so happy to be here thanks for having me um so if you've been attending the other sessions you've heard a lot about cloud um and yeah ai and analytics are kind of part of that trend so let's go ahead and talk about it so we got about i'll try to do it in 15 minutes and we'll leave five minutes for a question and answer um that that's a lot to cover for ai and analytics but i think i can uh give a pretty good um overview here so we'll start with just who i am so my name is casey johnson i'm with servicenow covering department of defense and intel communities currently focused on the army um but i have plenty of experience working with data and really my main focus is to make data actionable um so that you know governments can be smarter all right so why are we here and uh if you haven't heard of uh jack right it's the joint artificial intelligence center um it's really to how we're going to take ai in the dod over over the next you know decade um but it's really it comes down to decision making right and it comes down to data what access you have to that data in decision making and so this is a direct quote right to reduce to reduce risk and fielded forces and generate military advantage i put a picture of a radar on there because for some folks you know you never know who your audience is but i think of ai and analytics just like radar and even going all the way back to something like world war ii right you now have the ability to see something out there you don't know exactly what it is it's coming back right and you can use that information and you can do nothing with it you can just have a radar sitting in the in the room and somebody sees a ping and does nothing with it or you can have somebody manning that station and taking action seeing that i need to alert the captain you know that something's 200 yards out um we just don't know yet what it is and that's kind of where we are with ai and analytics right it's we're getting information back they're getting better every day the radar got better every year right and so until now you can actually identify exactly what kind of you know sub it is or whatever we're looking at if it's you know arc land um but it's really about hey we got this new tool data's coming back it's coming off the radar which is you know was an analogy for you know ai and analytics and how are we going to use that and how we're going to make that part of the mission and operationalize it and it's important if you don't and i put some key things down here the data needs to be relevant it needs to be timely you have to be able to trust the data and then even if you do trust it the way you build that trust is really to be able to validate that data and audit it in the in the future in case a decision went awry because of the data that you had and figure out why and then you can train on that and you can keep improving um what you're getting back so that's what we're looking at right we got this radar so but just to give context of where we are right the evolution of technology supporting the military because the data lives here right so way back when everything was on paper physically stored in filing cabinets difficult to correlate um from there as many of you know we went from two mainframes then we had on-premise networks and data centers and still to some extent have that um and now we're getting to the place depending you know if you're a civilian or or dod it's it's it's changing a little bit but we have the first clouds are becoming available and we of course we just had the presentation on how to procure those clouds um and now everything is really moving to cloud as a service and because of that um you know i saw the presentation from cloudtamer earlier right we have to tame these clouds um everything is you know all the data is becoming their data is available in real time um ai algorithms are becoming standardized so we don't need to train and have a custom solution for every single thing anymore right um and they're being and they're starting to be built into platforms such as servicenow um so there are some use cases where you know there i would say there's a spectrum of use cases for ai and one of them is um you know something as simple as you know doing some text analytics or predictive text analysis using ai and then it gets way more complicated for example i know i saw the corps of engineers for the army is now using drones to scan physical structures and you can then use ai to maybe predict you know when something could have a fail point in the future so there's a wide variety of problems that ai can solve and right now we're kind of in the beginning um if you read all the current blogs from you know the jack is you know we're collecting data and we're going to figure out the best way we can use that too to impact our mission but i realize when i talk to people you know ai is still this kind of foreign concept and i wanted to put a slide together that just gave a real clear simple example right so this is one where we have a solved problem kind of it's an image classifier right it's an already trained algorithm you literally send it an image from your computer or program or system and this is what comes back it comes back with hey it's an aircraft where it's a boeing and its score is 0.9982 so it's 99.8 percent sure that it's boeing but it's going to come back with not just for boeing it's going to come back with the other items that it's trained on right so other type of aircraft in this case it's pretty sure it's not the dornier right because that is less than a percent chance that that's it and that's it that's the blip on the radar screen right it came back now what do we do with it right and that's where i think uh for the dod side we're really looking to and then the and the dod has really focused on hey we have to rapidly prototype and pilot solutions we have to be ready to fail in order to succeed right because we're going to have to try to figure out the right use cases and that's this quote down here for what what's going to work and what is not going to work um you know somebody gave an example um of hey we could put a sensor in every single uh you know war fighters boots and we could sense how many steps they took or if they were having more pressure on one or the other and you could use ai to do all that and and you can it's just do do you want to is that the right use case um and that one's probably you know at this point it's not there and you know for example in your everyday life if you just take a look at um netflix i i you know netflix uses ai to to recommend to you what what you're going to watch because it saw what other people watched that you also watched i have no problem with that i think it's great that's a perfect use case and i've never yelled at netflix but i yell at my alexa at least once a week because it's a great use case it just also you know isn't quite there yet all the time and it drives me crazy you know um so that's where if we want to get into this rapid prototyping where we don't have to do huge procurements um a lot of the military already has servicenow and servicenow has machine learning built in it has um unsupervised and supervised um algorithms that you can use so you can think of it you like hey we already have some data here um data that that really can be used to take action right servicenow is the place where you go you don't just it's not just the radar blip blip blip right we're getting it's the place where you go to submit that ticket to make a change to put in a change request to you know to to do all of those things so with machine learning there's lots of small examples that you can do but it's a good way to i think pilot and probe what is possible and then you know use to and actually take action not just hey we're gonna you know we're going to uh you know go ahead and you know some something simple is a lot of you know servicenow just as the help desk really but it's quite a lot more than that right they have make hr modules they have it operations modules um there are there are agencies using servicenow for investigations right tracking you know working with agents tracking all the assets right think of it as just hey we have data we have a workflow we need to take an action we need to close it out um that's that's a lot of what the government does um and servicenow is really good at taking that data so i can i can provide one quick example where machine learning was super helpful um for an agency and it's it's a simple example um which is um responding back to congress so uh you know large agencies get lots of inquiries from congress you know they get phone calls they get walk ups they get emails and you better respond to congress consistently you can't give different numbers to different people um they don't like that so um servicenow they had the whole tracking uh congressional inquiry tracking system and servicenow and even though you know certain pieces of that information weren't together they used ai to go and find like tickets to make sure and go see hey last time we responded to senator grassley this is what we told them on you know the opioid crisis or something like that um and that's really helped and again so it's taking data helping them make the decision in real time and then also of course in real time means you can actually go walk on the hill with a laptop and show them the status of their inquiry and you know hey why is this request taking so long you know well we're pulling data from these three people and who's responsible and we should have that to you by the end of the week you know when you see um the leaders on congress see that their request is mobile and it's it's right there on the screen on the tablet they they just love it um so that's that's one of the examples i give down below but uh so the main point that i'm that i'm trying to get across is we got a lot of data we got ai and we got analytics providing predictions for the future or even recommending hey we think that's a boeing we think you know it's 99 accurate but we really have to operationalize it you have to get it as part of your day-to-day work how you can get smarter and better and more efficient and that's really what servicenow can help offer so the diagram you see on the right here you can see we have we have a host set of apps we have hr we have you know the help desk we can also very easily you know across go to different sources of information um you know maybe there's you know sensor data coming from uh ships you know that's reporting uh you know uh some something's running too hot something's running too cold um that can then um you know require maintenance so that's a good that's another good example because when that maintenance comes in it it's a pretty not all the time right remedy is also used quite extensively but you need you need to submit a ticket you need to take action on that and that's where i think right now we're at that place where we can there are really great use cases that we can go ahead and take um ai and analytics and take action on them and not just have them be part of something that was you know um fun to do and we can and we can fail at it and that's okay and that's really what we're trying to get to um so i have one more um federal use case i wanted to talk through here um which is the investigative identity revolution resolution which is you know something that comes through quite a lot where you have the same person you know a lot of times bad people are going to give you fake information they they lie and they're in different systems right they're in different siloed systems identity resolution is a tough problem and ai and analytics really help with that they say okay we're pulling all this together we work with your smes to identify what things are more difficult to lie about or how to correlate these identities together and then it provides them just like just like i showed with the um the boeing plane hey this is we think 98 of the time these two people are actually the same person but we're gonna let that agent that war fighter make that decision and then we're going to have an audit trail of that decision but now with those two pieces of data pulled together from the separate systems you can start to really like correlate other pieces of information together and which can lead to future decisions that that are important that we really need to know so um i guess that was my example of how we're breaking down siloed information systems you know usually it's it's years and years hey we're going to merge these together and they probably still should be merged together especially with all you know the cloud coming through and providing so much more functionality but you can still leverage ai in the meantime to really provide actionable information today is as quickly as possible um okay so i think uh that was my last slide so i was gonna ask for questions i think i'm right on time yeah thank you very much and um so a couple just kind of structural questions for you is um the solutions you're talking about the way servicenow approaches this is it um is it on-prem in most of these federal scenarios and then connecting to data sets that are in the cloud or is it in its own fedramp cloud environment and and connecting to the data wherever it lives good good multiple questions um but the question the answer is yes and yes so yes we've done um ai on prem yes uh it's the the service now gcc il4 cloud is um i think it's everybody's migrating there so it should be ready to use and that all the fedramp clouds already have the the ai built in and available okay and um and i'm curious you talked a little bit about a good silo busting example but is uh you know as a general rule are you seeing organizations trying to structure and ingest new data that you know sort of from the ground up thinking of your example of the army drones or the the you know potential boot sensors uh or is it about gathering and restructuring and and connecting all that legacy data that's out there yeah the that that is i guess the answer is it depends yeah um it depends how the data is stored that is true right ai needs the data to be um kind of in its raw state that's where it has the most power if it's working off aggregated data it it becomes more difficult um but yes like i think i mentioned before there are some problems where you know i use the drone example as one where hey that one might not best be suited for servicenow that might be hey we need a whole team we need a specialized team to be able to do this but it's going to take a long time so what i was trying to get to and what i'm trying to encourage the army to get to and i think everybody that i talk to is trying to get to is let's get some quick wins or let's get some quick failures but let's start trying because um we we have to be able to learn from our mistakes and figure out and and sometimes it's really even building those core competencies and you start identifying people within the army or within the air force hey these guys they did it they're really into it and you can build off of that um you know and of course always bringing in the right partners is critical you know no no one platform is going to solve at all i'm i'm only advocating that hey there are some problems where the tools are ready and we don't it's not going to take a huge you know contract that's been award over a year to do it it's it's there we just have to try or or have it as part of your mission statement to go try great and a couple questions that came in about the sort of investigations use case uh one was uh just a question of eight is is dod utilizing ai for for cyber related investigations that that you know of if you can speak to it um not that i know of uh not not by good experience you know either okay and um and then a more just sort of practical uh you know less specific um question which is can the retrieval of data from person to person apps like slack messenger that sort of thing be set up to be on demand for these investigations to avoid privacy issues be set up to be on demand yes i've never had that requirement but yeah they could be on demand for for privacy issues and i guess even if you wanted it further to make it increase privacy further um and i've actually i've gone into talks on servicenow privacy which is really fantastic um you could even have it just as as temporary so it never gets it's it's just you know stored in in memory and then it goes away and um and then you know you were talking about the the different sort of more traditional roles that servicenow has been involved in of you know of help desk and hr modules and things like that are you seeing are you seeing examples of the last few months whether in dod or elsewhere in government of using ai to sort of understand the the hr side of things in this new telework environment ooh um i haven't heard that one yet that's an interesting question um so no that that's interesting but i i haven't heard that yet but yeah you could use ai of course to to traverse logs and see kind of try to get a picture of digitally what people are how active people are at home and you know what what affects their day even i mean you can even start bringing in weather data for example yeah i know in other in other conversations we've had agencies have talked about trying to understand more from an i.t standpoint sometimes of just where the bottlenecks are in their systems as they go from five percent telework to 95 telework and yeah but just trying to trying to automate the analysis of that data to make smarter decisions and just seems like uh something that's going to be a a growing uh problem for agencies to have to tackle yeah and i then just to close out yeah servicenow is great at that right so if you had a form that previously had to be signed in person and now it's stuck because everybody's at home um you know dissolving those forms and then having them digitally signed is something servicenow excels at excellent well and i i think that's a good point good place to leave it which is good because we are at time but uh but casey i think uh thanks for um for digging in more on the on the ai as is something that people have touched on in passing all morning and it's nice to kind of really sink our teeth into it thank you thank you for having me everyone