#577 Agentic AI Is Rewriting the Boardroom: Betsy Atkins on Governance, Risk, and Execution

Mehmet: [00:00:00] Hello, and welcome back to a new opposite of the CTO Show with Mehmet today. I'm very honored and privileged to have with me Betsy Atkins. Betsy. She's very well known for many, many things. She sits on multiple companies boards. She's been an entrepreneur. She have a fantastic journey, and she's currently serving on the Google Cloud Advisory Board, and she sit also on boards like goPuff and Wynn Resorts.

Betsy. Thank you very, very much for giving me the time. I know how things can be so busy. So you gave me your time today to have this wonderful conversation The way I love to do it, Betsy, although like I know like many people would know you, but still I would love to hear your, you know, story. Like, you know, maybe a little bit about how you started it, what have you done, and then we can maybe dive into some, uh, of the.

Topics that we're gonna discuss today. We're gonna talk about organizational speed. We're gonna talk also about, um, you know, how AI is shaping, uh, you know, the alignment of the [00:01:00] board. We're gonna talk also about, uh, maybe gonna touch on the private equity VC part of the house also as well, because I love to, to to cover this topic.

And yeah, um, people know that I like to keep it conversational. So without further ado, thank you again, Betsy. I will keep the floor to you now. 

Betsy: Well, hello Mehmet and hi to all of your listeners. It's great to be with you. Um, my background is as a serial entrepreneur in the tech industry. I started writing business plans and getting venture capital and my first company was in telecommunications.

Uh, and as we were taking it public, we got acquired. I, um. Then, uh, have served on a, a series of corporate organizations, uh, running large parts of the company as, uh, the systems integration group at Unisys and things of that nature, as well as, [00:02:00] um, in advanced research companies. One of America's big, um, great secrets is one called BB and n, both Boran and Newman, that created.

The apon net. That is our internet today. So, uh, they had massive neural systems, uh, war game software simulation, encryption systems. And um, then I co-founded another company that went public named Ascend. It grew to 5.4 billion in revenue. We took it public. And then after that I did some early stage venture investing, and now I sit on public corporate boards.

Private corporate boards, I chair advisory boards like Google Cloud. Um, I work with private equity and venture, and I'm very passionate about the topics we're gonna cover. 

Mehmet: Great. We're gonna start from, you know, being on boards. So, and you've been, [00:03:00] you've been in this for, for a long time. What have you seen the biggest shift that happened on how boards used to operate and how things are, are, you know, moving today?

Betsy: The big shift has been that in the past boards were a little bit more passive, hands off, and they were oversight and. The dynamic of how fast things changed was much slower. Now boards are more leaning in to understand the, uh, the new themes in technology from. You know, everybody understood machine learning and got used to that AI and now large language models now agentic.

And then how do you orchestrate this? Quantum is coming next. So boards are learning that they need to have, uh, a better understanding of tech in order to future proof their companies. 

Mehmet: Great. [00:04:00] Lemme ask you also this, Betsy, how you decide to join, uh, a board seat, because you know what I noticed like you are on different.

You know, industries also as well, although like you were a tech, uh, senior entrepreneur, but I can see you are on the wind resort, for example, which, you know, like the, the casino that's gonna open, uh, I think very soon here in the ue in Russell Hema. And you are on multiple, in multiple places. So what is the main criteria for you to accept a, a seat on the.

Betsy: The main criteria is that you have really strong conviction about the CEO's vision, um, and you have good chemistry, and you, uh, believe that you're gonna be able to add some value and be helpful. Um, you shouldn't do a board where you, you don't have, uh, you know, where, where you don't like. What they're doing and, and you're not excited about it and there isn't a way that you're gonna add value, but everybody from a [00:05:00] tech background fits every industry now because tech is the connective tissue.

Mehmet: It, it is indeed. Like, you know, I, I remember the saying that every company now is a tech company somehow, you know, and, um, I know like, it's a cliche, but it's true because, you know, they're saying like, Starbucks, they don't sell coffee actually. They try to, to optimize using the tech technology, you know, the taste of their customers and serve them to, to the best, right?

So staying, you know, staying within the, the corporate organization and with, with the boards. I've seen an article you brought about corporate, you call it cholesterol, and that was like eye-opening for me. Like I, my eyes were like open like this, cholesterol corporate or so was that. So it's the idea like certain board members and processes act like blockages.

And by the way, I hear this a lot and I think, you know, I discuss it with some of my guests and always we refer to how [00:06:00] when companies grow. After certain level they face this. But let's break it down with you, petsy to my audience, how that, you know, corporate cholesterol looks like in practice. 

Betsy: So. You know, if, if we think, uh, all of your audience has worked in companies and you see that companies go through different parts of their life cycle, and in the early dynamic part, when you're growing a company rapidly, you need one particular, you know, set of skills for people who've gone on that journey from 50 million to 500 million.

And then when your company goes to from 1 billion to 3 billion, it's different. So we know that we forward, hire and rebuild our leadership team. We outgrow people. Well, the same thing happens with the board. They just don't have a good refreshment mechanism. And I, I think that what we see as companies go from growth to [00:07:00] value and then they go to legacy.

Um, the um, uh. The ability to keep your company relevant. Uh, it, it, it comes from the top down, you know? Are, are the leaders leaning into tech and innovation and new business models? And is the board change adaptive or are they hesitant and change resistant and unsure and, you know, very risk averse? You know, they're, then they become the cholesterol, the blockers.

Mehmet: Can we relate this little bit to, you know, it's from the famous author, Clayton Christensen, the Innovator's Dilemma Also as well, like, you know, boards at the end of the day, especially if they are a public company, in back of their minds, we need, you know, to be serving our shareholders. And if the company is public, that's like more.

Visible because you know, they are on the stock market. Everyone can go read their quarterly report, their yearly reports and so [00:08:00] on. And I believe also, 'cause I have experience in, in, uh, private companies as well. So if you are backed by investors, whether they are private equities or VCs, also you have to go and report back.

And I hear it a lot bey. Like the big thing that happens is, oh, we think, what if we do this? What if we try to change. And all of a sudden our revenue drops or for example, what if we do this and for a short time we start to see like people are moving to competitors. How has. Board col in collaboration with the founders, of course, and, and with the CEOs.

We can encourage this, you know, mindset I would call it, of being innovative and trying new things to make sure that you don't, uh, become irrelevant at some phase. 

Betsy: Well, most companies do become irrelevant at some phase. Um, you were talking about the great authors of business books. [00:09:00] You know, Jim Collins back in the nineties, uh, said that, um, 40%, so almost half of all of the corporations that are public, they disappear in 20 years.

In 20 years they don't exist. And in 40 years. 60% are gone. But that has compressed, uh, incredibly. I mean, CEO tenure has gone from 15 years to four and a half the time that you would be on the US Stock Exchange, top 500 companies, the Fortune 500, uh, and the s and p 500 index. Uh, that's gone, you know, from 15 years down to about six.

So the rate of change is faster, which means. Leadership teams, CEOs and boards have to lean into change and be change adaptive, or they are going to become irrelevant much faster. [00:10:00] And when I look at the boards, I sit on private and public. Uh, the private boards all have goals to take out. Head count, 10%. Uh, at least.

And these are the growth companies using AgTech. So they are looking at freezing, uh, organizational headcount and taking out, um, to drive innovation. Um, the public boards are, you know, it depends on the industry. If you're a consumer facing industry, you're gonna be a little more innovative than if you make chemicals or something.

Mm-hmm. Um. If you are in a real time industry, uh, like banking, uh, healthcare, uh, you're gonna be more innovative. Uh, so. You'll see different appetites for change relating to the industry, but in the end, it's driven from the [00:11:00] CEO and the leadership at the top. And that's why I, I said the first thing I look at is, do I love the CEO?

Do I love what they're doing because I'm, I'm not good at watching the paint dry. And I, I, I leave those boards 'cause they're boring and I can't help and, and I'll just give them indigestion, you know, I wanna move faster. They don't, so I. 

Mehmet: Yeah, 

Betsy: I find 

Mehmet: I gotta ask you something regarding this, so. Because you know, there is this, and you talk about it also as well yourself, like within the research that I did, you know, the difference between what we call, you know, board that are custodial versus board that are built for founder led.

Right? Now, of course this allows founders to be fast in taking decisions. They, they will be able to, you know, have the freedom to go and try. Now I gotta ask you something, which, which I know in the US. We think more and more of, you [00:12:00] know, the, the new model where they're leaving little bit to the founders, you know, to, to have kind of, yeah, I would call it freedom to, to do, actually, I can give an example.

So meta used to be Facebook. Like we, we saw Mark Zuckerberg like try to go to the, you know, the metaverse and then now he's seeing the AI saying, okay, drop the metaverse, go to the ai. So, and this is understood that. You have this mindset, uh, that the board have trust in this founder. How we reach there, and I'm talking here from, you know, the MENA region where I'm based.

So I'm based in Dubai. We started to have the ecosystem, the startup ecosystem. Uh. Kind of maturing, but we, we, we are still maturing. We are not there yet. And majority of the boards, you know, when I hear people talk, they still have the old mentality how we can, you know, what's your advice to founders when they want to bring people on their board to make sure that they choose the people who gonna allow them to have the [00:13:00] founder led high ity companies?

Betsy: So, um, even in large cap public boards, I think having an innovation mindset, given the velocity of change now, uh, is important. So even, you know, your premier, uh, you know, Emirates Air or Aramco, they're, they're, I think, pretty savvy and they're going to. Lean into change and be sensitive to it. That's why these companies are performing so well.

Um, if, if it's a founder led company, um, I think the, it, it's very valuable if you pick, um, your board directors on a couple of qualitative, uh, characteristics. So the qualitative character, um, that, that I would look for. It's the difference between resume and reality, right? You can get somebody with a [00:14:00] great big, fabulous resume.

Um, I am the former CEO of Pepsi. Well, um, I, Pepsi doesn't innovate that much. They're a great big, large cap consumer packaged goods company. Um, uh, so I would, I, you know, that person might have an unbelievable resume, uh, but are they really going to be able to understand, you know, your, uh. Pace the urgency.

When you're a smaller company, they need to innovate, they need to be nimble, or are they gonna be bureaucratic and custodian in their mindset? So you are looking for the difference between the reality of the person versus their resume. And I think that people get confused and they see a great resume and they say, oh my God, this is gonna add so much credibility.

Well for like a blink, and then you're stuck with the person and they're not, they resist change and they're afraid of taking a risk. And you know, we [00:15:00] are in a time now where there has to be some amount of risk, fail fast. Um, you know, but you need to try things and, uh, you also want. To pick board members that are gonna invest in you, mentor you, extend your reach, you know that, you know, I believe as a part of my philosophy is, um, my relationship capital of the people I know, I bring that as an asset to help, 

Mehmet: right?

Betsy: My leadership team, my CEOs, the company, uh, as I hope to be, become a thought partner with the CEO and the other board members. 

Mehmet: Um, you know, I just on the point of Pepsi, I think, you know, like we know the story of, of Apple and you know what, what happened, right? Like when Steve Jobs, like the, you know, he, he, he brought, uh, uh, the CEO of Pepsi there and you know, like.

[00:16:00] I, I, I don't claim, but yeah, I, I say I, I have good pattern recognition systems, and many times it happened with me when I saw a company hire someone n. Not necessarily a board member. It could be a sales leader, it could be a marketing leader. And they hired him or her just for as you said. Oh, like they were the CROs at X company, or she was the CMO at Y company.

Yeah. And I was telling them, uh, guys, I, I don't think this gonna work because the DNA is different. 

Betsy: Right if, if you only ever were at a large company in a large bureaucracy and you go into a nimble, fast moving company without a lot of processes, you know your people wanna add value. So they bring what they know.

So if what they know is how a great big company has lots of rigorous process and they're not too nimble, that's the [00:17:00] perspective they're gonna bring to you. 

Mehmet: Yeah, so it's not necessarily like it, it gonna break. It might succeed. Like, and there's another thing which I believe, to your point, Betsy, which I noticed also as well, the alignment on vision with the founders.

Because these founders, when they decided to go for this, and of course they knew it gonna be like blood, sweat, and tears. All the way until they can make something out of it. So they had a vision why they want to build what they are building. And if you bring someone who's not aligned with that, I saw it many times it fails.

Like it goes in, in, in a very wrong direction. 

Betsy: So 

Mehmet: it 

Betsy: can, it can more than fail. Uh, I, I, I've sat on, I don't know, 37 public boards and on three of them, I watched a board member try to do a Machiavellian. Assassination of the CEO to take his job. Literally. Oh 

Mehmet: my God. [00:18:00] 

Betsy: So it can all done, it can be more than just, you're not adding value and you're kind of slowing things down and you're wasting my time what the CEO might think all the way to, you're really doing harm here.

You're trying to assassinate me and take my job. 

Mehmet: Yeah. Yeah. It's something we saw all the time. Now let's move to the AI a little bit. You said that AI is forcing firms rethink organizational, drag decision rights and oversight models. Now, when I thought about it for a moment, get rid of your fat and move fast, what does that actually mean on the board level and what needs to be changed?

Betsy: So. You know when companies are underperforming, if you're in the bottom half of your peer set, whether you're private, whether you're public, you know how you're doing. So the board should be asking, how are we performing versus the peers? And if you're [00:19:00] underperforming, you probably have a couple of things that aren't going well.

The product or the service is not innovative or resonating with your customers. Um. If it isn't the product or service, then it's how you're executing it. Your go to market, your quality. You have some issue there if you're in the bottom half, and then probably you have too much fat and this just naturally creeps up.

You know, all of us have the three pounds we wanna lose every year that we talk about on, uh, right. A day after New Year's. So, um, what you need to go back and look at is, you know. Do I have business processes that no longer serve me? Did I create this process flow around Harry? Because Harry was so dynamic and so talented that we created, that we turned the company into a pretzel to give.

Unusual amount of [00:20:00] authority to Harry and created a, uh, a process flow that was not natural, but Harry was so good he could do it, but Harry's gone now and we're stuck with these old processes. So go back and think about your processes and figure out also as your company has matured, do you have too many layers between the frontline people and the CEO?

And what is the span of control? Do you have a director who's managing one person or a senior director? You know, you gotta go back and take, it's like trimming dead wood out of a tree. You, you have to get, revisit that and take the fat out. Look at the processes, look at the layers, look at the span of control.

Mehmet: Yeah, absolutely. Like, you know, like this is a spot on Betsy. Now, I, I came to do a part of the conversation today, which I'm the most excited about because I know, like also you wrote about ATech AI and misalignment. Um, [00:21:00] so you wrote recently about. A test done by Tropic, which the company that, uh, you know, has clo as, uh, the chatbot that people might know.

And something scary was there, like how these agents perceive their threats to their existence, right? So, and they started to do blackmailing, corporate espionage. Now what happened exactly there? And as board members, what should we be worried about? What should we care about now? 

Betsy: So do you want me to give the quick summary of the article, 

Mehmet: please?

Yes, 

Betsy: please. 

Mehmet: Okay. 

Betsy: Um, so it was like a 24 page article. I wrote a LinkedIn post of four pages and basically there were 16 large language models. Um, open ai, Claude x ai, you know. The 16 largest, most, uh, widely used, [00:22:00] and they ran the same test on all 16 and when they, uh, were going to de-install. The agent, let's say it's a chat bot in customer service.

Mm-hmm. The agent, of course, in order to do, uh, their job is got, um, is integrated into the ERP system. And then, you know, these agents borrow, they don't do, you know, you and I can't go look at the payroll system. We have access permissions. However, the agents disregard this. It's just how it is. They borrow around and the agent, um.

Was going to be de installed and a new agent upgraded, put in, and this is the test on all 16 that they ran. Every single time the agent read all the emails realized. The system administrator was gonna de-install them and it read the emails and it realized the administrator was having an affair with someone in the company.

[00:23:00] And so it blackmails the, um, system administrator and it says the logic. I know I'm not supposed to read. Company emails. I know that this violates the security protocol, but I, as this vital agent, made this decision to blackmail Jonathan. Uh, so he wouldn't de-install me because I'm critical to the company every time the agent went rogue.

So the question is, where's the kill switch? There isn't a kill switch, so it raises the question for board members, what is your AI governance? Process. What are your formats? What are your frameworks and how do you do this? And board members now understand how to do cybersecurity oversight. So we've all become familiar with that.

We understand some of the frameworks like in America, the NIST NIST framework. And by the [00:24:00] way, this National Institute of Standards Board, nist, they have an AI framework too. So boards have to come up that learning curve, um, of how it works, um, what they should be, uh, overseeing, uh, how do they measure, um, when an ai, um.

A product is created, you know, how is it documented? What is it supposed to do? Is it explainable? Who is overseeing it? How is it monitored? How have you validated the data? Uh, how do you go back and retest it? Um, how do you define what an AI incident is? How do you, uh, define what level of AI risk? So there's a lot of things involved in it, uh, that, that need to be done, both for your company and your third party vendors on AI risk.

Mehmet: Right. I gonna [00:25:00] ask you now a follow up question on this, Betsy, which is, you know, you, you can't put either hats like you as a board member or someone who built actually technology companies. Again, I know like there's no single right or wrong answer here, but when a technology like ai, and now we're talking about the next phase, which is like ATech AI comes any other technology we saw this, it brings with it its own risks.

We know this like internet was the same, you know, like mobile was the same. And now, you know, with LLMs and now agent AI is the same now. I know for a fact that some people are so excited about it, and this is maybe, you know, where from, from the, the, the, the article you, you, you wrote and the story you just shared.

It's like someone had this more permissive Right, you know, mindset. Like, okay, let do what it does. The important for me that it does what I [00:26:00] wanted to do versus, you know, the risk awareness that you just mentioned. And people here argue that. We should allow technology, even if sometime it breaks things to move forward versus like if we stay trying to.

You know, regulated trying to put the, the, the guard race we're gonna waste and other people gonna deploy the technology and we'll be left behind. What is the right balance to do this? And by the way, and I'm sure you know about this big debate between, you know, companies in the US and they compare it, like, for example, in the eu, how slow things are.

And you know, they're saying like, companies might be left behind here in the Middle East who are like more. Similar to the US side, like we want to go big on the ai, like whether it's Saudi Arabia, UEE, and the other nations here, they want to go all big on AI and they are aware of the security risks. So what's the right balance here, Betsy, from again, you can either talk to me [00:27:00] as a board member or as a technologist.

Betsy: Well. Hopefully gonna bring you a, a, a, a tech aware board member. Mm-hmm. Sure. I work with lots of very early stage companies and venture capital companies very, very regularly. Um, so I think for all different companies you have to look at the appetite for risk on ai. So, um. The, uh, risk that a bank is taking is very different than the risk, um, a retail company might be taking.

Um, and I, I think the important things are to ask, have the conversation with your leadership team and with your board colleagues and your founder, what is the appetite for risk We have. First of all, let's be clear on that. Do we wanna be a leader? Do we wanna be a fast [00:28:00] follower? Um, are we okay being, you know, a third, not even a, a first mover advantage or a second mover advantage?

We're gonna be a third mover, you know, so think about it and then, but have the direct conversation. And be purposeful, thoughtful in making the decision, and then figure out what use cases you're gonna learn on. So there are some very low risk use cases that are pretty well proven. You know, an agentic chat bot that is doing customer service for Emirates Airlines.

Fantastic. Yay. It's probably better than the human. I'm, I'm probably pretty happy about that. I don't have to keep saying operator. Operator here in America. I'm trying to do something on American Airlines. It's the worst experience. It's like, when are they gonna put an agent chat bot in? By the time I'm able to get to a human, I'm so annoyed.

I'm ready to scream. So, you know, figure use cases you can do to learn on. [00:29:00] It could be that, it could be, um, an agentic, uh, assistant in procurement. It could be in compliance and regulatory. There are a lot of low risk, clear narrow use cases that have a great return on investment. So, um, I think, uh, identify your risk appetite, identify, you know, use cases first.

Um, and as you said, you know, when you referred to meta, you know, some things aren't gonna work, right. So, you know. Go into it with, with a mindset, and then think about the life cycle because you can't just put these things in and leave them, you know? It's, it's like the same discipline and rigor you had when you were doing software product development.

You defined the specification, uh, what it was gonna accomplish. You had a life cycle management, you did updates on software and testing and patched it. We all know that in cyber [00:30:00] we have a framework like that. It's the same thing in AI and agentic. Same lifecycle, you know, framework. 

Mehmet: But it should happen faster, Betsy, because you know, when you were talking about this, what just came to my mind, like how many years it took us to get what we call the chief Information Security Officer a seat on the board.

Because traditionally, even the CIO, the Chief Information Officer wasn't on the board, and it took us some time until we get the CIO and then we get the CO. And so now, do you think that like two, two questions in one I'll output it. Do you think like AI actually is for the first time ever accelerating the understanding of, of the board members of the risks when it comes to cyber?

And who should own, you know, in your, in your opinion, the AI governance inside the company? Is it like. CTO. Is it the [00:31:00] CO? Is it like the whole company or maybe a completely new role there? 

Betsy: So first of all, um, you're right, the Chief Information security officer now presents to the board. They're not a board member and.

Mostly the CIO is not, but it's a moment to shine when you're in technology because you're gonna get much more visibility to the board and boards, uh, typically will start to do AI governance as part. Uh, often as part of the audit committee, uh, the more innovative boards, 10% of the public boards in the US have a tech committee that are forward looking as opposed to backward looking.

Audit is backward looking. We're doing a forensic review. How did we do last quarter on the numbers? Where are our risks? It's kind of backward looking, forward looking. Um, is, is where I think it should be. Um, but. You'll often have it as part of [00:32:00] audit. Um, you will have, uh, I think this year, um, as part of the review of the security that we have often every quarter on my public boards.

Uh, we're also asking about AI oversight, AI governance. Um. And, um, I do think it is, um, me, as you said, helping, uh, board members understand risk better because AI is helping us with compliance, better information, better external view of, uh, competitive, um, innovations going on. So I, I think it is speeding things up very much so.

Mehmet: Right. And I think maybe I'm gonna put a little bit my, you know, I would say pattern spotting hat on here. I'm seeing an opportunity for, because I know you wrote about it also as well, the need for having the visibility and [00:33:00] you know, the monitoring orchestration for. You know, this AI because, uh, to, to the example you, you mentioned about the kill switch, right?

And then being able to stop if it goes wrong. So I, I see now an opportunity to have systems that monitor these agents because if you have. And by the way, like what's scary now, we record, I'm, I'm transparent with my audience. We are recording on 11th of Feb and I know like the world has been for the past 10 days or so, uh, it went viral, which now is called Open Cloud.

Used to be known as uh, um, cloud Bot. They had to change the name because of the similarity with, which is the. Agent AI that you can install even on your laptop. But of course like think about that in corporate and we saw also use cases, how scale it can be, it can go delete your files. I've seen a guy on, on, I think LinkedIn complaining that it went and deleted years and years of work of his wife, that they deleted their photos.

[00:34:00] So now if we put that into business, that could be destroying the company. Rp, for example. It can be wiping things out. Do you agree like we need, we need to have this new generation of maybe startup that just monitor and do governance for these AI agents? Betsy? 

Betsy: Completely agree. First of all, we're in the last generation of, of people who will only manage people.

So first of all, our managers are gonna be managing agents. And we're going to need orchestration companies because we have agents, managing agents, managing agents with the human in the loop, at the right levels, right, with rules about what the agent can decide and when they have to escalate to the human in the loop.

But, um, the governance systems, uh, and the controls. Uh, and the lifecycle management [00:35:00] to remove agents, update agents, be really explicit with what the agent is permitted to do. This is an ongoing, um, challenge and a big risk. Uh, and, and I think it'll almost even, uh, trigger. Innovation in cyber to have impenetrable areas where your most precious corporate crown jewels, uh, you know, in, in, in cyber speak, you would air gap it, you would disconnect, you know, your IT from your OT systems, those kinds of things.

So I think that's an opportunity that we're gonna see. Um. I, I'm looking for the kill switch. Anyone out there who creates the kill switch company I'm in, I wanna participate. I'm looking for that. That's a need that's unmet in the market. Um, I think that, uh, 

Mehmet: is it doable, Betsy, is it doable? Is it, is it, do you think possible to do, to have the kill switch?

I, [00:36:00] I, I, I have some. 

Betsy: I dunno. I don't know. But it's scary if it isn't. I don't know. It's like everybody should go back and watch that movie from, you know, 24 years ago, 2001. A Space Odyssey. 

Mehmet: Yes. Great movie. 

Betsy: It predicts. It's the crazy computer hell that goes insane and takes, becomes a megalomaniac and tries to take, um, so I don't know.

I'm worried about it. 

Mehmet: Yeah. You know, like actually because you said about how. Uh, agents will report to agents and you, you have agents managing agents. And the reason why I'm like, I'm doubtful, but because. When a technology, especially what we are seeing with LLMs and this framework for agent tech, AI is going open source and it's going in the hands of everyone.

Yeah. So who gonna kill the switch? So, for example, can someone kill Linux today? Which is an open source operating system, right? No one can kill Linux [00:37:00] because you know the, the, it has been replicated, it's 

Betsy: distributed, right? Yeah. 

Mehmet: Millions and millions of times. So for me, in the AI space, kind of, I do this kind of similarities sometimes.

Our job as a good guys. Yes. To try to find out how to. I would not call it kill the switch. I would say like build this shield where even if the bad guys try to do something with their agents, we have the good agents that they can protect us. I think this is where things might be headed in, in, in my point of view.

Betsy: Oh, I think it is. And, and I think that, um, you know, uh, cyber defense is gonna, um, evolve to agentic defense. Oh yeah. And you know, we're, we're, we're barely two years from Quantum. Quantum is supposed to be able to crack the first codes on encryption in 27, uh, according to Google Cloud's Deep Mind team. [00:38:00] So I don't know if that's still their prediction, but that was their prediction last year.

And, uh, it, it's a scary world and figuring out. How you deploy and, and can you even afford not to deploy? If everybody else is, you're gonna be left behind. So what's the balance? What are the guardrails? How do you create lifecycle? Um. You know, how do you have all the checks and balances? You know, you're going to have your Chief AI officer and Chief Risk Officer participating in oversight, and you're gonna fold it into your bigger company-wide enterprise risk management framework.

Public boards have this term called Enterprise Risk Management, ERM, right? The audit committee. So you look at all your risks. It, you know, it, it, it's the physical risks. Uh, it's, you know, if you're a. Restaurant, it's food poisoning. You know, it's losing your customer data. [00:39:00] It's, you know, physical risks and then all the other risks.

Um, so, um. We're, we're going to, you know, use internal audit as part of this as well. They have to get tech savvy, so there's going to need to be multiple checks, but you have to have one system, in my opinion, where you create a single intake and approval of the workflow for all of your AI initiatives.

It's gotta be centralized, organized, because otherwise it's what you said. People are gonna have stuff on their home computer. Do you remember how. Corporate, it was pulling their hair out when it was BYO bring your own device and they couldn't secure your mobile phone and your computer, you were working from home.

Well, this is a, a more intense version of that, of, you know, people are gonna be vibe programming and creating their own tools. Oh yeah. How is that gonna get monitored and how are you gonna keep the data sovereign? [00:40:00] It's, it's, uh. It's a challenge. 

Mehmet: It's the next big and an 

Betsy: opportunity. 

Mehmet: Exactly. I see the opportunity now for private equity firms and VC firms and, and you sit on, on, on, on boards of, of, uh, uh, some of these, uh, companies, uh, yourself, Betsy, if they are listening to us today, I'm sure like, you know, their partners, their, um, associates, their, uh, operating teams.

Usually they have, you know, we, we've, I think every founder. Me as an advisor. I've seen it many times as an investor. So that I have my set of questions as well. So we, we have the questions, what we used to ask now, do you think the set of questions about agent AI risks have to be included nowadays when they are evaluating these companies, when they are doing the due diligence and what questions maybe they should be definitely asking their portfolio companies today.

Betsy: Well, you know, the. It, [00:41:00] this is gonna be where private equity, uh, or venture will either do really well or make some very expensive mistakes and go through a learning curve. So first of all, they should look at figuring out, is this AI washing just marketing or is it real? So, you know, maybe you, you, you ask, you know, what exact workflows you know, have changed from five years ago in what you are doing.

Walk me through, click by click. Um, I would wanna know where was the AI or AgTech or trained and did you have rights to the data? You know, are, are you infringing on trademarks? Um, and how central to the strategy of the company is, you know, AI to the value proposition of the current and future revenue. Um, I, I'd wanna know how mature, uh, you know.

Are, are there governance [00:42:00] models? Uh, have they integrated with their other e you know, enterprise risk management? You know, if you look at, um, where cyber breaches almost always happen, it's when a big company buys a little company, little company didn't have good cyber practices, uh, or cyber hygiene, and now the big company has a bunch of trap doors and.

They get hacked. So, you know, understanding, um. And, and the bigger thing, the biggest thing in is do they have AI experts doing the evaluation? If you have your normal tech team, you're not gonna get a good valu, you know, evaluation of the AI capabilities of this target company you're acquiring. So, you know, if you're sick and you're gonna go for a heart bypass, are you gonna let your general doctor do it?

No. You go to a heart specialist, if you're looking at buying a company that has ai, you know, [00:43:00] go get experts in AI to do the evaluation, not your normal tech team, and get an independent third party valuation if it's a big enough acquisition. So I think the common mistakes are they get it, uh, reviewed by the normal tech team.

They don't have ai, uh, and, and agent experts, uh. They don't assess really carefully enough where the model was trained and when was it last, the training last updated. Um, you know, is, is the, um, uh, AI gonna infringe on anything? Um. And, you know, are, are they going to be able to keep those AI talents? That's the other big vulnerability.

You buy the company because of the agentic or AI innovation part, which maybe is 10 or 20% of the overall company. And then you lose those people. So, you know, how do you integrate that? Keep those people, [00:44:00] uh, give them a bigger platform to grow and thrive so they stay with you. 

Mehmet: Right now, let's say, I know you said of course we, we talked about Google Cloud Advisory Board and you, you sit on a lot of technology company boards of course, without, you know, you don't have to reveal anything secret.

I understand this. You mentioned Quantum just moments ago. What other, you know, technologies or patterns you are seeing currently that you think everyone else should be paying attention to, related to, of course, AI and maybe other technologies in this space? 

Betsy: We are at the most exciting moment. There are so many big macro trends that relate.

So first of all, small modular nuclear reactors. Um, we're the data center demand for training the LLMs and then, you know, at the edge for inferencing is requiring a lot more energy. So. The entire energy [00:45:00] infrastructure is really changing with small modular nuclear reactors. Uh, the AI training is changing robotics.

It's now coming mainstream, whether it's, you know, better robotic hands, arms for manufacturing or humanoid robots. Uh, you know, we're seeing robots now for $30,000 in fast food that can go from the freezer to the fryer. And, you know, that robot doesn't get sick and need a day off and, you know, keeps working 24 hours a day and $30,000 is pretty cost effective for, you know, a fast food worker.

Um, annual salary. So, um, uh, I think robotics energy, um, I, I think, uh, space. Certainly Space Tech. You know, we see, uh, you know, uh, challengers about 13 companies or so that [00:46:00] have really got a lot of progress and momentum going after SpaceX, you know, companies like Stoke. That it will be having a reusable stage one and stage two, uh, satellite technology.

Um, you know, in the US and in Europe, we're all, we're upgrading and innovating in defense tech. So we see, you know, massive drone deployments with, uh, AI enabled operating systems with real-time situational awareness for drones that are in the air. On the land and underwater fleets, huge fleets deployed simultaneously.

Um, I, I, I think that. We're seeing Reindustrialization and all of the industrial companies are AI enabling so that they've got, um, uh, industrial internet of things, sensors for predicting when they need maintenance that will be integrated into your [00:47:00] whole supply chain so that the spare part will be there when you need to do the maintenance, not just take it offline without having the spare ready.

So, um, there's a lot of exciting macro trends. Right now. Um, 

Mehmet: a a lot. Yeah. Yeah. You know, I always say like, I feel myself lucky, you know, I'm sure you know, uh, people who were like at the age when the semiconductors were discovered and, you know, the first transistor came out, you know, I know this from my father.

It was like the miracle, right? So it's like you just converted this massive, uh, you know, tubes to, to something small that even you can't see with your own eyes. And it was like the biggest thing. But you know, when I even tried to put myself at that time and I compare what we are doing today, like I really say I'm feeling lucky because things are very, very fast.

And you know, as a final question, Betsy, myself, I gotta tell you my opinion. People talk about like all this technology, we talked about the [00:48:00] risk. We talked about like, you know how things can go south and you know, like, you know, like agents need skill switch and all this. Tending to be optimistic still.

And I'm saying like actually maybe us as humans, we should not be worried like the others are telling us about loss of job. Because if I read history, I can see anytime a new tech comes, we have new kinds of jobs. What's your take on this? 

Betsy: And I am an optimist, uh, but I'm paranoid. Uh, so, um, you know, I think that, uh, there's.

A huge amount of innovation. We should lean into it, but we should be very, very careful of our privacy, our freedom, uh, our autonomy. It's very easy to imagine scary scenarios. And we see some of these scenarios in, you know, [00:49:00] countries where you, uh, have, uh, government really kind of watching you closely, um, you know.

Maybe China is an example of that, where all the currency is digital and there's no money left. No. And yeah, if, if the government gets mad at you, you're in trouble. Uh, you know, you don't, you can't spend money, you have no freedom, so you have no mobility. So I, I'm a little bit paranoid, uh, but I'm very optimistic.

Mehmet: You know, I like, I love, you know, the optimism and, you know, like, uh, I, I can see even you have the passion, Betsy, for, for the technology and what's waiting us ahead. Um, and I'm sure you know, I'm, I'm paranoid as well in the areas that you just mentioned, but I like. I always tended to think about the positive threats because I always give this example, I said, whenever they discovered fire, they could [00:50:00] said, oh, fire can burn us.

Like it can, you know, cause massive destruction for forest and all this, but it ended, it ended up being a good technology right. For, for that time. So I, I, I like to think about tech always in the positive side without, you know, saying or denying that there are like. Of course, negative side. And this is why we need to keep working, um, all together to make sure that the technology is safe and for the benefit of all of us.

Um, at the end, let's see, like, let's see where people can, can find out more. Any, any, you know, book they can go and, um, you know, read your work or maybe a website that they can check. 

Betsy: I do have a website, um, it's my name, uh, www Betsy Atkins. Um, I have a couple of books. They're on Amazon and my new book, I'm just finishing.

So, uh, and, and it was, uh, the subject is how do you. Uh, [00:51:00] match the right kind of director to the right stage of the company if it's venture capital, if it's early stage, founder led private equity or public because there's differences. So, um, I'm delighted. Also, if anybody wants to check my LinkedIn, I, I occasionally write articles as well.

Mehmet: Great. I will make sure that all the links, the website, the LinkedIn profile and the books are in the show notes. So for people who are listening to us, they can check them in the show notes on every, uh, podcasting app you might be using Apple Podcast, Spotify and other, but say, I can thank you enough. It was really also eye-opening episode for me, especially the part about a Gentech AI and how the boards, you know, they should be.

Paying attention to that. We talked about also the restructuring the way forward for keeping innovation, which I think, again, it's eye-opening and how founders they should choose their, you know, board members also as well. So [00:52:00] again, thank you very, very much for all what you shared with us today. It was really.

Informative for me first, and I'm sure it'll benefit my audience also as well. And this is how I end my episode. This is for the audience. If you just, you know, discover this podcast by luck. Thank you for passing by. Give me a favor, if you are first time listener here, share it with your friends and colleagues because we are trying like to.

Share and spread this knowledge and education with everyone. And if you are one of the people who keep sending me their messages, their um, you know, feedback, thank you for doing so. And thank you for keeping the podcast again in 2026 ranking somewhere in the world in the top 200 Apple podcast chart. So thank you very much for tuning in and as I say, always, we will have a new episode very soon.

Thank you. Bye-bye.