefits of the benefits the value that you bring to the value that the value brings it is not that we reduce a medical professional hundred sixty hours of heavy lifting that they might have or might not have done it is actually that we give this opportunity to save more lives it is not that we get the regulatory people as suicidal and give them less spreadsheet to copy it's actually that we give companies retailers financial institutions telecommunications the opportunity to be faster to anticipate change and actually executed in their organization it's the training data revenue model revenue model of your clients benefits benefits you don't have them you don't have the case now the other part is selecting the technology so there's a lot of free technology out there right it's phenomenal so if you train a search engine you are training the search engine by every question that you give and every search that you do in every click that you do it if you train somebody else's AI in at AI is for free you're gonna pay it with all with your data you're either pay for something with dollars or your paper data that's how it is now that's not a problem but if you are training something really uniquely the knowledge of a process the knowledge of an approach that you want to keep yours you want to make sure that your data is your data so in IBM's case we've made a really important strategic choice we said you're using our technology in our capabilities you might even be using our technology and our capabilities plus a ground truth that we pre trained around it ground to the level of a college graduate who did financial regulatory or debt contracts or who did telecommunication or who knows how to work in a call center but that's not enough I'm really gonna insult some of you but I have some brilliant people working for me I only hire people that are brighter than I was at their age that's not difficult by the way but still but half of them is still useless in the first six to nine months when they come in here and I'm not talking useless like the complicate coffee I'm talking about letting them alone take something up and working it out because the secret sauce of how a company works how you do things in a company needs to be trained needs to be shared needs to be given and you need to have actually if your best people available to do that but the moment you make their training available on the highest skill for everybody that's the next part so if you're a law firm and you actually take your best people that train the technology instead of the new hires so the technology supports the new hires and actually the question that the technology wasn't trained for go to the experts the experts don't have time to come up with new concepts if you are telecommunication company and you can actually get your best field engineers available to do root cause analysis instead of continuously for us fighting that's what I mean with change now the point is again there's a lot of free a I and classify as there but there's very few platforms that say I separate your data and your knowledge from the rest and you choose if you want to sell that through my marketplace or you want to sell that to your clients but it's yours that's a really important message here now some of you say seriously as like seriously 2020 70% of the data is going to be from outside an organization that means that your data my data's someone else's data is going to be used by plenty of organization whether it is data or training knowledge neural network graphs whatever but data has just lost may according to economists surpassed the value of raw oil as a commodity pretty pretty powerful actually so if data is really something you can actually make money on just realize consumers realize one thing today if the product is free my data is the product there's these great cool websites where I can upload my CV and where my resume is there I want to know who looked at it but don't have to pay for it so actually I'll help you to make a product and then I have to pay you to see how you use my data it's a great model if you can get away with it but that model is about to change and you need to really focus on the value and keep this in mind so every time I have talks with people that our startups are in development that was lost week at the Canadian innovation exchange for normal event for those of you that live near Canada hey I'm definitely worth next year in Toronto to visit and plenty of people told me mark all I want is a sustainable competitive advantage preferably unfair I said sure that's not difficult that's no difficult you know what's going to be difficult you keeping your eyes on the prize because if you really want to be successful startup or you really want to be successful corporate incubator you need to do one thing you need to do that one thing better than everyone else and do not lose your focus and that sounds very very easy but the moment you make product one and you have basically customer number two and customer number two wants its customized your debt if you get it give in to that that's the moment you start finding a party who can do the customization that's the moment you start putting a customizable version later on the roadmap and keep the customer happy or you get another technology player to do that for you but the moment you step out of that focal point that gives you your value it's over and you belong to one of the ninety two percent of startups that doesn't make it but the eight percent you do do one thing and they really do it good and they don't dilute their approach so summarizing it's all about intelligent augmentation and I hear you think but that doesn't work for the arts and I yes it does so let's take four pieces of Watson tone analyzer not just the sarcasm but actually as my wife always yells to my kids I know that's what I said but you know what I meant right that's tone Elijah natural language understanding from Watson Watson beat bottom beat is basically like the last 30 years of billboards top hundred songs all their structure their their lyrics their peaks their waves their reception in the media their track records and we do some color designing for psychographic for making happy now we could have done it but it would not have been cool so we asked Alex the kid some of you might know him Nicki Minaj Rihanna we asked him to work with Watson beat the song is on Spotify now I do one slide after this to to wrap up and I'll just go directly when I take the person goodnight okay you having fun thank you [Music] is the kid and I've been tellin [Music] a culture plots analyzed millions of lines of text Wikipedia articles near times front pages social media blogs and more thank you the most pervasive things I'm covering the way people thought about them Watson also analyzed the layers of composition in over 26,000 top 100 billboard songs may be covering hidden patterns in most song structure from a devotion so much like having immediately self meeting we networks at once and then our course understanding social media which is conversation in general I could never do that [Music] to begin the music making process we gave them a prototype of something weird ability called Watson beat it understands and instructs music this is not fun I know we all make them [Music] if you want to build the billion-dollar company just take a process you really know well take one step out that has ever lifting put AI in it you make a shitload of money to come in five years so glad our communications are not here when they say that worked the best time to plant a tree was five years ago second best time is today thank you being here five minutes for questions arc so this very Inspiron technologies what is the hardest job so so let me take that in one step back so we have had three challenges before in building a platform I want to make sure actually that we meet regulatory so our data centers where both diseases of the world and we could actually serve our clients in multiple parts of the world of their data even in a cloud state in their physical Jia it's for our enterprise clients but also for those the big applications and solutions really important second is language and domain in spite of what you think it's really hard to just combine domain where we finished Watson with jeopardy we started Medical we used to Watson Japanese in fact all kind of medical journals in there because we thought that the generic training would help and I became to the conclusion at the bible and shakespeare are not the best medical advice given unless you live in certain part of this country so training the domain is hard and to make sure that the domain is supervised trained to get the capability you can by us in it right and to go through the main runway it's only going to work if we make our training tools open so our clients our partners can basically all say I'm gonna build the ground crew and I'm only gonna sell the ground truth to my land we're gonna put it down to Beckham awesome I think it's going to take five to seven years before we have telepathic spontaneously training to link that reads total was about might take longer so I think the biggest challenge is indeed domain training in our main ketchup you know in an acceleration the second thing has nothing to do with AI and everything what the data use is the power and capacity of batteries because as long as your phone dies at the end of the day and as long as even the best test models make more than 500 miles right we're going to have a substantial problem because in the end of the I might live in the cloud but it's delivered in your hand in your car or on your device on IOT devices and they mostly need portable power so the main learning power in power storage and then size and capacity of memory the two biggest issues come on I want something more sex it's a dumb thing this morning you're in my house already that first century [Music] constraints well sugars they read to them but it's one of my clients I come mention their name they are already in fire they have to spend the last year that's right so competitive interest and spend to go to all kind of catalogs of furniture and they mazing a patina for furniture so I'll see this beautiful design chair this one I like darn I want some table and stuff that matches that's right you know how hard it is that's even harder than playing that sure notice it because now we have to look to the style you look good calling it to look at sizes and lutely but it does that gives you the 70s right before and then I said okay this is only one item it's its first dish or an eBay or whatever I think those kind of retail access is going to be very powerful the circuit is much power of retail and please serve in your face who you like so Pittsburgh board Patagonia walk into a North Face snorts and I've got to hike it they know by the way I like to go hiking in the official record Patagonia around July by the way that's winter in the south of Chile the soon part of Chile we're in winter the Sun does not come up but you live in a great product and you're either extremely expertise in that or you were suicidal already and then you start talking about TBS by the way abs have you ever tried to find that air balloon system on the web because of 1980s you get all the carbons of habit but not the companies that sell basically the little backpack maybe five or six people in the North Face eventually would exactly know who what to recommend the name of spending our story they work with back office and they used to be cool but you know he would go back to the planet so the first thing that they did is they train all these questions in an iPad version that that we discover that we use for basically empowered to talk and the script version of death is now on the web and as in stores but I find the fact that Xen I never worked in the Lego store I work from Penn Station a year and I bought the Lego store and I know my youngest son wants something so I walk in and I said to go hey the lego city or probability yeah what city what there's a petrol Northup right New South Wales straight gonna Lego creator seer number 23 14 Wow you don't hold it with your clients so it's actually in the simple things I do think that a lot of is it's gonna hit us with a negative image for consumers or the inconvenience in processes it's absurd that do you know that 65% of the invoices for manual different-sounding companies research you know what because those clients who pay the exact amount with the right description they pay 50 percent now and 60 percent blades and they take the credit already the neighborly great so the renunciation is really hard unless you basically boss after year 96% reconciliation and all and it doesn't sound cool for you but if you're the person you'd be conciliation man you're like something gets better so I think the lawful gonna be little process that would make our life easier retail is really gonna hit us consumer products gonna really hit us we see the first step in policy conversations if you want to put your money on something if you aren't working on a book now organizational power right then you will be in two years one of many two delivers if you're working on an hour break make sure you deliver supplies but look for other AI capabilities that you can use that supers sir Oh so so yes so basically we have months of a caring services they appear global although wouldn't of them actually work in different regions we do nine languages now of course most services are whole service over the English I see if such services are in three in Japanese Spanish repeats French so I Chinese exactly you built your app as long as your app this is a cloud I really don't care much for patients I care for your vacation if you have regulatory restraints like we have clients that say we cannot take our de Tocqueville straight days out of America or my American clients are they in America right I want them in Canada so as long as you work with them the most of Western world we can handle we can read in Japan at the moment as well we have an airport joint venture so it's sort of good to Korea every joint venture so this is a real good point so princess Watson's out the platform so the way I see it I expected it coming to three years but it's too late we're gonna release that people will say I will mix and match 12% of amazon 8% of Facebook I won't give a comment about Microsoft I'm all for supervised learning I use some opening I hey I even wrote some code myself that's great the reason why we are a platform and all the people to build what what that's left when they can build about them or they can take components if you're a pure service company we're not going to build crucial thing which is market anymore or management or something it would be better to work with the partisans influential if you actually are technology company why not because every forget about Watson today if you work with technology from something or you would work walk into it Accenture or you the Infiniti would like to serve the report you would choose your niche to you know exactly even those priors you can open a specialization or your local part you were different change yourself because if you change yourself in the role market search of the arts but so IBM services has a niche but that doesn't mean there's no business for you oh by the way nobody is gonna lose market share so short answer yes you can be sure and I'm gonna say something that's already weird you should only do it at Walsall you should say where do I need one so to give the best AI power that meets in a platform behind that ensures the insecurities they and where do I use some moments that I can interchange the play with there was so so in the bluemix environment and there are people better than I am to articulating IBM direct apart from bluemix is more than Boston there's lots of time it's analytics in it it has a stake the platform services in it which basically is not just remember crunching data cleansing data standardization IOT connectors etc so there's a whole bunch there what we do is Watson before was really quarrel the AI and acknowledge the ground truth planted part space from there but the roommates platform make sure that it is much more structured database like the weather data and those kind of things so just go to a global common I think that's all the time that for questions for mark thank you [Music]