EPISODE 190
Augmented with AI (REPLAY)

EP 190: Augmented with AI (REPLAY)

Our bi-weekly Inflection Point bulletin will help you keep up with the fast-paced evolution of cyber risk management.

Sign Up Now!

About this episode

August 12, 2025

How should individuals be thinking about generative artificial intelligence at work and at home? Let’s find out with our guest Daniel Miessler, whose mission is “Working towards Human 3.0 so we can survive and thrive as humans after AI”. Your hosts Kip Boyle, CISO with Cyber Risk Opportunities, and Jake Bernstein, Partner with K&L Gates.

Website: https://danielmiessler.com/
LinkedIn: https://www.linkedin.com/in/danielmiessler/
X: https://x.com/DanielMiessler
Fabric: https://github.com/danielmiessler/fabric

 

Tags:

Episode Transcript

Kip Boyle: Hi. This is Kip Boyle. Jake and I talk about cyber resilience a lot on this podcast. You have to be ready for any kind of cyber-attack or cyber failure, and that goes for us too. We made a really great episode for you, but our systems got all messed up. And the audio, unfortunately, is completely unusable. So we went into the archives and found one of your favorite episodes. It's with our guest, Daniel Miesler, who talked with us about how individuals should be thinking about generative artificial intelligence both at work and at home. So I hope you'll enjoy this episode again, or maybe it's the first time you've heard it.

Speaker 1: Welcome to the Cyber Risk Management Podcast. Our mission is to help executives thrive as cyber risk managers. Your hosts are Kip Boyle, Virtual Chief Information Security Officer at Cyber Risk Opportunities and Jake Bernstein, partner at the law firm of K&L Gates. Visit them at cr-map.com and klgates.com.

Jake Bernstein: So Kip, what are we going to talk about today on episode 170 of the Cyber Risk Management Podcast?

Kip Boyle: This is going to be a very good episode. We've got a guest. We've got an excellent topic. We're going to look at the practical implications of artificial intelligence for individuals. There's so much conversation about what does artificial intelligence mean to humanity, but let's bring it way, way, way down. Let's make it hyper local and let's talk about how it's affecting people. And we're going to focus particularly on generative, so that would be OpenAI's ChatGPT, Anthropic's Claude. And we're going to do this with a guest. His name is Daniel Miessler and his mission is working towards Human 3.0 so we can survive and thrive as humans after AI. Daniel, welcome to our podcast. Thanks for being our guest. We would love for you to introduce yourself.

Daniel Miessler: Yeah. Yeah. Thanks for having me. So I've been in security for around 25 years, and at the end of 2022, I pivoted over. I'm still doing security stuff. I don't consider AI to be a thing itself. I see it as an augmentation of a thing. But I'm basically building AI now, applications and infrastructure for upgrading humans. So ironically enough, we could be more resilient against being replaced by AI.

Kip Boyle: Yeah. I love that vision of yours. Daniel, I knew about you. For many years ago you were a prolific blogger on the topic of cybersecurity, and so I would periodically read what you wrote, and then I saw you pivot over into being way more AI focused. And so I think we could also add entrepreneur to a hat that you wear. Because if I understand correctly, you now run a small business and generate all your own revenue. You're no longer a W2 employee. Is that right?

Daniel Miessler: Yeah. That's right. Yeah. At the end of '22, I went independent, had my last regular nine to five job, and yeah, now unsupervised learning is basically like the umbrella. It's also the name of the newsletter and podcasts and everything, but more so it's just the umbrella for everything I do, which is mostly building applications.

Kip Boyle: So you're building applications. You also have a membership. I've purchased a membership. I'm in there. I think it's very good. You've published software. I'm using some of it right now. And I want to also thank you for your service. I noticed as I was looking at your LinkedIn profile that you served in the US Army, you were an infantryman and an intelligence analyst. So thank you for your service.

Daniel Miessler: Yeah. Very welcome.

Jake Bernstein: I'm curious, I'm at a little bit of a advantage here because there's a whole bunch of things I want to go and look at, but frankly, my computer's going to be too slow to let me do that live. I'll check out the blog and this community. But I am curious also how we can link individual use of artificial intelligence to a security mindset and really a privacy protective mindset as well. And it strikes me that all of these concepts are really intrinsically linked together. If we want people to be resilient, we need them to understand how to use AI tools in secure and safe and protective fashion. I hope that's something that you both agree with. I'm curious to hear what you think about that. If that's going to be part of this conversation.

Daniel Miessler: Yeah. That's where I was hoping to take it actually, because the way that I'm thinking about AI is essentially optimization of goals. It's basically building methodologies for achieving goals, and that's why AI is not a thing. It's an optimization of a thing. So we could talk about the individual side, but it's the Cyber Risk Management Podcast so one of the things that I'm speaking on the most this year ... And I'm doing a whole bunch of talks, I'm going to do this at SANS next week. I am going to Switzerland to do this. Riyadh. So it's all about using the same method that I use for individuals, but doing it for security programs to manage risk and basically communicate with the C team and everything.

Jake Bernstein: Very interesting. So let's maybe talk a little bit about some of the AI applications and infrastructure that you have built. Kip mentioned that you've published some software and you write applications. Just give our listeners and me some idea of what that is.

Daniel Miessler: Yeah. So the first one that most people know about is called Fabric. It's an open source crowdsourced project. And essentially it's augmenting humans with AI is the tagline. But what it is really is a library of crowdsourced AI prompts. So think of it this way. Imagine that you have a problem and you talk to some chatbot to try to get it to help you solve that problem, but now you've got a really good way of doing it. You've figured it out .you figured out a really good prompt that anytime you apply it to that particular problem, you get a really good answer. What Fabric does is it takes instances of that and puts it in a category like really well named and allows other people to either improve that or fork it or make a copy for another thing. So imagine you don't have to reinvent good prompting every time and talk to some chatbot. Instead, what you have is you have a series of problems and you have a series of solutions to those problems, and you simply apply them whenever you need them. So it's like this giant library of solutions from AI that you can apply hopefully to a giant library of problems.

Jake Bernstein: That's so critical because prompting as everyone who has spent any time at all with the modern chatbots knows is super critical. At the law firm that I work at, we use a program called CoCounsel. It's all about the prompting and we are maintaining our own little internal database of prompts so that people can learn how to use it. And it's not always intuitive how to do it and what's going to make the best response. So that's excellent. And Kip, I assume that you have made use of Fabric?

Kip Boyle: Oh, yeah. Absolutely. Daniel put on a course in January this year and an introduced Fabric as well as his mission and the idea of augmenting humans with AI as opposed to a vision of replacing people with AI. So I was very attracted to that. Another thing that I like about Fabric is it also gives me a command line interface that I can use to tell Fabric what to do. So yes, it's got this fabulous catalog of prompts. And I can add prompts, and I have added prompts for my own use, but I can also use Fabric as a command line tool on my Mac. And there's some auxiliary tools. And the big problem that I've been having lately is that I've been using Fabric for is there's all this content out there on YouTube that I want to consume, but I don't have enough time to sit there and consume it.

And so what I can do with Fabric is I can have it go out to YouTube, retrieve the transcript of the episode, and then it can tell me ... It can rate it, it can tell me if it's how much quality is in there based on different factors, and then it will assign a score and it'll tell me whether I should consume it immediately, whether I should just put it on a list to look at someday. I can also tell Fabric to retrieve the transcript and then do something called Extract Wisdom, which was one of the first things I ever saw Daniel do, which blew my mind. So it actually is simulating me watching the video and taking hand notes as I watch the video, but I don't have to do it. It does it for me and it does it in seconds. And based on what it comes back with, I might decide, okay, this is great. This is what I needed to know. I don't need to watch the video now. Or it may come back and I'll be like, "Oh my gosh, this is super insightful. I absolutely now have to watch this video." And it just keeps me from wasting my time watching videos that ultimately don't impart enough value to justify me watching them. So it's fabulous.

Jake Bernstein: That's a really good example of I think augmentation. And on that point, one of the things I want to just mention, and maybe we can talk about it. And Daniel, I feel unusually unready to talk to you because I am realizing that you're famous on this internet, and I have apparently been living under a rock when it comes to personal use of AI because they've been so focused on the legal thing. But to that point, I wanted to just ... If you'll recall, Gartner maybe a year ago or so had made this big splashy in the media splash saying that in our estimation, 40% of legal jobs are going to be replaced by AI. They've pulled back from that. It's not true. But what has changed is ... What we tell lawyers now is, look, you're not going to lose your job to AI, but you will lose your job to a human, another lawyer who knows how to and does use AI.

And I think that is really maybe the core essential component of this augmented Human 3.0. It's not going to be sufficient to just be a person, a human. It almost isn't now, but within five years, it's going to be impossible to be an effective member of the economy unless you are augmented with AI. So maybe now is a good time for you to build off that and answer the question, why should people be using generative AI to augment their ability to perform work? And work in this case doesn't mean just your professional job, it could mean your personal life, but really, I think anything,

Daniel Miessler: Yeah. I think one way to do that, to answer that is to push back a little bit on what you were saying about the AI augmentation. I agree that it'll start with humans augmented with AI, that'll be who you're competing with. But I did a piece called ... Something like We're Getting It Wrong, or something like that around AI. I can't remember the name of it. But it's actually a bunch of workflows for regular work. It's like we're thinking about AI all wrong. I don't know if somebody wants to pull it up. But it's a series of graph workflows of what a person does in their day-to-day job.

So one of the examples I gave was, say, an analyst who approves or disapproves insurance claims inside of an insurance company. What they do is a number of steps where they're like, "Okay. I've got to figure out is this a legitimate thing? Does this person actually have a policy with us?" I need to review the details of their policy. I need to review all the evidence, then I need to detect if this is fraud or not. And then I need to make a decision of are we going to pay out? If so, how much are we going to pay out? Now, that is a manual process right now. Largely. There are definitely ... Yeah, that's right. That's right. Claim right. Do you mind me going a little bit scary, especially around the legal field?

Jake Bernstein: No. You should go scary. And actually I already know where you're going to some level, and I want to amend my statement, but you go first and I'll be extra convinced when I amend my statement.

Daniel Miessler: Okay. So I want to give an example of that thing that you just showed, that graph there.

Jake Bernstein: For those listening, since we don't publish video, I just showed a picture of ... The title of the blog post is we've been thinking about a AI all wrong from July 29, 2024, so relatively recent as of the day of recording if you want to go find it yourself. And we can put a link in the show notes, right, Kip?

Kip Boyle: Yep.

Daniel Miessler: Yeah. So basically what that image that you showed has is this series of steps that this human does for work. And my point about this is that AI, its main function inside a business is replicating those intelligence tasks. So that task of taking the context of the policy, taking the evidence and pictures and everything sent in by the person, determining if this should be a payout and how much it is exactly the same as that diagram that you just showed. Now, let me tell you how little and how much I know about the legal profession. One thing I know and I'm confident about, but I'm not an expert here, I remember old movies. I don't know if it's '80s or '90s, but there's a whole bunch of old lawyer movies mostly focused on trial lawyers. So it's like two competing lawyer teams fighting over this big thing. A huge part of lawyer work at that level in that game is finding details. So what these movies would have in the '80s and '90s is coffee, pizza, really stressed out legal teams and rooms full of boxes. And they would not sleep. The boxes are open, the lids are all over the place, and you see them stressing, and this is the stress part of the movie. They're like, "I don't know. We just can't find anything. We're screwed. They're going to beat us. They're going to beat us."

Jake Bernstein: Then someone finds the magic smoking gun and everything is ... And then there's always drama. So that's the discovery process, the civil discovery process. Over the last 50 years, it's been responsible for the vast majority of the increase in litigation costs. And curiously enough, my firm, back when it was Preston Gates & Ellis was one of the first to develop some of the original e-discovery tools, which then spun off into a separate company, which has gone through multiple iterations. But now more and more of those tools are starting to use foundational AI models. And we are very close to a future where the billable hours ... When I talk about the legal professions business model, those billable hours spent around that table with the boxes of documents are gone. They're gone. And what's super interesting about this is that we could look at that and say, "Well, damn, there goes all those hours. Our business model is hosed."

But on the other hand, you also have to realize that now a solo practitioner who might be a plaintiff's lawyer because they don't have all the conflicts ... And now I'm getting way too legal geek for this podcast. But I'm sure some of you know what I mean on that. And that individual solo practitioner now has the power of AI. So even though they could never afford to pay associates, most of those people in that room are. They can go pay a small fee to use generative AI to review documents, and that's going to lead ultimately to more work, more productivity. And I think that is where things are headed. But yes. I would say that the legal profession has been acutely aware of these replacements and this technology coming down the line.

Daniel Miessler: So if you take that further on, if you take basically all the different ways ... If you take the top legal firms or the average legal firm or whatever, how do they make money? So it's like creating contracts, analyzing contracts, looking for things that could hurt the customer or whatever. Revisions, recommending revisions. And then you have the trial lawyer part, which we talked about. All of those turn into Fabric patterns. They are use cases which produce outcomes which are useful to customers. So essentially what you have is a game where you need to get the context into the system and feed it to the AI. And the AI uses something like a pattern which says, find jaw-dropping detail that could massively help us in this case or whatever. And boom, it's able to pull out those nuggets and give you ... And even it could look at all your previous arguments in cases where you won and say, "Write me an argument exactly like that using the nuggets that you found from these 900 pages of content. Plus I took these 700 images of the crime scene. Analyze those as well. Give me new ideas for new research to do." So all of those just become Fabric patterns. You're not writing any AI, you're just using the AI that you already have. The patterns are already written. Find additional evidence, write the arguments, all these different things. So basically have this infrastructure.

Jake Bernstein: When you say Fabric patterns, that sounds to me like a special term. Maybe just to quickly explain what a Fabric pattern ... What you mean by Fabric pattern.

Daniel Miessler: They're just prompts. They're just locked in really, really good prompts.

Jake Bernstein: So those are the prompts that are stored in the Fabric database or GitHub repository?

Kip Boyle: Yeah. It's in the GitHub repository. And then when you install Fabric, you download copies of everything off of GitHub, and every now and then you should update it and so on. And you can write your own. But yeah, Daniel has ... He called it Fabric. And then the prompts are called patterns, and then there's a server called a mill. And so there's ... I don't know what you call it, an analogy I guess going on here.

Daniel Miessler: Yeah. So the reason for that is because on the command line, so there's another term called stitching, which is connecting things together. So I'll give you an example. You could say, "Take these 900 pages of evidence, send it to a Fabric pattern called find gotchas." Okay. That's the name of it. Find gotchas.

Kip Boyle: You'd have to write it.

Daniel Miessler: Yeah. You have to write it. You put it online, or you keep it private. Find gotchas. So that produces like 37 different gotchas. And then you pipe that one, the output of that goes into write argument. So now you've executed two patterns stacked together, and it took the content, the nuggets that were discovered and wrote the argument based on them. You can do this as many times as you want.

Jake Bernstein: And then at that point, we have a lawyer in a box, which is I think what a lot of these things are coming to do. And like I said, I don't want to focus this on ... We could make this entire episode on the legal industry's use of AI, but let's not.

Kip Boyle: And actually this is a great pivot. So let's pivot over to Daniel, what you talked about in the beginning, which is taking this idea that we've now been talking about applying an illegal context, but you said that you're about to give a series of talks where you're going to say, how can you do this to manage a cybersecurity program for an organization. So tell us a little bit more about that.

Daniel Miessler: Yeah. So this is where this other concept that we have is called telos, which is Greek for purpose. So the telos system is based on starting with a giant context document. It's a very large file, which is still small for AI, but a lot of the files I'm using are like a thousand lines, something like that of text. And what this does is it deeply describes and articulates the purpose, goals, strategies, problems, etc for an entity. So that entity could be you individually, which I have one for myself, my own personal telos file. You could do it for your family, you could do it for a company, you could do it for a city, you could do it for a country, a planet, doesn't matter. What you do is you put in there ... You explicitly describe, I am Daniel, I am about doing this. Here are my goals, here are my KPIs. Here's how I measure myself. Here are all my childhood traumas. Here are all my hangups. Here are things I really enjoy doing. Here's my best method for finding vulnerabilities in a website.

And I've explicitly documented all these different things. And at the end of it, I have my journal. My journal is a running list. Now in the future, I'm going to have this inside of HEPL, and they're probably going to reveal a lot about that in September. But I could be pulling in now my blood pressure and how many times I worked out. And so I could now say to my ai, which is looking at my telos file, Hey, I feel like crap. I feel really depressed. I haven't had any motivation the last two or three days. And it will respond back. Well, you haven't talked to any of your friends, you haven't been working out, you've been eating garbage and you haven't been working out. And this is known behavior. In fact, if you look back at all your journal entries, this always happens. You always feel like this four days after you fall off the wagon with your routine. Here's what you're going to do. You're going to stop reading this and you're going to go execute the following three actions and you'll feel fine in four hours.

Kip Boyle: Now, that's what it would be like for a person.

Daniel Miessler: That's for a person.

Jake Bernstein: I want to highlight the moment of speechlessness because I think that is a ... That was not just a pause. That was a moment of actual speechlessness though.

Kip Boyle: And for you, Jake, that's pretty good.

Jake Bernstein: That is-

Kip Boyle: Well done, Daniel.

Jake Bernstein: I never want for things. No. I think what I love about this, Daniel ... And just to bring up one thing I saw on the main Fabric side is the way you describe it. In other words, AI doesn't have a capabilities problem. It has an integration problem. I find that to be so true.

Daniel Miessler: Yes.

Kip Boyle: Well, that's slant on this whole thing, isn't it?

Jake Bernstein: It is. By the time this goes live, the first iterations of Apple Intelligence will likely be out there on people's new iPhones. And we'll see if Apple ... They typically will. Their whole pattern has been, let's take what other people have done and make it actually useful and great and a good experience. So far, Apple Intelligence that you can access, at least right now as of recording is pretty limited, but it shows signs of being very interesting. So let's go ahead and continue.

Daniel Miessler: Well, can I jump in here real quick?

Jake Bernstein: Yeah. Please do.

Daniel Miessler: Let me jump in here. So I am an Apple crazy person. It's pretty close to being my religion. So I camped outside for the first iPhone in 2007, and I have not missed a single launch. I camped for every single phone. I also worked there for three years and built a pretty cool program there. My take on Apple, and the reason that they're just crushing everyone ... And this is not from any sort of insider information, they are building life OS. So if you look at their health app, if you look at their ... They now have a mindful app. No. What's it called? It's actually called.

Jake Bernstein: I think it's called mindfulness or it's at least a feature of the health app.

Daniel Miessler: Yeah. It's mood. It's mood monitoring

Jake Bernstein: Mood. That's right.

Daniel Miessler: And the mood one, the mood actually looks at correlations across, oh, workouts, work. And some of that it takes from input, but it's also cross-referencing. So the number one thing that Apple does well is build an entire ecosystem and make it so that all the individual pieces benefit from the whole. And that is where the Apple intelligence is going to just multiply the value. Once they get it working. Like you said, they go slow and they try to get it right, but they're thinking about it from first principles from the very correct way anyway, because thinking of everything as an ecosystem.

Kip Boyle: So that's really how-

Jake Bernstein: Integration.

Kip Boyle: Yeah. So we talk about augmented, so people who use Apple devices, when Apple Intelligence is released and is flourishing and fully operating, that's going to be the easiest way for you to augment yourself. Because it's going to be all built in. Right now, what I have to do is I have to actually go over and I have to open up a terminal and I've got to interact with the AI on a dedicated basis and then take the outputs and then copy, paste whatever, pipe, however you want to say it, but then I got to take it over into another tool and do something with it. That's my current workflow. But what would be nice is if it was integrated in all the tools that I use, so I didn't have to do it as just another tool that it's woven into all the tools that I do use.

Daniel Miessler: Yeah. We're about to put it in Raycast.

Jake Bernstein: That'd be cool. For precisely that reason.

Kip Boyle: Now, Daniel, you were going to tell us about this presentation that you're about to make, about how you can use telos for managing a cybersecurity program. Did I say that right?

Daniel Miessler: Yes. That's right. So the detail that I gave on the personal side, now back up and realize, okay, what we're doing is we're doing context files and questions. That's the universal abstracted concept of telos. So now let's apply that to a security program, which is like my expertise on the security side is looking at the programs and how to fix. So what you do in there ... And this is what I do in the talk. I actually have a fictitious company in there. It's called Alma Security. And I've got the origin story, I've got the founder and where they are based out of. It's Redwood City. I've got the company goals, I've got the company mission, I've got the security team's mission, security team goals, how the security team measures itself. I've got the risk register. I've got the team and what projects they're working on, their budget, how they're doing on their metrics.

So for example, a big thing they're trying to do is get their number of criticals down in their core application, especially on crown jewel applications. And if you look at a lot of security programs as both of you, I'm sure a big problem is trying to communicate that you're actually doing a good job to the C team, but also to your engineering counterparts, also internally to the team. And most importantly, it's like, can I articulate what the core of this security org is? What is our mission? What are we trying to do? How would anyone in the org know if we were succeeding? So what this telos approach does ... We're actually ... I built a product around this. I'm not sure if I'm actually going to release it because I don't want to run a company, but I actually have a product around this that does it all.

Basically, you put all the content in there and then you simply ask it questions and it gives you the answers, including give me the current state of the program written for an auditor, written for the head of engineering, written for the C team, written for the internal org. All of those are just essentially Fabric patterns that run against the telos file. So now I could produce graphs on the fly for my progress on mean time to remediate criticals on crown jewels basically every single component of a security program where I need to have answers and narratives to give people, they all turn into these patterns that run against the central file. And once again, you can run a country like this. You can run a family like this. It's all the same concept. It's clear articulation of what you are and what you're trying to achieve combined with these questions, combined with this wrapper of AI.

Kip Boyle: And I think being clear in articulating who you are and what you're trying to do is probably the biggest hurdle for anybody to use something like this.

Daniel Miessler: It is.

Kip Boyle: Because most people are not, people are operating their lives or their businesses or their families more on intuition.

Daniel Miessler: Well, that's the joy that I've had from this is I've been doing this before AI. So I was making these files, these telos files. I was already doing security consulting with telos files before AI. AI is just this magic sauce that can look at the whole thing and keep it in its mind, and auto-generate a report. But to your point, actually forcing somebody to write down the contents of the telos file, you can not even ask any questions and it's already produced massive value.

Kip Boyle: Oh, yeah. Oh, yeah. Yep. No doubt about it. I went through that exercise when I started my company and I have since then started a different Podcast. I started different initiatives, and what I found was the process ... I found a template from somebody that I trusted and just the process of saying, why am I starting this? What do I hope to get from it? How will I know if I'm successful? Just going through that process has been amazing. It has actually been so helpful in orienting me to what I want, that I've been able to materialize those things. Whereas where at the time that I was writing it down, I was like, "I feel like I'm shooting for the moon here when I write down that I want this, this, and this," but it wasn't shooting for the moon. It just felt like it. So I'm excited about this idea of telos because I've seen firsthand what happens when you write this stuff down. Now coupling that with AI, and to be able to ask it questions, how am I doing? Should I be doing something different.

Because in the most recent version of your course, which I took, one of the things that you talked about is building processes and then actually having AI tell you how your processes should improve. Either because there's a new tool available or because there's a new technique available, or maybe the approach that you're using is now obsolete and there's a completely different approach you should be using. And having the AI almost as an agent, keeping an eye on all that stuff for you, and then prompting you when you should make a change. Did I get that right?

Daniel Miessler: That is right. That's absolutely right. And if we take it a little bit further into operations ... And this requires that agents get a lot more mature and come down in cost. But ultimately, if you look at the security operations side, like the technical side of doing this, it's actually something relatively similar. So I have this piece from early 2023 called SPQA. So watch, this state, policy policy is essentially the telos file. So policy describes everything about the company, everything about the security program. Date is actually how many servers are currently running, what software are they currently running, what applications, what versions. What is their current state? Every Kubernetes pod, everything. Network, device, everything. So that state. You have policy, which is the telos thing. Q is questions, which is all these patterns that hit it to ask questions. It could be business questions, IT questions, security questions. And then A is actions. So watch this. What does a vulnerability management program look like in this future, which is happening essentially right now? Here's a crazy stat for you. McKinsey, 40% of their business is now doing this for companies. 40% of their business is AI consultant.

Jake Bernstein: Wow. That is a crazy stat.

Kip Boyle: Yeah. Already.

Daniel Miessler: It just started. It just started.

Jake Bernstein: It's like less than a year old or about two and a half. No, not even two and a half. A year and a half. Are we two years? Maybe we're two years. Has it been two years since ChatGPT?

Daniel Miessler: Probably less than that since they started the practice.

Jake Bernstein: Oh, for sure, for sure. Yes.

Daniel Miessler: Yeah. So that's a crazy number. So just imagine being able to apply AI to your entire program. So here's what VM looks like. You basically say, "Look, we have this issue with doing business in California. These regulations right here really mess us up. By the way, any vulnerabilities in this type of application because it relates to this type of data, that's our worst ones. That's how we prioritize. So we can't have any of those. These types, you know what, we really don't care about because that's not a real attack surface. Even though the vulns exist, it doesn't really matter." Well, now you take all the vuln data and throw it in there as well.

So it looks at these tens of thousands of vulns ... You remember how entire companies would come out just around prioritization. That's just a detail. Now, in this world, that's a detail. That AI does it automatically because it understands what matters to you. So we can automatically sort those, and now you can just ask a question, what do I need to fix first and why? And it just becomes so much easier. And the A piece is like if that hooks into Slack, if that hooks up into GitHub actions, if that hooks up into Jira now, it can cut tickets, it can change code, it could prompt a human to make sure they approve the code that gets changed or whatever. So you have this orchestration now that deeply understands what's trying to be accomplished. It could see everything, which is the state, and what it's doing is it's comparing ideal state, which is policy and current state, which is state, and then using the A for actions to get them in line,

Kip Boyle: Right. Yeah, I love that. So text to action is one of the new things that I've heard as far as forget about text to text, text to image, text to audio, how about text to action that I can type something and that it'll actually turn into a real change somewhere, like a code fix or something like that. This is fascinating. And Daniel, what's really interesting to me is that what you've described about current state versus policy or desired state and gaps and prioritization, all that stuff, we do this all the time or we wish we could do this all the time if we had enough hours in the day and if our brains weren't already melted because we went to all these meetings and did all this other side stuff. And so what we're doing is by doing an augmented approach, it's actually enabling us to do these activities that we've always wanted to do that we know are good, that there's virtue in all this stuff, but now we can actually grab it and we can actually do it and I think that's pretty amazing.

Daniel Miessler: Yeah. Imagine describing this. I like your point because it's exactly correct. Imagine describing to a very smart alien who came to visit and they're like, "Hey, so what's up? I got a clipboard. Just tell me what you're doing." And you're like, "No. This is amazing. Watch this. This is brand new amazing stuff. So what we do is we figure out what we want, we look at things that are bad, and then we change it so the bad things are the things we want." And the alien's like, "Yeah. Yeah, that's what the whole galaxy is doing. What were you doing before?"

Kip Boyle: Welcome.

Daniel Miessler: Yeah. So it's like all of this is so obvious, but because our jobs are so bogged down with all these details and meetings or whatever, we don't have the breathing room to think clearly about, Hey, what are we actually trying to do? And I use the alien for another reason like that too. Alien comes down and visits an average human in 2024. "What are you about?" And they're sitting there with a clipboard and a pen. And the person's like, "Yeah, I go work. I drive into work, I get a coffee, I check the spreadsheet, I make sure the spreadsheet's updated." Blah, blah, blah. And they're like, "No, no, no. That's all stupid. What are you actually about? Who are you? What are you trying to accomplish?"

Kip Boyle: What do you value?

Daniel Miessler: Yeah. People don't have answers.

Kip Boyle: Yeah. We're really lost.

Daniel Miessler: The telos process, it's an exercise of self-discovery for organizations, for countries, for people, for families, whatever.

Kip Boyle: Yeah. Gosh, there's so much that we could talk about. I'm trying to be mindful of the time here. We're coming up on about 40 minutes, and so I feel like I need to start winding down the episode, but I'd love to have you come back sometime. But Daniel, as we wrap up, I want to check with you on something. So when I think of, Hey, we need to augment our capabilities with AI, one of the things that I think of because an old guy is I think of VisiCalc. Because that was the first electronic spreadsheet. And people these days don't really know unless they've watched a movie where you could actually see it, spreadsheets used to be these giant things on a wall that you would have to walk up to with a grease pencil, and we did spreadsheets, but it was all manual.

And then here comes VisiCalc. And so VisiCalc turned these enormous spreadsheets that had to be done by hand by multiple people, and now one person could do it and think about recalculating a cell when you change a value and how the recalculations can flood across a spreadsheet and change all kinds of things. We used to do that too with a grease pencil on a wall, but it would take forever to do even one iteration. And now with VisiCalc, we can do thousands of iterations in a day. An unparalleled productivity. And unfortunately, all the people who got paid money to use their grease pencils on the wall, they don't have jobs anymore because now that's been subsumed into this electronic spreadsheet. Same with word processors, same with PowerPoint.

I remember when PowerPoint first showed up in the late '80s, early '90s, and I'm not kidding you to make the predecessor of a PowerPoint slide before we were using 35 millimeter cameras to take pictures of things, and then that's how we did slide presentations. Of course, now every student in every school can make their own slide presentations. And so I think conceptually what we're talking about is the same. We're saying AI has to be a new tool in your kit of computer literacy. What do you think about that, Daniel?

Daniel Miessler: I think that's right. Unfortunately, the conclusion that this starts bringing us to is going back to that post, you start to realize that more and more of everything we do, even as you move into really advanced jobs like the professions like being a lawyer, they start to look very similar to updating the field in the Excel sheet. And then the natural tend of realizing this tends to go start with this. Holy crap, this really sucks. But then imagine the alien again. The alien will pipe in and be like, "Hey, so everyone gets depressed right about now. Let me tell you the good news. In 20 years or 50 years, you're going to be lifting grandkids, playing in your garden, coming up with screenplays to launch a new movie, which you could use AI to actually release. You could design game levels for people to go and enjoy themselves." Turns out the jobs that you're going to lose, they're actually crap jobs anyway. We shouldn't have been doing that stuff as humans anyway. It's time for us to move on to more valuable work.

Kip Boyle: I really appreciate that. And there's a whole other episode that we could do about just that. The disruption to people based on the loss of these types of jobs. Word processor used to be a job title, not a category of software.

Jake Bernstein: So was computer. Computer was a job title.

Kip Boyle: Computer was a job title, right. Exactly. And that creates a certain amount of distress for people, and that's a whole other podcast that I think we would want to tackle that in. But Daniel, thank you so much for being on our show today. Now, if people want to reach out to you, if they'd like to talk with you about some of the things you've mentioned, how should they do that?

Daniel Miessler: Danielmiessler.com is the site. I'm at Daniel Miessler on X. And you can find me on LinkedIn pretty much anywhere.

Kip Boyle: Great. And then in the show notes, we're going to put the URL to Fabric, which is up on GitHub, so you guys can check that out. Your LinkedIn profile, the blog post that Jake mentioned at the top of the episode, and anything else that we can think of that we talked about. So really appreciate you being here, Daniel. That wraps up this episode of the Cyber Risk Management Podcast. Today we talked about the practical implications of generative AI for individuals, but we also talked about it for programs, cyber security programs, and we even talked about how lawyers are using it now. But we did all of that with our guest, Daniel Miessler. See you next time.

Speaker 1: Thanks for joining us today on the Cyber Risk Management Podcast. If you need to overcome a cybersecurity hurdle that's keeping you from growing your business profitably, then please visit us at cr-map.com. Thanks for tuning in. See you next time.

Headshot of Kip BoyleYOUR HOST:

Kip Boyle
Cyber Risk Opportunities

Kip Boyle is a 20-year information security expert and is the founder and CEO of Cyber Risk Opportunities. He is a former Chief Information Security Officer for both technology and financial services companies and was a cyber-security consultant at Stanford Research Institute (SRI).

YOUR CO-HOST:

Jake Bernstein
K&L Gates LLC

Jake Bernstein, an attorney and Certified Information Systems Security Professional (CISSP) who practices extensively in cybersecurity and privacy as both a counselor and litigator.