Tracy Bannon: The organizations, all of them, are in there are a ton of them that are listed there. Pulling them under a Pentagon level CTO is a smart thing to do.
All of them do fantastic work on their own. But imagine if you get more cohesion between them, more intersectionality, between them, whether it's prototyping, operationalizing, each one of them plays a part in that. So now what you're doing is enabling all of the organizations that are part of what I call the value strain, all of those that are part of the value stream to come together.
I think you're gonna see a lot of parallel execution. Think of this, like, an agile, when everybody started to do agile, and you started to scale that out, and you started to need to have a scrum of scrums or a meeting of meetings. Individual teams met, then their leaders needed to roll up so that there was enough cross pollination to break down barriers that were there.
Carolyn Ford: Tracy Bannon and I started So What? Our government tech series, here on Tech Transforms, to take the biggest headlines and ask the question most people skip. So what actually changes? I'm your host, Carolyn Ford, back with your co host, Tracy Bannon.
We took a break, the world kept moving fast, and now the stakes are even higher. So What is back and today, we're jumping in with a big one.
The Pentagon is consolidating R&D, AI, and data under a single chief technology officer, and on paper, it's massive. Tracy is in this world every day, briefing at the Pentagon, seeing what's real, where the friction is, and what's still fiction. We talk about why this might work, where it could break, and how culture, data ownership, and speed matter more than the or chart.
This is So What? We're back. Let's get to it.
The Pentagon just rolled out major reforms in R&D and AI.. Secretary Hegseth is consolidating several tech offices under CTO Emil Michael. So, but the orgs that he's rolling up under this, and a lot of these, I'm gonna be honest, I've never heard of them before reading this article, but DIU, which scouts commercial innovation...SCO, which prototypes emerging capabilities, CDAO, the AI and Data Powerhouse. Do you agree with that definition?.
Tracy Bannon: That's what we need, that's fine. Their role is to drive the implementation and adoption. So, yeah, they are key players, big players.
Carolyn Ford: Okay. And the list keeps going. 
Big one, DARPA. This one surprised me. O S C and the new Mission Engineering Group at the same time, he's launching new AI initiatives from dronesworms to Enders's Game Style Sims, to Gen AI for the entire workforce, and breaking up massive Advana database did not had not heard that name.
Ooh! I know. So here's what I need from you.
I need you to take me to school, and especially with your vantage point. You're someone who is at the Pentagon on a regular basis. You're briefing at the Pentagon on a regular basis.
So help us understand what this new structure means, and what these groups, why these groups, under a single CTO, I mean, I like the idea of having a belly button, you and I have talked about, this is one of the problems in government in general. Nobody wants to own it. Everybody wants to own it, so nobody owns it.
So now we're kind of saying,Okay, we got Michael. He's now the owner of this. So talk to me.
Tracy Bannon: Well, you actually pretty much nailed it. The organizations, all of them, and there are a ton of them that are listed there, and there are others that will be associated with this. Pulling them under a Pentagon level CTO is a smart thing to do.
Right now, they've been very loosely coupled. Sometimes they worked together. All of them do fantastic work on their own.
But imagine if you get more cohesion between them, more intersectionality between them, I can see whether it's prototyping, operationalizing each one of them plays a part in that. So now what you're doing is enabling all of the organizations that are part of what I call the value string, all of those that are part of the value stream to come together. It doesn't mean that things are going to be hyper serialized.
I think you're going to see a lot of parallel execution, but think of it, like, I always start using my software engineering terms. Think of this, like, agile when everybody started to do agile, and you started to scale that out, and you started to need to have a scrum of scrums or a meeting of meetings. So individual teams met, then their leaders needed to roll up so that there was enough cross pollination to break down barriers that were there, to get better momentum, when there were things that needed to be dealt with.
This is a smart move. For what it's worth, the changes that are being pushed under Secretary of War Hegseth, and his organization, right? He's not a single man coming up with all these ideas on his own, but his organization.
They are addressing a lot of the things that we've talked about for multiple years. Right. For multiple years, and not just with this organization, but with a prior memo that he put out, I think it was the 9th of January.
And right before the holidays, the rollout of Generative AI at scale for everybody here. Boom, it's available to you on your desktop.
I'm going to use the term for he, is pushing fast, and we've wanted that momentum. I think a lot of people are going to get very nervous in building momentum. Because, you know, when you run, sometimes you fall.
There are going to be oopsses. They're always there, we always knew the potential was there. What we have to be ready for is how to react to those things when they hit us.
Carolyn Ford: So I like the way you describe how this is gonna change things in practice. still feels like we're trying to boil the ocean, though. I mean, that's a lot of really big groups that now we're rolling up under one person. Is it, do you think it can really happen or are we just talking about org chart shuffle?
Tracy Bannon: Um, I don't know if it will end up being org chart shuffle. I don't know that. I know that CRO has been moved.
It was originally under the, it used to be the DOD, under the DOW, under the CIO, CDAO was in that organization, in the CIO shop, then it got moved out. So equals, then this last year, not very long ago, it got moved under the office of R&E. So they, as a group have been moved around, but the charter of what they need to do hasn't changed. So think of it this way. The charters of all those organizations haven't changed. They still need to do. They still need to accomplish what Congress has asked them to do, what the government, what the military is asking them to do, what the war fighters need.
They're still all chartered to do those things. But now they should have visibility and to what one another are doing, without it being so formal. Right now, in the organizations as they stand, I'm not a government employee, I work for a nonprofit that works for the government. So for me to talk to certain people, I actually have to go up the food chain and then down the food chain. Multiple people just to have. 
Carolyn Ford: Really?
Tracy Bannon: Into another organization? It depends on what the conversation is about, whether I can go direct, its the same thing for the government folks, they have to follow protocols.
This will help, I believe, tear down some of those preexisting. Right. Now people are going to be sharing the same letterhead, so to speak.
Carolyn Ford: Well, I love that, and the article that I read is in Breaking Defense, and I'll drop the link in, the show notes. I mean, some of the friction points that the article calls out are some of these groups. Hegseth doesn't have the authority to move them.
Congress does. So there's that. Are there other friction points that you see with this?
Tracy Bannon: I think we always have to be cognizant of the friction that comes with different organizational cultures. Research organizations are different than those who are building and deploying something into a battle zone.
They just are?. Now, trying to serve the same purpose ultimately, but they have a different culture.
In a research organization, higher degree of failure. Why? Because it's research, we can fail.
When I'm building a weapon system and deploying software onto a new platform? Failure takes on a different modality, right? And we have to be much more careful.
So, risk, resistance, risk tolerance, even the way the organizations themselves rally internally, is going to be very different. So we'll see how much autonomy they have to stay to keep their corporate cultures, so to speak, in this bigger organization. Are they going to be like the different states that all make up the United States?
Or do they lose their identity? I don't think that they'll lose their identity. I don't think they can afford to lose their identity.
'Cause that's what helps to keep them who they are. They are gonna have to scrub some rough edges. That's the same no matter what.
As for the organizations, whether or not, the DOD, whether the DOW, whether Pete Hegseth has the authority or not, I think he'll, as we've seen with the art of the deal of this administration, I think'll make it It't matter. You make the move, you make the move, and then people will things will either get pushed back or they won't, but stretch the envelope, stretch the envelope. I don't know that it's the worst approach.
I don't know that it's the worst approach, because bringing them, I do believe that bringing them together and giving better visibility is a good idea. How does he get Congress on board with that? I don't know.
Carolyn Ford: Right. So. I mean, like I said, I and, like, you and I have talked about even for years, we all agree we need, um, to put it crassly, one throat to choke.
We need a leader that takes responsibility and that seems like that's what this is. You talked about generative AI, and Pete Hegseth saying everybody's gonna use it, making it accessible.
Even in corporate world. I know companies that even still do not allow the use of generative AI.
They're too scared of it. So the fact, so when you say he's telling everybody to use it, what does that look like on the ground? What do youmean?
Tracy Bannon: It's controlled. So, he's not saying, "Go out and sign up and just use the free tier." He's not saying that. Repeat, he's not saying that."
Carolyn Ford: Well, can I just say none of us should use the free tier either?
Tracy Bannon: Well, that's correct, Unless you're, like, say, "Make me some chicken jokes or I need a recipe, and this is all I have ingredients in my refrigerator.
Yes. Correct. Correct.
These have been, they're rolling out four different options, four different services or service providers. So, right now, they've rolled out Gemini from Google. They are working on the conTracyting for the other three, which is Grock, right? Which is X, open AI, and Anthropic. Claude Anthropic. And it's not just a model.
All the players. That's what's important about this all the players. And they're not just conTracyting for access to the single model and the single chat interface.
That's what they're ruling out first, because that's the quickest win. I can get this onto the desktop right now. If I log in to my nipper, nipper net, um, which is a non government network.
If I log into that, I can go to the link, and I can go to a chat interface, and I can be interacting with Gemini at the classification level that is appropriate to the network that I'm on. That's important too.. They've negotiated this so they are at the correct security classification level.
So what's he doing? He's actually tearing down the risk. He's burning down that risk.
He, again, he. They, in their smarts, are burning down the risk by saying, we're going to do this fast, because what's been happening for the last couple of years is piecemeal parts. Somebody might go out conTracyt for one organization, they might get a small conTracyt, they might go out and just shy of credit carding it, right?
Putting it on the purchase card. or worse, somebody goes home and says, "I'm just gonna use perplexity myself, and my own, I'm just gonna do this on my own." So they are incentivizing people by making tools available to them and training. There is a community practice that's already stood up on this, that's available to anybody with a dot mill account.
So you can join the community practice. They're teaching people how to use these things. So I think it's smart, but I want to touch on one other piece about this kind of organizational structure.
You also have to consider that it is going to make a big difference with conTracyting an acquisition, because each one of those organizations right now. may have to do it on their own, but you're talking about building an authority that can buy on behalf of the whole DOW. \\
Carolyn Ford: And just make it accessible. It's accessible now.
Tracy Bannon: And just make it accessible. So there's some sexiness to it. You're going to see some friction here or there.
Data ownership is always a friction point. Who owns what data? I mean, even when I'm dealing with a specific mission area, I have different organizations that own different parts of the data.
And if you create the data, you own it. You're that steward. And sharing that, there's a it's my system, therefore, it's my risk mentality.
We'll see, that'll have to be worked done. That'll have to be worked on.
Carolyn Ford: The Breaking Defense article made me think that this data is supposed to be shared and easily, easily shared between the organization,.But that seems like, again, I love the idea, and how is that gonna really happen?
Tracy Bannon: Well, um, I can't go into any details, but Sorry.
Carolyn Ford: No, I know Tracy, listen, we've been friends long enough.
I know I know the game. 
Tracy Bannon:  So, um, things that I can tell you. There are, in the Department of War, there are there's this construct called being a systems of Record.
And there are other systems. Doesn't mean they're not equally as important, but a system of record is where data originates and it's considered to be a gold copy of that, whatever that data piece is. Now, data gets propagated all over the place, copied, the same thing is that we have.
If you think about the IRS, or if you think about big health tech companies or healthcare companies, they have the same kinds of problems. Everybody has those kinds of problems. But the system of record has certain responsibilities.
There are other very important systems, whether they are for planning, whether they're HR, that might not be considered a system of record. and how they get the data from the system of record or how they collaborate with it or coordinate with it takes a bit more, takes a bit more effort. So what we're gonna see is we're going to see a renewed focus on data ownership in understanding the data universe and in democratizing access to data. doesn't mean everybody gets everything. It's still a need to know.
It's still based on classification level, but putting into place the infrastructure, in what I call the data backbone, right? The way that you can plug in and I can discover, oh, let's make up a silly example. I need to find all of the rideshares that are available in my town.
I can go out to this list and get a list of those ride shares. And that's a democratized list, and people keep it up, and I can just subscribe to it whenever it changes, I get an update that it's changed. Like, that's a great little data service that I could get.
Imagine how much data there is in the DOD.
Carolyn Ford: Right..
Tracy Bannon: And making it available, safely, they've already been working on it for a while. This is not as though it's a magic change that they're just waving the wand right now. There are existing platforms I know of entire groups that do nothing but focus on data management and data governance within the federal space.
Carolyn Ford: So I want to shift gears a little bit. I mean, this new org under Emil Michael, like I said, I'm a fan of having a leader and it's all war focus. get it..
It's Department of War. Okay. It Tracyks with the mission, but what about all of the other problems that need to be solved?
AI for humanitarian problems, solving disease, world hunger, global warming. Where does that all go? It's, I think it's still siloed under a dozen or more different groups, right?
Tracy Bannon: Well, remember, at the end of the day, this is defense
Carolyn Ford: That's right.
Tracy Bannon: And they're making the play to make it as effective and efficient as possible. Defense is not the job of defense, it has not been to solve humanitarian challenges, though it is to support our humanitarian efforts.
So it doesn't change what these orgs have been asked to do either. Defense is still defense. Who owns AI for humanitarian and global problems?
I think that the CTO will need to kind of treat it as out of scope a little bit, but keep apprised of it, right? So it's not theirs to solve.
Carolyn Ford: Do you think we see a CTO on that side of the house that Emil Michael, like there were peers?
Tracy Bannon: Outside of the DOW, I don't know. So the DOW is a little bit more top down than all the rest of government. So where would that CTO sit over what would they sit?
Would they be over? Health and Human Services, right? Would they be under Robert Kennedy Jr.'s organization?
Well, there could be one there. That would make sense, and maybe centers for Disease Control, Medicare, Medicaid. Like, I could see one in that area, that domain.
And then what do we do with the treasury and IRS? Well, that's another domain. And I worry there is very different.
I'm not worried about humanitarian. I'm actually worrying about how much compute that they need for all the AI that they're going to need to deal with all the problems that they have, right? When you dry the world's Lakes.
Carolyn Ford: If we keep building these data centers. I know, we are, and as we're trying to solve the world problems, it's like we're in this, this paradox, this, you know, deadly cycle that we're using AI to try to solve the problems that AI is causing. We are oversolving with AI right now.
Tracy Bannon: Yeah. I've had multiple teams that I've met with one very recently again and something that they were attempting to do.
It just needed an algorithm. It didn't need a large language model. We didn't need to find two in the large language model.
We didn't need to generate tens of thousands of training set questions, no algorithm, just math. Let's do some math. Get up on the board, let's do some math.
We are attempting to solve things that are not problems. We're still in the hype cycle. I like to say that we're tobogganing down into the trough of disillusionment right now.
And I can't wait till we hit rock bottom till more people are like, we waste a lot of money. That was really risky. And I say this about everybody.
Government or non government doesn't matter. They're all just jonesing on AI to do magical things. It's not magical.
It's marginal, it's cool, it can be really effective. But like you and I talked about earlier, I don't know that I could be as effective as I am right now without it, but I also know how to use it, when to use it, and I know when to turn it off. 
I know when it's actually harming what I'm doing, when it's like you realize that it will amplify you. So if you care and you're having a very detailed back and forth conversation, you will get a better result than if you're kind of not invested in it, and you'll kind of start to get garbage. And the language model will start to really get tunnel vision in how it's responding to you.
Carolyn Ford: Oh, yeah. The conversation goes on. It becomes an echo chamber of garbage. I want to go back to what you, I think you already touched on this. So, genai.mil.
We talked about how talked to me a little bit more about how it's going to scale securely. Are these just contained instances? And maybe you already told me and I just missed it.
Tracy Bannon: The providers right now, we've already had the architectural constructs in place for the infrastructure. Think about AWS has a gov cloud, and they don't just have one. They have them multiple because they have data centers that are located in the United States with vetted American citizens only running it.
And I'm going to tell you that an important part of what we do in the DOD is having vetted Americans that are working on these things. It's a really important part of it. So they've already had a cloud that was available in that way.
There are networks that are available at every classification level. AWS provides them, Microsoft provides them, and these others do the same thing. So they have, they are protecting, they're providing instances within each one of those classification levels.
It's not as though there's one big mega AI to rule them all and that you just filter out into each one of the environments, depending on the vendor, depending on which organization is, they probably have it implemented on the backside slightly differently. But there's a tremendous amount of effort that goes into what we call infosec, Knowing exactly how the information is flowing and the security infosec, the security that goes with it. So every boundary that that little piece of data goes across, who can see it, who should see it, who can't see it, is it getting written to a log file somewhere that we have to secure it?
All that stuff is being dealt with with each one of these. So it is very secure. And it's only available in those environments. I can't from Nipper, I can't just go to regular chatGPT. Like, there are things that are blocked, and they should be blocked at these different environments.
Carolyn Ford:Most of the most of the environments are not on the regular internet. We've talked about all the big players are coming in, Gemini, Grock, and. OpenAI. There we go, and Anthropic.. That's all of them.
So I think it should be obvious, but let's talk about why that's important that all the big players are coming. I guess start there. 


Tracy Bannon:So the couple of reasons.
I hate it when people talk about how much they hate vendor lock in. Yeah. But this is one of those times where we need to not have lock-in.
We need to have a better way to say it is we need to have options because no one model is perfect. No one I use multiple models every day, multiple models, because different ones are better for different things. Same thing has to be available to the Department of Defense.
We need to know different models for different purposes. So I think that that's one of the biggest pieces of it. Plus, what if somebody decides to drop us, right?
You can't have all your eggs in one basket. It used to happen way back in the day that, especially when I was doing steak government work, they might go all in with a particular vendor. And when they went all in with that particular vendor, and there was all of this benefit of having proprietary XYZ, it's so fast and efficient, and they're managing it all for you, and when they went belly up, or were sold, your data was locked away, you lost those things.
So I think where there's a bit of protectionism in us making sure that we are broadening and not narrowing the industrial base. 


Carolyn Ford: I just think about human nature. 
Even about my own practice, I bounce around between them too, and I have my favorite, where I spend most of my time. And if that favorite went away, it would hurt for a minute. I mean, but I just, I can't help but think, even we talked about these different organizations cultures, they're probably gonna find their favorite and go all in on that one anyway.
Does it matter? 


Tracy Bannon:Does it matter if they go all in? No, it doesn't and it doesn't.
If you have a need, find the right tool for the need. period end of story. That's what this is allowing them to do. There are organizations right now who had already started to use what's called Ask Sage?
Ask Sage is Nick Shalan's offer. And really, what he's doing, I'm going to oversimplify it. He's brokering all those different models at the different classification levels.
And he got in there right away. And so he enabled people to get out a number of different models, a number of different kinds of tools. But ultimately it was up to you to define what's your why.
What are you trying to do and which one would be right for you? This removes, there's no broker there in the same way. Now, Nick still has some really awesome capabilities that GenAI is not going to be providing anytime soon.
Carolyn Ford: Well, he's kind of built, like, the agent, for lack of a better word, right?
Tracy Bannon: Yeah, he's built the layer, a beautiful layer on top. For those, kind of a broker, all kinds of stuff.
He's got all kinds of good stuff there. And it's all positive. It's all goodness.
And he's still a key player in this space. If you think about the Army conTracyt, I believe it was Army conTracyt that he got last year offering it out to the PEOs. I think, at free.
It was something amazing, but it was provided to them by the Army. They bought it, said,Here are your tokens or your subscriptions or whatever mechanism that they used. Some of them are now going, have this.
Now you're offering me Genai.mil. That's going to be less expensive for my program, for my own personal budget, but I don't have all the capabilities that I have over here yet. They're not there yet.
So they're really desperate to understand road maps, because if they can if they can depend on centralized, you know, providing of capabilities as opposed to having to conTracyt at themselves, hopefully this should also run the prices down. Right. If you're buying that large at mass, it shouldn't be a $600 hammer.
It should be a $6 hammer.
Carolyn Ford: So, under this new consolidation of R&D and AI, one of the things is this Advana database. I think it's a database.
It's a database. It's a mega database, right? Like I said, I didn't know what it was till I read this article, andl talk to me about the significance here.
It looks like it's gonna be split into three different groups now. I mean, do you have any concerns? Do you think that it's a positive thing?
Should we even care?
Tracy Bannon: I can tell you what Advana was supposed to be, and I will temper my judgment on if they achieved it.
Carolyn Ford: Okay.
Tracy Bannon: Advana is this enterprise data and analytics platform. It's been around for a bit.
I don't remember anymore, which consultancy or which organization implements and runs it on behalf of the government, matters not. It was supposed to be a centralized place for all that authoritative data. Remember, I said there were systems of record?
Authoritative data should go there. And there should be deduplicating, so we don't have duplications at all and what's the phrase we use? Oh, single pane of glass.
Single pane of glass, right? Especially when it came to Congress facing things. I will say that.
From observation, it's become very complex ecosystem. It has many different types of use cases. Sometimes mixing some storage and some analytics and some applications together.
So I think there was some blurring that has gone on. I think there's sometimes you get to a point where it's like Mama Bill, and they broke up Mama Bell and the baby bells. Right?
I think there's a little bit of that happening here, to have clear architectural rules. You know, I talk about Conway's law, right?. Your organization, and your architecture in the way that you operate will start to take the shape of your organization, right?
So back in the day, I'd have a middleware team, and that was separate from a database team, and that was separate from my XYZ team. I think he's trying to get to clearer data ownership, I think. I think that's a piece of it in making sure that everything is secure from the ground up.
Carolyn Ford:It looks like, I mean, exactly what you said. Like, he's breaking it into three different, for lack of better term, use case groups. That's what it looks like.
So, I think this goes back to data sharing, and I know you work a lot with CDAO. He said,Hegseth was quoted in the article saying, "No more denying CDAO data requests." So, again, this sounds like a good thing to me, where we're sharing the data, but in practicality, what do you think?
Tracy Bannon:Okay, so, even though they may not look at these are rose colored glasses, and the rose colored glasses. The rose colored glasses believe that if we define clear data ownership and we have strong governance and well defined interfaces with a focus on security, those principles sound modern data architecture. That sound modern data architecture.
If those principles happen, as we're dividing into these three groups underneath, I think it could be a wonderful thing. Okay. If it happens in name only, and you still have I call it data hoarding, right?
Carolyn Ford: Right..
Tracy Bannon: It's my data, you can't my data. Or if there's cultural resistance to sharing, then we will see a slippage. We'll see things kind of fall back to the way they are. This is no different than any other time where we have grown to a point where we need change because it's unmanageable or unwieldy.
If Advana is being split up, if Ma Bell is being split into the baby bells, there's a reason for that monopolization to be split up. And what I'm seeing, right, and what I'm reading, I'm reading the same things that you're reading. I'm not actually having conversations about the Advana breakup. is that they are looking to take it into those different domain areas, those different missionaries.
Carolyn Ford: Right,. Okay, so. current director of JAIC, former director of JAIC, Jack Shanahan, Mike Groen, both express deep skepticism about this whole thing. So what are the signs that you're going to be watching to see if this boiled the ocean strategy actually delivers results?
Like, what are the first signs that you're like,Yep, this is working?
Tracy Bannon: So I don't know that it's a boil the ocean strategy, so I don't know if I agree with you on that.
Carolyn Ford: Okay.
Tracy Bannon: I think I because there's a lot of big messy areas. And so trying to do mass cleanup is, I don't necessarily think it's boiling the ocean. I don't want to take Mandarin or Cantonese.
I just don't want to learn those languages. It's not part of what I want to do. A little bit of an interest in Russian, but only because my daughter studied that for a while in college.
I'm going to continue to look for the same things that I look for right now. What's the attitude? What are the values that we're looking to bring?
How do I measure it? If I have those things, if I can observe the culture, if I can see how things are changing, and if I can see a measurable difference, boom. I mean, those are the only three things that we ever look for.
People processing tech, yes,So I'm going to be keeping a watch on those pieces of it.
If I see that people are getting, you know, a boundary and hiding what they have, it's gonna speak, you know, negatively. If I think that likelihood of having some challenges with policy as it's emerging, the policy is emerging and changing. I think that's probably the I think, most difficult area.
Because policy is already difficult to understand and implement, and now we're changing things up and putting different voices in the mix. So I'm going to be watching, I'm gonna be watching the people. Are they playing nice?
Do they intend to change? If you think about the Hegseth memo that came out on the ninth, he essentially said, "Go do this." Like, he didn't say, "It would be nice if he didn't say, "These are only policies."
He said, "People are getting trained.You need to use this now. This is available to me. Prove to me that you can't."
Carolyn Ford:And you're witnessing that happen, Mm hmm. Right? So I am. That's the other thing. Okay, since our so what series, this is the beginning of the beginning again, of our so what series, I will make a note that we're going to check in, let's even just say three months from now, because you're on the ground.
You're going to experience this in real time. I mean, one of the things that maybe you'll be able to see is that ability to laterally connect with people that you haven't been able to. That's crazy to me, that you're still doing that archaic up and down the command.
Tracy Bannon: Military. It's military. It's the command change in the military. Right.
You know, if I meet you up at the local bar at the coffee shop, yeah, I'm going to have the conversation. But, I mean, anything that's formal has a formal route. Right.
Carolyn Ford: So that there's one thing that you and I can Tracyk is your ability just to talk to people and to get things done in between the different orgs. What's another little thing we should Tracyk?.
Not little. They're not little. Like, they will make a huge difference in how you do your job and your ability to get things done faster.
Tracy Bannon: I think part of it is going to be... a combination of speed of adoption  and mistakes that come along with that. Those two things, because we want people to adopt these things in a safe and effective and efficient way, not just use it because I told you, you had to, but use it because it makes sense.
Use it because it will help you. So those are the things that I would be looking for is adoption rate.And that being a placemat of the different kinds of folks, I can tell you right now that all of the Department of War is not magically generating all of the code, right? They're not.
They're not. Nor should they. Right. They're using it appropriately.
So I think it's going to be watching that placement. And then all of the surrounds for it, all of the surrounding things is the policy catching up with how I want to do these things, for example.
Carolyn Ford: All right, I've written them down.
We're going to check in, in three months back just on any episode. We can just check in on this. Okay, we have to do our questions because it's my favorite part.
You are handed the keys to Ender's Foundry, which, by the way, I was a fan of what he's doing just for this, because Ender's game is one of my favorite. My favorite books of all time, but I digress. You're handed the keys to Ender's Foundry.
What simulation do you create first?
Tracy Bannon: I think.. Congested, contested logistics, and cyber degradation scenarios.
That's where I think that's. There's a lot of stuff for us to look. Like, you know, disaster logistics and those things, but. cyber attacks and focusing on dealing with cyber attacks coming at us, people don't realize the volume of those cyber attacks.
I tell a story that is over a decade plus old, where I was working with a state. I wasn't even working with the federal government. I was working with a department of Health and Human Services in a particular state.
We had a production environment, and we had a staging environment, and everybody said how how secure our production environment was, but I showed them, gave them all their on their phones, had them pull up a nice little dashboard to show them where people were hitting that site from the staging site, because it was a publicly facing,cause we had to test it in the same way with accesses. Turns out, we were getting hit from China and from a lot from Germany, we were getting hit from the Russian area and why would anybody care about a state's Health and Human Services site? It's because it's cyber rich.
So, right, in terms of, I want information about your population and that was one of the largest states. So my point is they're attacking us. You just don't know the sheer volume and depth of what's going on right now.
That's where I want to see us.
Carolyn Ford: All right. That was another thing that I really liked about  what I think is coming out of this restructure.
Is that we're willing to say, they're attacking us and we're calling a spade a spade. Like, stop. And we haven't in the past.
So, okay. All right, you create agents all the time, so you've probably already had this happen. But your worst case scenario, an AI agent, has access to your calendar.
Like I said, I think this is a real life scenario for you. It is a real life scenario, and decides to optimize it. What chaos ensues or has ensued.
Tracy Bannon: For Tracy Bannon, and personally, I don't let it have access to my calendar. I won't use it on my calendar or on my email. I don't connect into any of my personal things.
I don't connect into any of my communications streams. That's intentional. Now, that having been said, if you were watching this over at the weekend, I think it's Maltbook.
Have you followed this?
Carolyn Ford: No. I don't know what it it Maltbook.
Tracy Bannon: I think it's Maltbook, M-O-L-T. Yep. Okay. M-O-L-T, B-O-O-K. AI.
Okay. There are two things that happened in parallel. Maltbook is a social media site that can only be used by AI agents, a social media site for AI agents.
It just came about. And there's also something called claw bot.
The problem is, he's got so hot and so popular in the span of five or six days, Hundreds of thousands of downloads. And for this malt bot, where I say, I have an agent that I've created, I'm gonna watch it.Cause you can watch. Humans can watch what's happening on this social engineering platform.
Literally, they claim that over a million agents have been connected in, was it turns out, every single moltbot agent, every single one has exposed the API keys of the owners. Because remember, your desktop, your own personal desktop. has not been secured for the type of interconnectedness and system of systems things that we think we want.
So it's already happening. It's happening. It's happening today.
So. No, I don't. I don't.
I use agents all the time. And I don't let them. I certain rooms in the house are locked.
Yeah, they can't go in.
Carolyn Ford: Well, you're the one that first alerted me to the zero click exploit. And I'm like, "Yeah, this is why you do not let AI answer your email.
Tracy Bannon: That's correct.
Carolyn Ford: All right, last question. Okay.
If you could assign one AI project to tackle a real world problem outside of defense, what would it be?
Tracy Bannon: There are two. One is dicier than the other.
I'll give you the dicey one, but then I'll move on to the one that I think is better for humanity. The dicey one is what happens with social engineering. I would like to double down on removing social engineering, figuring out ways to obfuscate who Carolyn Ford is so that people can't go after you, figure out your demographics, Target market you, the way that they do, manipulate you, manipulate you with a different kind of algorithms.
So I would love to solve that, because I think that one of the biggest saddest things that's happening in the world right now is how each one of us is getting a message that's been tailored to what the algorithms think we want to hear and turn our opinions that way. And I still felt for it, and I know it's happening, and I still fall for it..
Tracy Bannon: I love my Instagram feed. Holy moly. But I won't tell you what's there.
But I love it. What I really think would be also helpful is take all of the different type of tech that I'm already doing and apply it to disaster logistics. So similar, we have we have missions where we need to figure out what, sociotechnical aspects of it.
Same tech, different stakes. but immediate human impact. So how do we make FEMA actually helpful?.
How do we do that at the different state levels? How do we do that internationally? So that would be the second place where I would focus.
Carolyn Ford: I love it. All right, where can our listeners connect with you and learn more about your work?
Tracy Bannon: Um, the easiest way is to find me on LinkedIn.
Just look me up. Tracy Bannon, my email address is there. It's Tracy at TracyBannon.tech. Shoot me an email if you're not already connected with me on LinkedIn, and you can also check out my website, which is Tracybannon.tech
Carolyn Ford: I mean, it's worth going to your website just to see that cover picture. I love it so much. You're baiting people.
Tracy Bannon: That's fine. I totally baiting people.
Carolyn Ford: I also dropped your email in our show notes and your website in our show notes.
So thanks, Tracy. I'm so glad we got so wet back. We're back, we're back.
Tracy Bannon: We're back. Woo hoooo. I know.
Carolyn Ford: All right, listeners, thanks for joining us. Share this episode. Please give us a review.
It helps this content reach more people that could benefit from this conversation. Tech transforms is produced by Show and Tell, and tell next time, stay curious and keep imagining the future. Thanks.