WEBVTT 00:01.809 --> 00:03.809 Hi , I'm Bonnie Evangelia . I'm the 00:03.809 --> 00:05.659 acting acquisition and assurance 00:05.659 --> 00:07.603 director for the Chief Digital and 00:07.603 --> 00:09.659 artificial intelligence office . I'm 00:09.659 --> 00:11.715 Trevor Austin , the chief technology 00:11.715 --> 00:13.881 officer at the CDAO . We're here today 00:13.881 --> 00:15.659 to talk about Open Dagger , our 00:15.659 --> 00:17.770 framework for doing rapid acquisition 00:17.770 --> 00:19.326 of new technologies and new 00:19.326 --> 00:22.069 capabilities in CDAO , and we hope more 00:22.069 --> 00:24.236 broadly at the Department of Defense . 00:24.236 --> 00:27.149 Open Dagger is a framework to enable us 00:27.149 --> 00:29.270 to acquire new technologies and new 00:29.270 --> 00:31.840 capabilities more quickly . The main 00:31.840 --> 00:33.951 way that we want to do that is making 00:33.951 --> 00:36.118 sure that new capabilities are able to 00:36.118 --> 00:38.229 snap right in that we have predefined 00:38.229 --> 00:40.118 interfaces where we can adopt new 00:40.118 --> 00:42.229 things without needing to do a lot of 00:42.229 --> 00:44.007 translation , a lot of building 00:44.007 --> 00:46.220 adapters , a lot of re-engineering of 00:46.220 --> 00:48.775 existing systems in order to . Get new 00:48.775 --> 00:50.886 things in the door , prove them out , 00:50.886 --> 00:53.053 and then scale them up . Importantly , 00:53.053 --> 00:55.219 Open Dagger is a framework . It is not 00:55.219 --> 00:57.164 a specific program . There's not a 00:57.164 --> 00:59.455 single RFP this refers to . It is an 00:59.455 --> 01:01.622 approach that we're developing at CDAO 01:01.622 --> 01:03.844 and that we hope will be either adapted 01:03.844 --> 01:06.066 or adopted across the department . that 01:06.066 --> 01:07.677 we're applying to all of the 01:07.677 --> 01:09.899 acquisition work that we're doing going 01:09.899 --> 01:12.066 forward and we're developing it in the 01:12.066 --> 01:14.288 open . We're sharing the materials that 01:14.288 --> 01:16.455 we have and the learnings that we make 01:16.455 --> 01:18.622 to try and have an integrated approach 01:18.622 --> 01:20.677 that combines the acquisition pieces 01:20.677 --> 01:22.788 and the technology pieces you need to 01:22.788 --> 01:24.955 get . New technology adopted quickly , 01:24.955 --> 01:27.066 right , and sometimes that's the hard 01:27.066 --> 01:28.805 part is to make sure that our 01:28.805 --> 01:31.027 technology approach is aligned or has a 01:31.027 --> 01:33.165 nice marriage with the acquisition 01:33.165 --> 01:35.955 approach . So this is an intentional 01:36.355 --> 01:38.299 attempt to make sure that both are 01:39.230 --> 01:41.430 Nested nicely with each other . That's 01:41.430 --> 01:43.652 right . In the end , none of the things 01:43.652 --> 01:45.652 we're doing here are entirely new . 01:45.652 --> 01:47.763 Everyone wants to go fast . There are 01:47.763 --> 01:49.986 best practices . This is our attempt at 01:49.986 --> 01:51.986 combining them and synthesizing the 01:51.986 --> 01:54.300 lessons and then sharing them out and 01:54.389 --> 01:57.860 having a name to call it by . So one of 01:57.860 --> 01:59.749 the main things from a technology 01:59.749 --> 02:01.749 perspective we need to do to enable 02:01.749 --> 02:03.980 this is look for natural break points 02:03.980 --> 02:06.500 in applications or in solutions where 02:06.500 --> 02:08.619 things could be handed off from one 02:08.619 --> 02:10.786 provider to another and make sure that 02:10.786 --> 02:13.580 there's a clean , well defined , 02:13.970 --> 02:16.259 importantly documented and government 02:16.259 --> 02:18.259 controlled interface between them . 02:18.479 --> 02:20.940 Commonly this takes the form of an API . 02:21.610 --> 02:23.721 But importantly , you can't just have 02:23.721 --> 02:25.999 any API . You can't have , for example , 02:26.669 --> 02:28.780 you might have a custom plug for your 02:28.780 --> 02:30.725 laptop that only fits your kind of 02:30.725 --> 02:32.891 computer and nothing else . We want to 02:32.891 --> 02:35.002 make sure that we have standard plugs 02:35.002 --> 02:37.225 in standard places , so not the Apple , 02:37.225 --> 02:39.336 not the Apple dongles . We don't , we 02:39.336 --> 02:41.509 don't want DOD dongles everywhere . We 02:41.509 --> 02:44.470 want to have . We want to have , you 02:44.470 --> 02:46.869 know , Lego bricks with the standard 02:46.869 --> 02:49.091 interface that all snap together . This 02:49.091 --> 02:50.702 is important because from an 02:50.702 --> 02:52.813 acquisition perspective we tend to be 02:52.813 --> 02:54.925 very hyper focused on the application 02:54.925 --> 02:56.813 layer . We're very focused on the 02:56.813 --> 02:59.036 capability and delivery of a capability 02:59.036 --> 03:00.702 and often what that means the 03:00.702 --> 03:02.925 unintended maybe consequence of that is 03:02.925 --> 03:04.647 we're buying vertically closed 03:04.647 --> 03:06.869 integration technical stacks behind the 03:06.869 --> 03:09.119 application layer a lot . And without 03:09.119 --> 03:11.063 any malice on anyone's part , it's 03:11.063 --> 03:12.897 often just easier and quicker to 03:12.897 --> 03:15.008 develop things in a silo . So what we 03:15.008 --> 03:17.175 want to do is provide the structure in 03:17.175 --> 03:20.279 our own contracting and also open 03:20.279 --> 03:22.960 guides to help . Other departments that 03:22.960 --> 03:25.750 are acquiring things know where to ask 03:25.750 --> 03:28.529 for interfaces to be created so that 03:28.529 --> 03:30.696 you can get the application capability 03:30.696 --> 03:32.807 you need in a way that doesn't commit 03:32.807 --> 03:34.862 you to deeper things under the water 03:34.862 --> 03:36.918 line that they are going to slow you 03:36.918 --> 03:38.973 down going forward . What we want to 03:38.973 --> 03:41.196 produce in the end is an open ecosystem 03:41.196 --> 03:43.251 where at these various layers of the 03:43.251 --> 03:45.251 stack with these different kinds of 03:45.251 --> 03:47.418 capability across the interfaces , you 03:47.418 --> 03:49.585 can have lots of different providers . 03:49.585 --> 03:51.529 Either competing or just providing 03:51.529 --> 03:51.369 parallel solutions that address 03:51.369 --> 03:53.480 different things so you don't have to 03:53.480 --> 03:55.536 get um if you're buying something in 03:55.536 --> 03:57.647 one layer doesn't commit you above or 03:57.647 --> 03:59.480 below you can keep adopting more 03:59.480 --> 04:01.536 solutions at different layers in the 04:01.536 --> 04:03.758 stack , right ? So the key here from an 04:03.758 --> 04:05.869 acquisition approach is again because 04:05.869 --> 04:07.570 we're not buying a front end 04:07.570 --> 04:09.626 application and everything behind it 04:09.626 --> 04:12.360 we're allowing solution providers if 04:12.360 --> 04:14.839 they . Don't have some other part of 04:14.839 --> 04:17.006 the technical stack , then that's OK . 04:17.019 --> 04:19.186 And we're trying to , we would , would 04:19.186 --> 04:21.186 call this modular . We're trying to 04:21.186 --> 04:22.963 make this create a more modular 04:22.963 --> 04:24.630 approach to our acquisition 04:24.880 --> 04:26.970 methodologies and make sure that 04:27.640 --> 04:30.179 Vendors don't have to commit to giving 04:30.179 --> 04:32.440 us the entire tech stack no matter what 04:32.440 --> 04:34.607 their solution is , right ? We want to 04:34.607 --> 04:36.551 be able to say if you have a great 04:36.551 --> 04:38.662 autonomy solution , we don't need you 04:38.662 --> 04:40.884 to develop a whole infrastructure layer 04:40.884 --> 04:40.619 below it . You can use ones that are 04:40.619 --> 04:42.820 already provided and the way that you 04:42.820 --> 04:44.876 develop your great autonomy solution 04:44.876 --> 04:46.709 isn't going to be constrained by 04:46.709 --> 04:48.876 infrastructure choices that we've made 04:48.876 --> 04:50.931 earlier . That doesn't mean that you 04:50.931 --> 04:50.799 know a full stack provider can't 04:50.799 --> 04:52.910 provide solutions . Lots of layers of 04:52.910 --> 04:54.910 the stack . We just want them to be 04:54.910 --> 04:57.021 able to be remixable so that when new 04:57.021 --> 04:59.132 capabilities come on , no one's hands 04:59.132 --> 05:01.132 are tied by the decisions that were 05:01.132 --> 05:03.299 made in the past . So that's a benefit 05:03.299 --> 05:05.355 to small businesses because they can 05:05.355 --> 05:07.355 now integrate into our platforms or 05:07.355 --> 05:09.529 environments and but it's also great 05:09.529 --> 05:11.862 for government buyers because they have , 05:11.929 --> 05:14.250 I would say , more choice and they can 05:14.250 --> 05:16.250 make intentional decisions around . 05:17.029 --> 05:19.196 Some of those deeper parts of the tech 05:19.196 --> 05:21.251 stack that might be impacting cost , 05:21.251 --> 05:23.473 schedule or performance , I would say . 05:23.769 --> 05:26.410 So the devil is in the details on these 05:26.410 --> 05:28.929 things in order to make that work , we 05:28.929 --> 05:31.130 do have to provide a certain small set 05:31.130 --> 05:33.570 of universal services that help define 05:33.570 --> 05:35.681 where those interfaces are . So we've 05:35.681 --> 05:37.403 got some of them listed here . 05:37.403 --> 05:39.600 Importantly , unlike other efforts . 05:39.994 --> 05:41.938 within the CDAO to implement these 05:41.938 --> 05:44.161 things because we aspire for them to be 05:44.161 --> 05:46.114 universal and enable this 05:46.114 --> 05:48.058 interoperability , we want to make 05:48.058 --> 05:50.003 versions of , for example , a data 05:50.003 --> 05:52.003 catalog that are as lightweight and 05:52.003 --> 05:54.170 unopinionated as possible . We want to 05:54.170 --> 05:56.281 accommodate data owners wherever they 05:56.281 --> 05:58.503 happen to be so that they're as easy as 05:58.503 --> 06:00.725 possible to integrate with . That means 06:00.725 --> 06:02.670 sometimes holding back on having a 06:02.670 --> 06:04.781 richer feature set and saying we want 06:04.781 --> 06:04.774 that to be a thing that you can plug in 06:05.114 --> 06:08.299 and the universal layer should be . As 06:08.299 --> 06:10.589 lightweight as we can make it , so 06:10.589 --> 06:13.170 because Open Dagger is focused on going 06:13.170 --> 06:15.200 quickly , uh , there are technology 06:15.200 --> 06:17.089 choices we want to make , but the 06:17.089 --> 06:19.144 largest portion of that is about how 06:19.144 --> 06:21.200 you do the acquiring and so Bonnie's 06:21.200 --> 06:23.256 gonna speak to how we get it going , 06:23.256 --> 06:25.422 right , so they're . Kind of taking it 06:25.422 --> 06:27.559 back a step , thank you for giving us 06:27.559 --> 06:30.040 some technical details there , but we 06:30.040 --> 06:32.040 now when we're thinking about maybe 06:32.040 --> 06:33.929 business decisions or acquisition 06:33.929 --> 06:36.040 decisions because maybe we don't have 06:36.040 --> 06:38.040 all the answers yet and uh like you 06:38.040 --> 06:40.151 said devil's in the details and we're 06:40.151 --> 06:43.750 not sure how what standards or maybe 06:43.850 --> 06:45.850 specific technical nuances we might 06:45.850 --> 06:48.017 need on our acquisitions yet we have a 06:48.017 --> 06:50.239 set of principles we are practicing and 06:50.239 --> 06:52.559 living by . There are certain outcomes 06:52.559 --> 06:54.679 that we believe these principles will 06:54.679 --> 06:57.040 uphold essentially . The first is we 06:57.040 --> 06:59.207 have to be clear about how to bring in 06:59.207 --> 07:01.207 emerging technology , clear pathway 07:01.207 --> 07:03.040 acquisition pathways to initiate 07:03.040 --> 07:06.010 piloting at speed . The trade win 07:06.010 --> 07:07.843 solutions marketplace is a great 07:07.843 --> 07:09.510 example of this where we have 07:09.510 --> 07:11.799 communicated to industry that this is a 07:11.799 --> 07:14.230 path . by which we will leverage and 07:14.230 --> 07:17.029 bring in emerging technology that is 07:17.029 --> 07:19.029 meeting a mission need or some some 07:19.029 --> 07:21.196 kind of operational gap . So that's an 07:21.196 --> 07:23.307 example of where we need to make sure 07:23.307 --> 07:25.251 we have those clear and accessible 07:25.251 --> 07:27.630 acquisition pathways . Second , for we 07:27.630 --> 07:29.686 need to make sure we have government 07:29.686 --> 07:31.686 owned technology infrastructure and 07:31.686 --> 07:33.852 services . That's the backbone we were 07:33.852 --> 07:35.852 just talking about , right ? And we 07:35.852 --> 07:35.470 have to do this in a way that is 07:35.470 --> 07:37.700 protecting vendor IP . So if we ask . 07:38.989 --> 07:41.510 An autonomy vendor to come into our 07:41.510 --> 07:43.390 platform or government or owned 07:43.390 --> 07:45.390 infrastructure we have to make sure 07:45.390 --> 07:47.829 that their IP is protected , but we 07:47.829 --> 07:50.051 also need to make sure the integrity of 07:50.051 --> 07:52.385 the government data is protected . Next , 07:52.385 --> 07:55.029 uh , continuing the getting things 07:55.029 --> 07:57.429 through quickly to initiate piloting is 07:57.429 --> 08:01.000 not , uh . That's not the end of the 08:01.369 --> 08:03.591 the road we still have a lot of work to 08:03.591 --> 08:05.813 do from an acquisition perspective , so 08:05.813 --> 08:07.591 making sure we have clear cyber 08:07.700 --> 08:09.660 security requirements , ATO 08:09.660 --> 08:11.940 requirements , and making sure that . 08:12.570 --> 08:15.010 We're able to implement a rapid 08:15.010 --> 08:17.609 assessment environment is key to that 08:17.609 --> 08:20.170 as well . So in order to make these 08:20.170 --> 08:22.170 pilots successful , we also want to 08:22.170 --> 08:24.570 provide a common set of tools and 08:24.570 --> 08:26.459 capabilities that new vendors can 08:26.459 --> 08:28.514 leverage . So you're not reinventing 08:28.514 --> 08:30.126 the whole wheel , you're not 08:30.126 --> 08:32.237 reimplementing the whole stack if you 08:32.237 --> 08:31.850 have a best in breed solution . 08:31.980 --> 08:34.091 Particular kind of problem we want to 08:34.091 --> 08:36.313 enable the vendor community to focus on 08:36.369 --> 08:38.536 just that thing and not have to tackle 08:38.536 --> 08:40.702 a bunch of ancillary things , and that 08:40.702 --> 08:43.250 means providing tools to everyone that 08:43.250 --> 08:45.194 comes in and probably standards or 08:45.194 --> 08:47.539 tools and standards and business rules 08:48.289 --> 08:50.400 these are these are all things we are 08:50.400 --> 08:52.345 anticipating I would say from that 08:52.345 --> 08:55.010 perspective . And then lastly we need a 08:55.010 --> 08:57.066 clear operating model . So let's get 08:57.066 --> 08:59.859 into that on the next slide . This by 08:59.859 --> 09:03.669 no means is it , but this is again our 09:04.109 --> 09:06.276 line in the sand . This is the start , 09:06.276 --> 09:08.442 I would say , of an operating model by 09:08.442 --> 09:10.900 which we're going to be living uh the 09:10.900 --> 09:13.020 open dagger principles by , uh , 09:13.179 --> 09:15.260 there's nothing new , super new like 09:15.260 --> 09:17.316 you were saying earlier here from an 09:17.316 --> 09:19.260 acquisition perspective , you have 09:19.460 --> 09:21.404 major acquisition muscle movements 09:21.404 --> 09:23.016 starting with requirements , 09:23.016 --> 09:24.793 development all the way through 09:24.793 --> 09:26.904 transitioning . The key I would focus 09:26.904 --> 09:29.210 on this slide are the gray boxes 09:29.549 --> 09:31.909 underneath the yellow boxes underneath 09:31.909 --> 09:34.076 each of those major muscle movements . 09:34.076 --> 09:36.187 This is where we are trying something 09:36.187 --> 09:38.353 different or we are taking an approach 09:38.353 --> 09:41.190 that we believe is nuanced or specific 09:41.190 --> 09:43.260 to emerging technology or scaling 09:43.590 --> 09:45.590 emerging technology within the open 09:45.590 --> 09:47.701 dagger construct . So underneath step 09:47.701 --> 09:50.000 one . How we are doing requirements 09:50.000 --> 09:53.539 development , uh , we tend to focus on 09:53.539 --> 09:55.483 problems and which is a little bit 09:55.483 --> 09:58.140 different and this is where we are 09:58.140 --> 10:01.119 doing work . To figure out problem sets 10:01.119 --> 10:02.897 all the way from the edge , the 10:02.897 --> 10:04.675 tactical edge through strategic 10:04.675 --> 10:07.469 enterprise problem solving . And we're 10:07.599 --> 10:09.766 enabling those in novel ways , right ? 10:09.766 --> 10:12.030 You have the Trade Winds marketplace 10:12.030 --> 10:14.719 where problem sets are posted for open 10:14.719 --> 10:16.886 comment . You've got where vendors can 10:16.886 --> 10:18.719 come and post their solutions to 10:18.719 --> 10:20.997 problems that they think we might have . 10:20.997 --> 10:23.330 We've got the guide series of exercises . 10:23.330 --> 10:25.497 We've got the sipper vehicle . There's 10:25.497 --> 10:27.608 a wide way of a wide range of ways of 10:27.608 --> 10:29.775 identifying . These capabilities right 10:29.775 --> 10:31.886 and that's right , and that's a great 10:31.886 --> 10:34.190 segue into because like identifying the 10:34.190 --> 10:36.301 problem is number one , then you have 10:36.301 --> 10:38.523 to find solutions very , very quickly . 10:38.523 --> 10:40.579 This is where step 2 comes into play 10:40.579 --> 10:42.523 and we all those clear acquisition 10:42.523 --> 10:44.690 pathways or rapid acquisition pathways 10:44.690 --> 10:48.570 need to . Be a parent and it needs to 10:48.570 --> 10:51.130 be , uh , it's , it's not just a parent 10:51.130 --> 10:53.352 to the government buyer but to industry 10:53.352 --> 10:55.463 as well , especially small businesses 10:55.463 --> 10:57.130 where if you really want that 10:57.409 --> 10:59.849 innovation or that competitive 10:59.849 --> 11:01.849 ecosystem at that application layer 11:01.849 --> 11:04.729 that needs to be extremely apparent for 11:04.729 --> 11:07.359 under step two . And then with step 3 , 11:07.530 --> 11:09.752 this is where I think a lot of the work 11:09.752 --> 11:11.752 that your team is doing will really 11:11.752 --> 11:13.919 shine . We , how do we know what right 11:13.919 --> 11:16.086 looks like ? And part of that is going 11:16.086 --> 11:18.086 to be that when we do the pilots we 11:18.086 --> 11:20.479 want to get them as much as possible 11:20.479 --> 11:23.080 out of toy environments , out of 11:23.080 --> 11:25.520 prototypes and into we have a working 11:25.520 --> 11:27.619 version that is in production with a 11:27.619 --> 11:29.675 test set of users and if we have the 11:29.675 --> 11:31.786 right technical infrastructure we can 11:31.786 --> 11:33.952 lower the barrier to entry for that um 11:33.952 --> 11:36.119 both from a technology perspective and 11:36.119 --> 11:37.952 from an acquisition and like ITT 11:37.952 --> 11:40.063 perspective and so you can have those 11:40.063 --> 11:42.119 pilots be piloting with real users . 11:42.119 --> 11:44.286 Real data so that when you move on and 11:44.286 --> 11:46.286 doing them in a way that we know in 11:46.286 --> 11:48.230 advance is going to meet our cyber 11:48.230 --> 11:50.397 standards is going to be interoperable 11:50.397 --> 11:52.119 with other systems through the 11:52.119 --> 11:54.063 completion of our pilot when we're 11:54.063 --> 11:53.984 doing our evaluation , we're ready to 11:53.984 --> 11:56.206 make big decisions instead of trying to 11:56.206 --> 11:58.373 cross the chasm , right ? So step 4 is 11:58.373 --> 12:00.625 really being postured to scale quickly 12:00.625 --> 12:03.465 when we find something successful and 12:03.465 --> 12:05.849 something . Can just be that like a 12:05.849 --> 12:08.179 small piece of success . It doesn't 12:08.179 --> 12:09.957 have to be sometimes we go into 12:09.957 --> 12:12.650 piloting thinking I need I'll just , I 12:12.650 --> 12:15.570 need some , I have a big idea of what a 12:15.570 --> 12:17.570 capability is , but maybe there's a 12:17.570 --> 12:19.681 small kernel of success from that big 12:19.681 --> 12:21.903 idea that is worthwhile to go into that 12:21.903 --> 12:24.289 step for like big decision making . So 12:24.289 --> 12:27.929 there's ways in which we're going to 12:27.929 --> 12:29.929 posture contracts . There's ways in 12:29.929 --> 12:32.151 which we're going to have relationships 12:32.151 --> 12:34.262 with the services so that we can move 12:34.262 --> 12:37.369 into that . Transition or production , 12:37.500 --> 12:39.780 you know , pick your technical term 12:39.780 --> 12:42.880 environment . So this is how we're we 12:42.880 --> 12:45.320 in CDAO have been trying to make Open 12:45.320 --> 12:47.542 Dagger real . We have multiple lines of 12:47.542 --> 12:49.840 effort where we either started in a 12:49.840 --> 12:53.840 prototype or we started uh there were 12:53.840 --> 12:56.599 either existing use cases that were 12:56.599 --> 12:59.400 I'll say at the operational or tactical 12:59.400 --> 13:01.729 level that found success and we built 13:01.729 --> 13:04.039 upon that and either went into pilot 13:04.039 --> 13:06.206 quickly or we went into helped them go 13:06.206 --> 13:08.428 into production very quickly uh I think 13:08.428 --> 13:10.809 the key here is we're . Eating our own 13:10.809 --> 13:13.330 dog food as you and I were just 13:13.330 --> 13:16.650 discussing and we're trying to either 13:16.650 --> 13:19.010 get to that step 4 . We're either 13:19.010 --> 13:20.954 developing lessons learned , we're 13:20.954 --> 13:23.010 developing best practices , or we're 13:23.010 --> 13:25.330 moving things into a more production or 13:25.330 --> 13:28.030 enterprise ready state so that the 13:28.030 --> 13:29.880 department can take advantage of 13:30.250 --> 13:32.194 certain capabilities . And each of 13:32.194 --> 13:34.609 these first three you mentioned the 13:34.609 --> 13:37.409 Maven smart systems . The Advana and 13:37.409 --> 13:39.530 the tactical data services , they all 13:39.530 --> 13:41.752 have provisions for third party vendors 13:41.752 --> 13:43.586 to come in . They're providing a 13:43.586 --> 13:45.929 framework for them to pilot their own 13:45.929 --> 13:48.520 new solutions , and then we have both 13:48.710 --> 13:51.080 the acquisition vehicle and the 13:51.080 --> 13:53.650 technical platform to scale up those 13:53.650 --> 13:55.080 pilots that proved successful .