Host Intro (Gene Kim)

So next speaker is Justin Fanelli. He is the acting CTO of the US Navy and the technical director of PEO Digital. So I am so delighted that he's going to be sharing some stories about how he's bringing incredible innovation to, well, some of the more boring aspects of the US Navy — sort of like the back-office functions. And yet he will argue that these enterprise support functions are some of the most critical things that enable hundreds of thousands of sailors and civilians to do their daily work. So try, imagine getting work done when your laptops don't show up for, say, months, or take double digits — minutes — to boot, when as he says, people are just trying to save the free world.

I am in awe of how he's helping bring innovation to the edge. Justin will share how the US Department of Navy is laying the foundation to push innovation, supported by a courageous and incredibly irreverent centralized group that provides mandates, air covers, and guidance. What they have achieved, I think, will be genuinely inspirational to any large enterprise who think that they have world-class bureaucracies built up over decades or maybe even centuries. Here's Justin.

Justin Fanelli

Good morning, family. This is my first year, but I've been taken aback by how welcoming and how connected everyone is. It's one of my favorite things in the world, to be a part of something bigger. It's how I felt when I joined the military 25 years ago.

It feels big every day in the Department of Navy, Department of Defense, and federal government. One thing we know about BIG is, it's hard to change. And it feels every bit of that, at times.

So: 670,000 users, top to bottom. We're talking about anything from connectivity all the way up to endpoints and everything in between.

Something has changed. We've been in one place on a trajectory for a long time. It's something that industry realized a long time ago, but what we were doing, we finally realized, stopped scaling. So we were at an impasse. We were trying to climb a tree to get to the moon. It worked when we were trying to get to a roof, and then no longer. We were in need of a revolution.

That revolution was a public-private-sector mesh revolution. We knew it. We knew what we needed to do. We knew where we needed to adopt and leverage differently. But could we do it well?

You see, in a software-defined world, we almost have no choice. Comms matter more than before, and our integration timelines were going the wrong direction. We knew layer-two tools and instrumentation needed to be changed as soon as possible. And there was one more wrinkle, or confounding factor.

The Davidson Window — a term coined by Admiral Davidson for when China would be prepared to enter Taiwan, a sovereign nation and linchpin for modern technology and everyone's economy — was 2027. That is not traditional government speed.

We needed to move with more partners, to more commercial off-the-shelf, out of our comfort zones every day. And even over the vaunted Valley of Death, time and time again, we needed to bring much more innovation forward.

We needed a bridge. That bridge would come in the form that is, in fact, probably most familiar to everyone in this room: it's a race track. And again, this is the government, so this couldn't afford to be a lazy eight. So we needed a lot more laps, drivers, and cars.

The magic of reps — this came here in the form of pilots, prototypes, and operations. The mission was to replace key parts of layer one and layer two, so the dominoes could fall, with the intent to scale the approach not just across Department of Navy (which is Navy and Marine Corps), but across DoD and federal through partnerships. Does that sound ambitious enough?

The first place we unleashed pirates and cowboys was naturally DevSecOps pipeline. We were the last service to have a pipeline-developer-on-day-one. Looked more like developer-on-day-90. We've gotten down to day 20 and dropping. You inspire us, we will get that down to day one.

And then ATOs. ATOs strike fear into everyone they haven't already scared off. We are now saving an average of 150 days per project, cumulatively over hundreds of projects, and that will keep compounding.

Everyone's very familiar with the linchpin of zero trust: identity. We are so good at that, that we have 1,500 different identity management solutions within the Department of Navy. Virtually one per application — perhaps more. A designated enterprise service to rule them and replace them all has now been put into place for enterprise use cases, and we've proved out disconnected and afloat use cases. The reason that we've pushed off for so long…

From one of the largest contracts ever scrapped — despite an awesome name, some folks may have heard of JEDI — we moved to a portfolio approach. A front door for all of our mission partners, for all major IaaSes and PaaSes — that entryway exists. And as of this morning, we are pulling in SIPRNet, or impact level six, partners. I got the email about three hours ago. This is the largest cloud instance that will be named soon the exemplar of zero trust in DoD. And we're hoping to share that a lot wider than DoD.

Over 1 million endpoints. And we did the math: more different configurations than particles in the known universe. Through unification and simplification, we narrowed to three paths for connectivity: cloud-managed virtual desktop, and BYOD, for everyone.

This last move amplified the next issue. Every organization — which is hundreds in the Department of Navy — are responsible for their own devices. Who are the two groups that carry around two cell phones? Drug dealers and government employees, right?

Endpoints-as-a-service are now real, and being scaled across bases, homes, barges, and ships. And the culmination of these activities, across 300 worldwide installations and almost as many ships, is simplified and standardized offerings for workstations, telephony, edge compute, tactical edge connectivity.

The connectivity on the USS — hmm, I could be on that ship right now, a little dry — the connectivity on the USS Roosevelt: 20 times reporting, 20 times better comms, and less friction, based on how that ship is moving and how the folks on those ships are able to work and get their daily tasks done. We want them to be more focused, in every case, on the mission at hand. So to Gene's earlier points: if we bog these folks down with minutiae, we are distracting them from keeping us all safe.

When we talk about this ecosystem being massive, you can think about the 670,000 users. We can think about the 30 to 40,000 industry partners — so many that we have to estimate the number of industry partners we work with. We can think about the 2 million folks who are working throughout the ecosystem. And then we can think about the billions of folks affected by how well these sailors and marines do their job.

This maneuver includes two of the ships that we've deployed the tactical edge on — tactical edge compute that didn't exist eight months ago. And every class of ship here has tried it out. In some cases 40% of functions that they're getting, with a 2% cost. 24 months of grinding — but instead of grinding as carpenters trying to build everything ourselves anymore, we have learned to grind as gardeners instead, learning to adopt instead of build.

Our business model no longer fit. And this is an important part: we could no longer deny it.

So you see, we have plenty of innovation, but adoption is the harder maneuver. Adoption, as you all know, is a layer-three problem. Darn humans in their social circuitry. So we had to create layer-three solutions, and we had to show that they worked for cowboys and pirates and they can work for everyone.

So what we did was, we said: okay, if we have an innovation solution, but we have an innovation adoption problem — well, we don't send warfighters into theater without a kit. They kit up. So how can we send innovators into the government without a kit? So we created the innovation adoption kit for all of them to become gardeners.

We started with the pipeline, or adoption funnel. The question I get absolutely most often is, "Can I see a roadmap for that?" And the roadmaps, in reality, are thousands of disparate groups all working. And so we had to view them as seeds and plants. And some plants grow faster than others. It's not what we started, or how much money we spent on one — it's what's ready to support our sailors and marines.

And so, by using a simplified adoption pipeline, we have said: hey, let's scout. Let's use other people's money — that's up to $15 billion. Let's pilot. Originally we were in a situation that I think probably some of you find yourselves in. We were in all-O&M — all operations-and-maintenance organization. So we were trapped. We were hostages to inflation. We were never going to get out of that do-loop without piloting. So we scrapped and scrambled and tin-canned for piloting money — horizon two. And that has allowed us to recycle and allow more shots on goal — those more laps and more drivers and more cars — to disrupt ourselves on a monthly basis.

And then the people. We wanted to find, from a Bayesian perspective, what percentage of people could change the way that they think. Some are born that way. There are fence-sitters. How far could we get up the stack? And so we started with about 1% of the organization. You're gonna hear from more of those right after me. And then we said: who else can we bring along? How big can we make this movement? Now we've worked with 10 different departments and agencies. We've worked with folks from every stretch of the imagination. AWS has been an awesome partner. We're pulling in folks who 10 years ago said, "Hey, I don't think the DoD mission is for me." Now they're saying: deterrence is peace, let's get after this together.

One of the things that we have the most trouble with — and I think this is probably only the government, this won't sound familiar — but prioritization is very hard. And saying no, or stopping something, is very hard. We have opinions. We have some data. The opinions usually win. And so we baselined how we're looking at decision-making and prioritization.

Very scary stuff. We have programs of record. People have heard this term. And your job is to preserve your program of record. To shift to capabilities, we had to work with Congress, and we had to say: hey, if we look at this in terms of portfolios and capabilities out, then we can really get somewhere. So we had to start looking at our projects as cattle instead of pets — not something to protect, something that allows us to make things happen.

So this is the framework we use for our decision-making. We've tried this in a couple different portfolios for, now, two years. We now know how many pounds of operational resilience $1 buys. It's a quantitative prioritization that we're using for all piloting and all execution. And that's spreading.

Here is how we present to every executive. I don't know if we're unique in this, but sometimes different executives like different views. What we've said is: here are the outcomes. If you'd like to dig deeper, we can show you, but how about we focus on the outcomes? And that has worked surprisingly well for an older, staunch, and hierarchical organization. So we are a dashboard-only presentation group.

And those shots on goal: structured pilots. Anyone can be a pilot if you bring an operational champion. Anyone is now allowed to sponsor a structured pilot — if you have the WAMs (the outcome-driven metrics, the World-class Alignment Metrics) to say "I think I can get something from horizon three to horizon one," there's a catch: if you can't get there in three to six months, if you're not on track, you go the way of cattle drive.

We now have pilots in almost every functional area, and we have folks from different organizations plugging these in. What we did was: we proved it out for the first 25. In 2022, we did two pilots. In 2023, we did 25 pilots. And now we have about a hundred active. And what we're saying is, we are going to measure you transparently, quantitatively, centrally. And if you're ready to pass Go, then we'll get together.

Generative AI is several of those pilots. We don't have a lot. We're on the wrong side of the K-curve until we turn that around. We have now made the case that AI is moving faster than fashion, and we have to get on this. So we are going to fortify that portfolio of generative AI pilots, specifically after yesterday.

Structured challenges. We needed to figure out who's unleashed in our organization. Like we said, we're massive. Who is there? Who can we open up the mind of, and how do we open our arms to accept more of them in? So we've opened up structured challenges. And most recently this is now in congressional language: everyone has to do hackathons. Now we are going to do more of these. We're going to share, we're going to partner, we're going to do those with other nations. We are pulling in academia like we've never had before, and industry was already there.

Adaptive roadmaps. This is a crazy idea, specifically within the government. So we love our waterfall requirements — and in three years and 2,000 pages, you will know exactly how to build something that you could buy off the shelf. Here we have tech-informed concepts of employment. So on the off chance that someone is stuck, not exactly sure how this should work after sitting at a desk for the last 10 years, we're asking the technologists what is the art of the possible, and then shaping top-level requirements and execution off of that. Very backwards. Very not-waterfall.

And my favorite, structured divestments. There's a quote from Reagan: "The nearest thing to eternal life on this earth you'll ever see is a government program." Our parking lot has been full for years, and we keep trying to pull new cars in without turning anything off — and it's not working. Everyone's tech debt is bad. This is an area where push comes to shove. There are crisis moments that force folks' hands. We're finally there. And so despite it being tough, we have cattle drivers — and the ghost of Rickover — as well as policy language and references that says this is now possible. We have turned off over — what — nine figures' worth of capability in the last 16 months, and we plan on going up from there, without any real trade-in what's being delivered in terms of outcomes.

Here's the deal. We are doing this and want to continue doing this through partnerships. We realized the hard way that our business model didn't fit anymore, and that we couldn't do it alone. We got by for too long being able to do it alone. So we are now working with all of the services. We are now working with more organizations than we've ever worked with before. For the last 15 years, since we've recorded the metric, we have decreased the number of organizations that we worked with. This is the first year that we're up, and we will continue to grow.

No one in our space has ever been around an expanding industrial base. It's been a generation. We are growing how we think about the public-private mesh. And in a world where whole-of-nation and whole-of-society is required for some of these daunting problems, we need to team, to grind, and to partner like we've never done before.

That's where it comes to you.

Number one: I'd ask all of you to revisit and reframe how you think about working with the government — how you think about the government in general, how you think about what's possible from the government and what needs to be possible with the government — in an increasing public-private mesh, as it relates to AI, as it relates to any major undertaking that we're doing. It's been 80 years since the Manhattan Project. We are together in pockets more than we've been in a long time, and we can up the ante on that.

Number two: contributing structured AI pilots with divestments. There are no new cars in without two cars out. So help us. If you have commercial off the shelf, if you have something, figure out which parking spaces you can better occupy with a compact car. And then quantitative wins. If you have brags in this space — whether it's Department of Navy or not — tell us about those brags. One thing the government isn't great at doing is telling our story. And so we're going to do that through stories and data. We want that scoreboard to post up. And it's going to take, again, a whole-of-society approach.

And then three: share. Share that this is not your granddaddy's US government. Share that we can do more together. Share that we need to belong to reach our potential — as an organization, as a country, as a world.

Thank you for your time.