Larridin Blog

Bask Iyer on Who Owns the Bot: AI Accountability and More

Written by Floyd Smith | May 12, 2026

See our AI Impact podcast episodes on YouTube.

Who owns the bot that just discounted your inventory 30% on Thanksgiving? In the latest episode of AI Impact, Bask Iyer, former Global CIO at Johnson & Johnson, Honeywell, VMware and Dell, tells Larridin co-founder and CEO Russ Fradin that most enterprises can't answer that question, and that's where AI accountability actually breaks down.

They also cover the “pretotyping” framework that cuts a nine-month proof of concept to one month; why matching your competitor's tech stack leaves you with no budget and no differentiation; and why "the tool saves me time on weekends" is not a result the CFO should accept.

Key points

Bask started his career selling refrigerators and air conditioners, studied computer science, worked for NASA, and went on to serve as global CIO at six major companies on both coasts. He now advises about a dozen AI companies through his firm, BaskMind. Several key points emerged.

Somebody has to own the bot

Bask sees AI redefining organizations, and pulling the CIO, CFO and head of HR closer together. Large companies already struggle to track who works for them: employees, contractors, consultants, contingent workers. Now they also have bots and agents running around, some of them talking to each other and making critical decisions.

If an agent sees that Amazon has cut prices for Thanksgiving and drops yours by 30%, that might be a valid call. But if it goes badly wrong, who made the decision? Who authorized the agent, who updates it, and who fixes it? Bask says those three leaders together need to be able to answer.

Agents instead of monoliths, and guard your data

When Bask ran IT at companies like Honeywell, build versus buy was simple: get the best tool at the best price, which usually meant buying, from email to ERP. Now, he says, you don't need a monolithic, million-dollar system to handle something like accounts payable; a task-specific agent can do it, and your own team can often build it. (He also says AI has made programming fun again, like solving a sudoku puzzle.)

But he urges caution on partnerships and data. Companies that put all their data into the cloud and the big LLMs risk losing their differentiation over time, so as companies grow, he favors hybrid computing and a hard look at what's proprietary.

Take chances on startups, and don't copy your competitors

Bask coaches both sides of the startup-enterprise divide. Founders, he says, need the humility to leave their zip code, get on a plane, walk a factory floor, and learn the actual business problem. CIOs need to take more chances: a startup's founders will work on your project personally, which a giant vendor never will, and a $60 or $600 license is a small bet. "Just buy the damn thing and use it," he says.

To keep vendor evaluation manageable, he set aside a weekly Friday innovation block for startup pitches and brought his whole team. And he warns that if you run the same stack as your competitors, from ServiceNow to Salesforce to AWS, the subscriptions eat your budget, and you bring nothing distinctive to the table.

Prove it fast, and call the bluff on cost avoidance

Many CEOs now feel they're behind on AI and ask Bask whether to add a chief AI officer. His advice: either back your CIO or replace them, but don't stack up competing leaders. Instead of telling a CEO they're wrong, Bask brings a framework he calls pretotyping: assume the AI claim is right, then design a one-month test, with the CEO and CFO involved, that proves or disproves it, rather than a nine-month pilot.

Half the time, the CEO concludes on their own that it isn't working. Russ describes CIOs whose business leads raved about their new AI tools until the CFO asked for headcount savings to pay for them. Bask says leaders have to call that bluff. Working five days instead of seven is cost avoidance; the real test is whether costs came down or revenue went up for the same number of people.

Conclusion

See the full episode on the AI Impact Podcast playlist on the Larridin YouTube channel. To get a demo of Larridin, contact us.

Quotes

  • "Do you know who the bots are, who authorized them, and how it worked?" (Bask Iyer)
  • "The real test is, did actual costs come down, or did you produce more revenue for the number of people you have?" (Bask Iyer)
  • "If you do everything your competition is doing, if your competition has ServiceNow, Salesforce, Cisco, Amazon Web Services, whatever, and you have it too, what do you really bring to the table?" (Bask Iyer)
  • "Everyone's trying to retrain their global workforce all at once, with a new set of tools that didn't exist six weeks ago in some cases." (Russ Fradin)
  • "Every CEO feels like they're way behind." (Russ Fradin)
  • "Is it making you more productive or is it not?" (Russ Fradin)

Q&As

Russ had questions; Bask had answers.

Q: How did you used to think about build versus buy versus partner, and how do you think people should think about it today in AI?

A: Previously ... your job was to get the best tool at the best price to get the job done. ... As you go further now, it looks to me like you don't really have to buy a monolithic large system like we used to. You can buy these systems that are specific agents. ... For accounts payable, we can create an agent. ... On partnership, I would be very careful with data. People are going to the cloud, putting all their data with the AI LLMs and so on. You lose your differentiation after a while.

Q: How do startups do a great job partnering with CIOs, so that a pilot becomes a long-term relationship?

A: One is a little humility. ... You started a company and worked in the same zip code. ... Go to the East Coast, go to the rest of the world, and learn the business problem. ... And I do the same thing with the CIOs on the other side. ... Take the risk of working with some really clever people who are trying to solve the problem. ... Just buy the damn thing and use it. Put some money in, put some skin in the game, and take a chance.

Q: What's the best way you've found to work with very smart, hard-charging CEOs who feel they're behind on AI?

A: You can't go to most CEOs and say, "Hey, this is too early. We're not behind. You're wrong." ... Instead of saying no, I came up with a framework for success. ... I call it a pretotype. ... Let's assume it's right, but what would prove it as quickly as possible? So we have one month to prove whether this thing will reduce our workload by 50% or not. ... Halfway through, 50% of the time, the CEO himself will say, "I don't think this is working."

Q: What advice would you give younger leaders today, in terms of having a successful career in a world of AI?

A: I would recommend some risk-taking. ... You can't get anything without taking risk. ... The second thing I tell them is, don't bet so much on traditional vendors. ... The second thing I tell them is to spend some time in Silicon Valley. ... And the last thing I tell them is ... don't spend so much time at one company, because your learning comes down.

Topics in this episode

Key topics in this episode include:

  1. AI impact: Bask says something real is happening with AI tools: he expects big productivity gains, even 50% reductions in some workloads, soon, if not yet, and says now is the time for companies to figure it out.
  2. AI tool usage: Russ describes CIOs whose business leads loved their new AI tools and wanted to renew, until the CFO asked for headcount savings to pay for them.
  3. AI costs: Bask warns that subscriptions to the same big vendors and cloud providers as your competitors leave no budget for innovation, and urges CIOs to just buy promising startup tools at $60 to $600 a license and try them.
  4. AI ROI: Bask rejects cost avoidance, like working five days instead of seven, as a result; the real test is whether actual costs came down or revenue went up for the same number of people.
  5. AI workflows: Bask expects companies to build more agentic work themselves, such as an accounts payable agent, instead of buying monolithic, million-dollar enterprise systems.
  6. Human productivity with AI: Bask, who left programming because it got boring, says AI has made programming fun again, like solving a sudoku puzzle.
  7. AI agent productivity: Bask asks who owns the bot that drops prices 30% on Thanksgiving, and says CIOs, CFOs and HR leaders must work out who authorizes agents and who fixes them when they go wrong.
  8. Enterprise AI: Bask's pretotyping framework gives big companies one month, not nine, to prove whether an AI tool delivers what it claims, with the CEO and CFO involved.
  9. AI and the Board: Bask's counsel to boards and CEOs who feel behind on AI is to either back their CIO or replace them, not to add a chief AI officer alongside.

Full Transcript

See this latest episode of AI Impact.

Chapter 1: Bask's career arc from NASA to global CIO to AI advisor (0:00)

Bask: I'm actually not surprised that a lot of CIO jobs still exist, because it is changing. Most traditional companies expect a CIO to be a bridge between technology and business. Previously, people fell into the trap of command and control. It's impossible to make that call, because technology is changing so fast. Everybody's job requires technology.

Russ: How did you used to think about build versus buy versus partner, and how do you think people should think about it today in AI?

Bask: I would be very careful with data. People are going to the cloud, putting all their data with the AI LLMs and so on. You lose your differentiation after a while.

Russ: What advice would you give younger leaders today?

Bask: You can't get anything without taking risk.

Russ: Bask, it's great to see you again. Thank you for taking the time. Just to dive right in: you've had such an impressive background. Could you tell the audience, in two minutes, your career arc, how you wound up CIO of so many amazing companies? And then I have a ton of questions about CIOs and AI.

Bask: I was pretty much a businessman. I started my career selling refrigerators and air conditioners. Then, accidentally, I came to this country to study computer science, and computer science just took off. I came to Florida Tech, I started working for NASA early, and the whole computer industry went up. So it was very difficult to get back into business. I mean, no complaints, because my career took off. I spent about a third of my career at large companies like Johnson & Johnson, GlaxoSmithKline, Honeywell, etc., largely on the East Coast and in the Midwest. And for the last 15 or 20 years, I've been on the West Coast, working with companies like VMware, Dell, Juniper Networks, etc. So it's the combination: I've been a global CIO for large enterprise companies, and then for these typical startups, which are startups no more, because they've grown so big these days. And then, for the last couple of years or so, I've had my own company. I do boutique consulting on digital transformation with AI, mainly bridging traditional companies with startup companies in Silicon Valley. So I'm an advisor to about a dozen AI companies, like Zoom, Cohesity, ThoughtSpot, Automation Anywhere and a few others whose names you wouldn't know. That's what I do now.

Chapter 2: What the CIO role actually requires across enterprise and tech companies (2:20)

Russ: So before we get into what makes a great one, what's your definition? What does a CIO actually do? It's one of those titles we hear, but it's very different from Honeywell to VMware to Dell to J&J. What is a CIO's core job?

Bask: I was one of the first few CIOs. There was no such title as CIO when my career started. Typically, they took the most technical person in the company and put him in that job. So initially it started off as a technical person who could make decisions on what to buy for the company. And then it transitioned. When I was at a company like Honeywell, they wanted a general manager who knows technology. They didn't want a technologist; they wanted a business person who could do technology. And for a large company, you need leadership. You're working with presidents of companies and CEOs and so on, and they smell weakness if you're not there. So a lot of leadership is required. Most East Coast companies, or most traditional companies, expect a CIO to be a bridge between technology and business. That's how they look at it. They look at you as a leadership team member. These days everybody knows technology, so they all believe they're CIOs. But typically, in a Fortune 50 company, you are the technology leader for the company. You're expected to lead, and to take charge. There's some serious stuff, too. You have to manage cost. You have to manage the security aspects. You're the single throat to choke if something happens on your e-commerce site, or things like that. But there's a lot of creativity in the job, too. On the West Coast, in Silicon Valley, your founder has a PhD in computer science, so he or she knows more than you do. So the role is different. You don't try to compete with the person who's technical. You have to have the modesty and humility to say, hey, this person knows what he's doing, but you bring your strengths to the table. In a technology company, some of the technology CEOs have challenged me, and I always explained to them: if you go to a music store, the guy who runs it can play all the instruments. He can play guitar, keyboards, drums, etc. But he's not Eric Clapton. So, I am Eric Clapton, is what I tell them. This is what I do for a living. I may not be the inventor, but I'm the person who's responsible for your security and so on. So we're not trying to compete on who knows what better, but you've got to take charge. I do it jokingly, but you have to land the point, because everybody in a technology company believes they know better than you, and they do.

Russ: And also, it's gotten easier with the cloud. Or it seems easier with the cloud now. It's not like the real old days.

Bask: No, no. And I think that's one of the things that has changed. Previously, people fell into the trap of command and control: I make all the decisions, etc. And it's impossible to make that call, because technology is changing so fast. Everybody's job requires technology. Even finance jobs require technology; HR jobs require technology, not just product development. So you become more of a mentor and collaborator now, rather than saying, "Thou shalt not do this, and thou shalt do that." That style doesn't work.

Chapter 3: Why CIOs, CFOs, and HR must share accountability as AI agents multiply (5:52)

Russ: So as you think about organizational design, especially as it relates to AI, everyone's trying to retrain their global workforce all at once, with a new set of tools that didn't exist six weeks ago in some cases, and didn't exist three years ago in all cases. And then you have concerns about job displacement: what's going to automate my job? How are people going to lose jobs? You have people caring about cost, people caring about security. What do you think the relationship will look like in the future for CIOs, CFOs and heads of HR? I think those are really the three key roles that are going to be impacted, especially at larger companies, and that are really going to have to change the way they work together. So if you think about your experience, how did you used to work with the head of HR, versus how you'd imagine a CIO today has to work with a head of HR and a CFO?

Bask: I've always worked very closely with the CFO and the head of HR. They're pretty important positions, because a CEO looks at them as their chiefs of staff: get the job done, because I have to worry about sales and customers and other things. Don't come and bother me with all the stupid things. So it's good to have good collaboration anyway, regardless of what you do. But as you put in more and more AI, organizations are going to be redefined. The HR job becomes very complicated. At the same time, HR jobs are being replaced. I can see a future where there aren't HR specialists, or even IT specialists or finance specialists. AI is coming after a lot of the back-office administrative jobs. It's already there. I'm actually not surprised that a lot of CIO jobs still exist, because it is changing. In most jobs, if you're doing help desk support, that's going to go, because people are going to just go and search. It's a lot easier to search and find than to ask someone to tell you. Claude is going to tell you. You're not going to go into your ServiceNow and search for how to fix this. So I think those jobs are gone, but the jobs that remain are jobs that require human intelligence, or EQ: collaboration. That's where Claude hasn't caught up yet. Even though it acts very friendly, sometimes it gives you advice that's totally wrong, but it very confidently tells you.

Russ: Like a lot of co-workers.

Bask: Like a lot of co-workers. Exactly. You actually can't knock AI, because I can find other bad examples in humans.

Chapter 4: The agent accountability gap: who owns the bot, who authorized it, who fixes it when it goes wrong (8:15)

Bask: But I think a good CIO, a good head of HR and a good CFO will come together to redefine the organization: what should the function of G&A be, who are the critical people, and then also governance and other kinds of things. It's tough enough to know how many people work for you: who the consultants are, who the contractors are, who's full-time, who's showing up to work, who's a contingent worker, all these categories. We still haven't gotten a hold of that in a large company. Now you have robots, and different bots running around, and I'm thinking, well, do you know how many of these are running your organization, making critical decisions? Who's in charge of them? Who owns them? Who updates them? What if one of them is talking to another one, like agents? This agent is talking to the other agent, and all of a sudden your e-commerce site doesn't work, or you're discounting your goods by 30 or 40%. Who made the call? Did the marketing person make the call? The finance guy? Did somebody make the change? A lot of times, the agent would make the call and say, "It's Thanksgiving. Amazon has already dropped the price. Let's drop our price by 30%." That's a valid call, but what if it does something really bad, and you're losing out on revenue or operating income? So, do you know who the bots are, who authorized them, and how it worked? I think this is going to require a combination of these three individuals, primarily.

Russ: Yeah. I mean, certainly we're going to have the merging of the human worker and the tools they use alongside the digital worker, and that's just going to force more collaboration between CIOs and heads of HR. There's no other option. And the nice thing is, none of us really knows the answer, which makes us work together.

Bask: Yeah.

Chapter 5: Why task-specific agents are breaking the case for monolithic enterprise systems (10:06)

Russ: I'm curious. If you think about 20 years ago versus today, how did you used to think about build versus buy versus partner on technology needs? Today, on the one hand, you see a lot of companies deeply partnering with Anthropic or Google or Microsoft or whomever on the AI side. And you certainly see a lot of companies thinking, "No, no, we need to build our own. We need to have our own model. We need to build our own context. We need to be really locked down." How did you used to think about build versus buy versus partner, and how do you think people should think about it today in AI?

Bask: Previously, especially if you weren't in a tech company, if you were at a company like Honeywell, your job was to get the best tool at the best price to get the job done. So make or buy was a simple decision. If I can do it in two days and it takes the vendor four days, or it costs half the money, I'll do it myself. But generally you did a lot more buying, because it wasn't your job, or you didn't have the people to keep building several of the tools. So most of the time, whether it was email, calendaring or ERP, you bought a lot of tools, including hardware and so on. As you go further now, it looks to me like you don't really have to buy a monolithic large system like we used to. You can buy these systems that are specific agents. So now you have to be clever. Are you buying an entire enterprise management system, or just something that does accounts payable? For accounts payable, we can create an agent. You see a bill, you pay a bill, and you have some rules. Do you really need this monolithic, million-dollar system just to handle your accounts payable or expense reporting? So I think you're going to see more agentic work being done, and you'll have your own folks do it, because it's not rocket science, like you were saying. I loved programming. I got out of programming because it didn't pay enough money, and it got boring after a while. There were too many languages and things to remember. But now I love programming again. I said, "Huh, this is easy. You just have to ask questions, and it tells you what to do." [laughter] And it's like solving a problem, like a sudoku puzzle. So I thought, wow, programming is fun again. So now you're going to have more agentic stuff, and you're going to do a little bit of building. On partnership, I would be very careful with data. People are going to the cloud, putting all their data with the AI LLMs and so on. You lose your differentiation after a while. So I would be very cautious. First of all, I'm thinking even about computing: I'd do hybrid computing. I wouldn't put everything in the cloud. If I'm a startup, I put everything in the cloud, but as I grow bigger, I think, wow, should I really put everything in the cloud, especially data? What is proprietary data? And then you have to make some calls on certain agents. I described a simple agent, like accounts payable. I'm sure somebody has written that already, or if it's not already there, it will be. Just buy that. You don't have to write it.

Chapter 6: What founders consistently get wrong about selling outside Silicon Valley (13:14)

Russ: That's an interesting question, actually. I'm curious how you'd think about this. I know a lot of your more recent career has been partnering with early-stage startups and helping them grow, pre-AI and certainly post-AI, helping them figure out how to sell to someone like you. How do I partner with CIOs? A big question that every VC in Silicon Valley is worried about today: fundamentally, you've seen explosive growth of these AI startups. The truth, as folks know, is that what you've seen is explosive growth of lots of proofs of concept. You're seeing giant organizations that used to move very slowly, because of the fear of being left behind in AI, which I know you've talked about in the past, now moving very quickly to do lots of pilots. So you have this interesting conundrum as a VC. On the one hand, you're seeing a company growing faster than any company you've ever seen grow. On the other hand, all the revenue is brand new, and these are all pilot deals. So how do you think about that world, as a CIO, where CIOs today have had to develop a whole new muscle around trying new vendors at a volume they've never done before? How do startups do a great job partnering with CIOs, to give themselves the best chance that a pilot with Toyota becomes a long-term relationship with Toyota?

Bask: Yeah. I advise two things. One is a little humility. Most of the CEOs, the founders, are super bright. They're all very clever; there's no doubt. Unfortunately, a lot of them know it, too. What I tell them is, you're very good, but you started a company and worked in the same zip code. You've not even left the zip code. So you need a little curiosity to partner with a CIO outside of Silicon Valley. Go to the East Coast, go to the rest of the world, and learn the business problem. Don't assume that if you build it, they'll come. The product validation part: I spend a lot of time making sure the product is validated, that it's something they want. They do it, but I think you have to shock them a little bit to say, "Dude, you've got to go out and do it. Don't talk to your Silicon Valley friends and everybody else and assume you know the problem, because you really don't know the problem." And I do the same thing with the CIOs on the other side, to say, listen, don't bet on the same existing companies all the time. If you make a decision to go with Salesforce or one of those companies, you're going to meet the salesperson. That's about it. You're not going to meet the top people. You may see them for one minute, but they're not going to be interested in your project. A million-dollar project is not a big deal for those companies. So I say, come down. Yes, it's a new company. It's not known. It's in Silicon Valley, but these are two bright people. The CEO and founder, and the co-founder and CTO, are going to work on the project for you. You'll never have that luxury again, because soon they're going to become big, they're going to have 30 other customers, and they're going to behave like everybody else. So I tell them, take the risk of working with some really clever people who are trying to solve the problem, who think of your problem as the most important problem for them, because they want to get it done. Take some risks. I see a lot of CTOs and CIOs who come and talk to the startups, but they never take a chance. They're still talking about, "Should I do a prototype?" or whatever. It's a subscription model: 600 bucks for a license, or 60 bucks for a license. Just buy the damn thing and use it. Put some money in, put some skin in the game, and take a chance, is what I say. And two things happen. One, you push the envelope and see what your requirements are and what's missing. And you get the very top folks to work on it. I've profited with some companies because I took a chance early. I got these licenses for next to nothing, and forever. Right, Russ? I know you've been an entrepreneur for a long time. If I partner with you early, I get a product, and you become huge, you're never going to drop me. You're going to remember me for life.

Russ: 100%. Yes. There's nothing better for startups. Look, the ideal relationship is always a startup with a large company where it turns out everyone bet right, and you wind up with these very long-term partnerships. For sure. I imagine, to your Marc Benioff story from earlier, he still picks up the phone for his first five clients, even if they're not Salesforce's largest clients today.

Bask: He does. I don't blame him, either. There's a scale issue, but he'll always pick up the phone for the first five people. He never forgets them.

Chapter 7: How Bask structured a weekly Friday innovation block to evaluate vendors without derailing his team (17:53)

Russ: So how do you think about the large companies that are working with AI? Is it CIO-driven? Do they set up an office for AI transformation, digital transformation? Is it happening at the department level, where the head of sales is picking the tools he needs for his team? It's an extremely complicated world. If you were a CIO, my guess is you could fill every minute of your calendar, every week, with 30-minute slots just to meet new vendors you'd never heard of, all day, every day. You'd never get anything done. So how should companies be doing this? Because you can't try everything. Forget that much of it is useless. You just can't try everything. It's impossible.

Bask: Yeah, that's right. I think you put some discipline in your calendar. Say every Friday, or whatever, you have an open innovation day or something. Friday afternoons, if you're on the East Coast, can typically be a little lighter, and it's still Friday morning on the West Coast. So you can always get some startups to come and tell you what they're doing and how they're solving the problem. And it does two things. I typically tried to get my whole team to come and attend those sessions, and some of my team members complained when I set it up. They said, "It's a waste of time. I don't have time for this nonsense," whatever. But after I left the company, almost everyone came and thanked me for it. They said, "Oh, Bask, you forced me to get out of that and go see this new vendor, and that opened my eyes." But they never admitted it at the time. So I think it's a good team-building thing. It's a good one for the whole company, to be innovative without investing money or getting VC funding. It's nice to just pick your brain for half an hour without having to invest in your company.

Russ: I'm curious what you think of this. I've spent my whole career selling things to large companies, and to your point, one of the things you see with founders... your advice to founders is to go out there and understand what it's actually like to work at J&J. On the other side, a thing I've seen frequently from people who've spent their whole careers at large companies is this: when you're at a company like J&J or Honeywell or IBM, it's actually very possible for you to spend your entire life only dealing with other people who work at that same company, which isn't a great way to grow your overall knowledge of the world.

Bask: Exactly. I think you're dead right. I always tell people that when I was at Honeywell, I knew the 120,000 people at Honeywell. Well, that's how I felt: I knew everybody, and not one person outside the company. There's just no time. And when I worked at Silicon Valley companies, I didn't know everybody in my own company.

Russ: Yeah. Wow.

Bask: The reverse is true. And for networking, career development and professional development, I think the latter is better. So somehow you have to get to a situation where you have that network above and beyond. That's why I tell CIOs to reach out. It's helped me, it'll help you, and it'll help your team members a lot. And likewise, I coach the founders. You've done this for a long time, so you're comfortable, but a lot of founders starting now just do a Zoom call or an audio call, because it's convenient and it works fine. I tell them, just take the plane and go. I know it's a pain in the ass, but Honeywell is calling. Just go. Yes, it's a waste of a whole day, and you could have done it more efficiently in half an hour, and I know you all work from home. I understand all that logic, but just go. The amount of learning you get by walking a factory or a distribution center, you can never get otherwise. So I think both sides have to come a little closer together. And then you look at Europe. I'm just talking about the East Coast and West Coast, but Europe has to spend more time, too. The innovation is happening in Silicon Valley, and I may exaggerate, but a lot in Silicon Valley, Israel and Bangalore. You have to be in those three places somehow if you want to be in technology.

Chapter 8: How to handle CEOs who are convinced they are behind on AI without losing their confidence (21:53)

Russ: So how do you think about one other thing you've spoken about in the past? It feels like it's becoming a much bigger issue now when I speak to CIOs. They have this feeling, or this feeling is pushed on them by their CEOs, that every CEO feels like they're way behind. Every CEO feels like their competitors are using AI better, cutting costs better, selling more, acting more efficiently. If you think about your career, there have been other points of giant technological dislocation. What's the best way you've found to work with very smart, hard-charging CEOs to help them understand what's actually happening in the market? Are we behind or not? Frankly, should we just take a deep breath? How should you think about digital transformation at a time when everybody feels behind, when every analyst wants to understand your plan with ChatGPT to lay off half your employees or triple your sales or whatever?

Bask: Yeah. It's a difficult question, because the best CEOs are also very opinionated. That's what makes them good, and that's what makes them bad CEOs as well, because sometimes you need strong opinions and the ability to drive, and it can work both ways. So you can't go to most CEOs and say, "Hey, this is too early. We're not behind. You're wrong."

Russ: Right. "You're wrong."

Bask: Yeah. That's not going to work. They'll go talk to the CFO or somebody else and say, "My guy is not that good." In fact, a lot of CEOs are calling me now. Previously, a lot of CEOs used to call and say, "My IT guy is behind on budgets," and blah, blah, blah, "Can you be a mentor or coach?" or whatever. Now the people who know me are saying, "My IT guy is great. He's handling his projects, but I don't feel like we're going fast. Should I hire a chief digital officer? Should I hire a chief AI officer? Should I hire a consultant?" And I always tell them, well, first you have to give the person a chance. You have to be very clear about what you want. And you don't want two heads of HR, or two heads of IT. I don't believe in two of these jobs, but some companies just say, "Let the IT guy run the systems, and let's hire a chief AI officer." I would rather replace the person, if he or she isn't good, than add another person. So that's the counsel I give most boards and CEOs: either your person is good and you have to give them a chance, or you've got to replace them. You can't have three people, because they'll all come and say something different.

Chapter 9: Pretotyping: the one-month framework for pressure-testing AI claims before a nine-month pilot begins (24:18)

Bask: But one of the things I've done with CEOs that worked is, instead of saying no, I came up with a framework for success. So I said, "Oh, you think such-and-such LLM is going to replace 50% of your workforce? Let's put it to the test. Here's the framework. Here's the pilot. Here's what we're going to do. And I want your CFO and other people in this project. I want you to be involved." And then go through it and ask, did we eliminate it? That way, you're doing the right thing, giving it a proper test, not just nodding your head. And secondly, if it works, you give the credit to the whole team. It's not you who found a cool new tool. And honestly, job elimination, those kinds of transformations, are not a one-person job. If the CFO or the HR person or the head of procurement doesn't want to change, no system is going to fix it. So I always come up with a framework that says, okay, let's assume this is right. I call it a pretotype. It's like prototyping, but pretotyping. Let's assume it's right, but what would prove it as quickly as possible? So we have one month to prove whether this thing will reduce our workload by 50% or not. We're not going to spend nine or 10 months trying to see if it does. We're going to accelerate and run it. What are the criteria? What should we do? And that seems to work, because halfway through, 50% of the time, the CEO himself will say, "I don't think this is working," which is what you want, rather than you saying it doesn't work. Or it has been successful, too.

Chapter 10: Why cost avoidance is not a real business result and who needs to call the bluff (25:52)

Russ: We had a dinner recently with about 20 CIOs, and they all told some version of the following story, so I'm curious how you think about this. They decided they were going to do these AI tests. They bought a bunch of AI software and deployed it everywhere, and all the business leads they'd bought the software for came back and said, "This is making us way more productive. This is amazing. Thank you. The software is great. Renew. I need more. This is making us way more productive." And the CFO said, "Great. To pay for it, we're going to cut 15% of your headcount." "No, no, no, no, no. We're not ready to do that." Well, is it making you more productive or is it not? Are you going to have a higher revenue number? No. Are you going to have a lower cost number? No. So what really is the plan here, with all of these tools?

Bask: The truth is somewhere in the middle, but I think the CEOs and CFOs have to call the bluff. Cost avoidance is something I've always had an issue with. Almost everybody says it's wonderful, and then, what's even worse, they'll say, "I can't reduce people, but what happens is, I used to work every Saturday and Sunday, and now I'm working only five days a week." That all sounds like cost avoidance. So to me, the real test is, did actual costs come down, or did you produce more revenue for the number of people you have? That's what you have to do. Otherwise, what happens? Yes, all these AI tools I'm using definitely make my life better. But will I take a 20% cut because my job is easier now? No freaking way. I'm going to be protesting. So without being too judgmental... I don't want to be too dogmatic, because the tools do work. I remember at J&J, I think it was very early, I had to go justify email to the CFO. Just email. And he made me go through hell. He made me cry. "What? You think I'm going to type all my email and get a message, and somehow, magically, we're going to reduce three secretaries and save so much money because we can send emails?" And, "What's the cost of not doing it?" He had so many questions that I had no answer for, as a young CIO. And I said, "Okay, if you don't want email, more power to you." [laughter] So you can go to the other extreme. And I tell CEOs, CFOs and other people, "Hey, there's something real happening now with AI tools. This is real. I do see it reducing people by 50%, productivity gains. I don't see it now, but I see it happening soon, and it's a good time for you to figure it out." That's what I tell them. So I think you have to be that way. At least one person, maybe the CFO or the CEO or somebody, should be the bad guy and say, "Where are all the people going? How much money is falling to the bottom line?" Somebody has to ask the question.

Chapter 11: Three pieces of career advice for aspiring leaders in an AI-first world (28:42)

Russ: Obviously, you're an AI bull and think these tools really are having an impact, and obviously at Larridin we believe that and spend a lot of time helping companies measure that value. I'm curious: if you think about yourself 30 years ago, or as you deal with younger people in your various consulting projects, what advice would you give younger leaders today? Not really CIOs so much, but younger aspiring CIOs, or aspiring heads of HR, or aspiring heads of anything, in this world of so much technology dislocation. What advice would you give people who've decided to enter this business world, in terms of having a successful career in a world of AI?

Bask: I would recommend some risk-taking. I feel like my colleagues and my comrades have become so risk-averse now. If you look at the CIO or CISO, the definition is always a person who doesn't take risks. It's become that bad. So I'm telling them, you can't get anything without taking risk. There have to be risks that you take. Yes, it's a pretty serious job. You've got critical systems, and all that stuff is fine, but you've got to take risks. And the second thing I tell them is, don't bet so much on traditional vendors. Remember the saying from a long time ago: you can never go wrong by choosing IBM. That thing is still going on with 30 other companies. You can never go wrong with ServiceNow. You can never go wrong with Salesforce. So I tell them, especially when you're young, take those chances. Try to attack these systems with agentic AI and other kinds of stuff while you still have the energy. Because most CIOs aren't technical like they used to be. When I started, you had to be technical. So they go very safe. They say, "What does Gartner recommend? Is it in the quadrant or not? Who are the top companies I should pick? I can't get fired if I choose these five." And when you look at your budget, especially with subscriptions, you're paying so much money to all these large vendors and cloud vendors that you have no money left to do anything. So I tell them, you need to pick a few. What are you going to target? If you do everything your competition is doing, if your competition has ServiceNow, Salesforce, Cisco, Amazon Web Services, whatever, and you have it too, what do you really bring to the table? So I tell them, don't take on everybody, but figure out where you want to take a chance, because that frees up so much of your money for other kinds of innovation. So that's one thing I tell them: take some risks. The second thing I tell them is to spend some time in Silicon Valley, if you haven't. Get out of the office, spend the time, build your network and everything. Because the other thing you get in Silicon Valley, when you spend time there, is that you can quickly validate. You don't need a vendor pitch. You can quickly ask, is this company real or not? You can have coffee with somebody, and they'll say, "Russ is great," or "Russ is crap." You don't need LinkedIn. You don't need anything for that.

Russ: Right. It doesn't have to be that complicated.

Chapter 12: Why staying at one company beyond three years slows your learning curve (31:47)

Bask: Yeah. So those are two things I tell them. And the last thing I tell them is general advice. I spent a lot of time at large companies, for a long time, and I tell people, don't spend so much time at one company, because your learning comes down. Pick a time, three years or whatever, because the world is too big, and you need to know it. The nice thing, again, about Silicon Valley is that a lot of people have gone from one company to another: they work at a cloud company or Tesla, and then they go to Apple. So they bring a lot of knowledge from different companies, and that's very helpful to both the individual and the company. So I tell them, don't get too comfortable in a job. After the third year or so, you've learned everything, more or less.

Russ: Well, I really appreciate you taking the time to talk to me today. Thank you very, very much.

Bask: Thank you.

This transcript was prepared by AI and may contain mistakes.