Why Most AI Transformations Fail — and What Leaders Must Do in 90 Days | Charlene Li の全文書き起こし

first things that came across in your book that I resonated strongly with was you don't need an AI strategy. 00:03:15.629 --> 00:03:20.750 What you really need is a business strategy that's enabled by AI. 00:03:20.750 --> 00:03:24.349 Can you just unpack that a little bit by what that means? 00:03:24.349 --> 00:03:28.109 And what does it mean for the CEO looking at his business right now? 00:03:28.590 --> 00:03:36.349 Well, in so many organizations, the CEO is thinking about their business, and they think over here on the side is AI. 00:03:36.349 --> 00:03:38.030 And we got to do something with AI. 00:03:38.030 --> 00:03:39.549 What's the RRI of AI? 00:03:39.549 --> 00:03:41.469 And all the questions are on AI. 00:03:41.469 --> 00:03:50.829 And they're looking at it like a technology, like a thing that has to be dealt with, versus saying, what can AI do to support my business strategy? 00:03:50.829 --> 00:03:54.909 How do I do the things I'm trying to accomplish with my business strategy? 00:03:54.909 --> 00:04:05.389 How do I get to it sooner, so faster, more efficiently, so cheaper, uh, and and and better, and maybe do something completely different and even safer. 00:04:05.389 --> 00:04:11.629 AI can help you deliver this in a safer way too for yourself and organization and for your customers. 00:04:11.629 --> 00:04:16.430 So they're not asking and connecting AI back to their business strategy. 00:04:16.430 --> 00:04:21.709 So instead of an AI strategy, think about how AI can be used to support your business strategy. 00:04:22.269 --> 00:04:26.269 We had uh Stephen Wonker on the podcast uh a while ago now, actually. 00:04:26.269 --> 00:04:41.870 You know, talking about uh the model that HelloFresh did, where they completely changed their business model, which is I can inspire by because I just sadly get the impression that most people perceive this to be a cost-cutting tool and therefore people are nervous about their jobs. 00:04:41.870 --> 00:04:45.069 You know, the HelloFresh case study just disproved that. 00:04:45.069 --> 00:04:53.470 Can you tell us any other examples that you came across when you were preparing for your book about how people reinvent their business model? 00:04:54.030 --> 00:04:56.990 Yes, um, one of my favorite examples is from Connector. 00:04:56.990 --> 00:05:03.069 Um, and they're a global call center, um, business process outsourcer, BPO. 00:05:03.069 --> 00:05:05.790 And again, as you can imagine, they have all these call centers. 00:05:05.790 --> 00:05:12.189 And they immediately said, we're going to use AI, not to cut people, because our problem isn't that we have too many people. 00:05:12.189 --> 00:05:19.949 We don't have the people trained to the levels we want them to be at in order to deliver the services we want for our customers. 00:05:19.949 --> 00:05:26.670 So they used AI initially to raise the quality of their people, the performance across the board. 00:05:26.670 --> 00:05:28.349 And the numbers were

just astounding. 00:05:28.349 --> 00:05:33.550 Error rates in particular just went away because the quality increased so much. 00:05:33.550 --> 00:05:35.949 Training times decreased significantly. 00:05:35.949 --> 00:05:39.709 And what they also said is what are the new things we can do now? 00:05:39.709 --> 00:05:48.430 We have these long lists of services we've always wanted to use and to create for our customers, but we've never been able to do them. 00:05:48.430 --> 00:05:52.750 Now we have the time and capacity, but we also have new capabilities. 00:05:52.750 --> 00:06:11.550 In particular, they looked at all the things that a certain group of clients, for example, law firms, who don't have the capacity to put in significant AI tooling, and said, could we actually take our new capabilities with AI and deliver these new services that didn't even exist before to our clients? 00:06:11.550 --> 00:06:13.310 Because we know them so well. 00:06:13.310 --> 00:06:22.189 So I think that's a great example of a business that you would think normally would be decimated by AI and just cutting people left and right. 00:06:22.189 --> 00:06:26.509 And their commitment was to say, no, how do we do things better and in a different way? 00:06:26.509 --> 00:06:29.230 Right from the very beginning, that was their strategy. 00:06:30.189 --> 00:06:42.829 So one of the techniques that people could use to help think about themselves and their customer, maybe something that they probably should have done, but maybe this is a wake-up call, like a customer journey map. 00:06:42.829 --> 00:06:48.670 Did you find any examples of companies doing that to reimagining how they relate to their customers? 00:06:49.069 --> 00:06:55.469 Well, I I think one of the things that Nestle did in particular, they started trying to understand customers. 00:06:55.469 --> 00:07:04.269 They're historically have been very keen on doing a lot of customer research to understand what customers are thinking about, what are their problems. 00:07:04.269 --> 00:07:07.310 And what they did is they used a company called outset.ai. 00:07:07.310 --> 00:07:11.149 And they use AI to conduct these consumer interviews. 00:07:11.149 --> 00:07:19.550 And what they found is that they could go super deep because in research, you either had to trade off quality versus quantity. 00:07:19.550 --> 00:07:28.670 So you could do qualitative research that was highly bespoke, highly informative, but only to a small number of people in the tens, maybe. 00:07:28.670 --> 00:07:32.670 Or you could do huge quantity to get statistically reliable results. 00:07:32.670 --> 00:07:34.990 But with AI, they could do both. 00:07:34.990 --> 00:07:36.670 They could actually do both. 00:07:36.670 --> 00:07:39.230 And that completely changes the game. 00:07:39.230 --> 00:07:49.550 So the types of questions that you would even begin to ask about the customer journey, about what is a consideration, that consideration black hole. 00:07:49.550 --> 00:07:50.189 How does it work? 00:07:50.189 --> 00:07:51.629 How do you think about it? 00:07:51.629 --> 00:08:02.589 And because they could do it at scale, but also in a qualitative interview-based style, it completely changed the way they even thought about how to do customer research. 00:08:02.829 --> 00:08:06.829 So is it still a human asking the questions, or is that the AI asking the questions? 00:08:07.069 --> 00:08:09.629 No, it is an AI asking the questions. 00:08:09.629 --> 00:08:16.189 But you see a screen on your computer or on your phone, and it begins by asking some questions like what are your biggest issues? 00:08:16.189 --> 00:08:21.629 But it's trained to follow these trails to say, well, tell me more about that problem. 00:08:21.629 --> 00:08:26.829 Tell me, like that seems like an interesting thing, or to pick up nuances of the things that we're saying. 00:08:26.829 --> 00:08:28.269 You seem uncertain about that. 00:08:28.269 --> 00:08:30.670 Tell me more about why you're not so sure about this. 00:08:30.670 --> 00:08:51.070 And so the nuances that you can pick up, and then doing it consistently across thousands of people gives you the statistical wherewithal to say this percentage of people actually were picked up on this particular trend because you don't know what questions they ask at the beginning of that research. 00:08:51.070 --> 00:08:57.629 But with the open-ended responses that you can give to people, and also very importantly, people respond by speaking. 00:08:57.629 --> 00:08:59.230 They don't respond by typing. 00:08:59.230 --> 00:09:04.909 They literally hit a button on the microphone and they say they respond after reading something. 00:09:04.909 --> 00:09:07.550 And they can respond in any language too, as well. 00:09:07.550 --> 00:09:09.230 And the AI understands them. 00:09:09.230 --> 00:09:14.269 So it's fantastic for research done in a way that's scalable. 00:09:14.269 --> 00:09:22.990 And they understand it's it's an AI, obviously, asking the questions, but it feels like the AI is really listening to them and not just following a script. 00:09:23.470 --> 00:09:29.470 Well, it's fascinating to see how people responded so positively at a really early stage in this technology. 00:09:29.470 --> 00:09:31.550 So you know, that's an interesting finding. 00:09:31.550 --> 00:09:35.550 Great for companies trying to find out what the bottlenecks are when they're serving customers. 00:09:35.550 --> 00:09:39.390 I'm sure Nestle benefit benefited from that. 00:09:39.390 --> 00:09:50.590 But you're a CEO, you've now discovered that you know that you've got some opportunities to enhance or even change your business model and you've got some ideas about what to do. 00:09:50.590 --> 00:09:55.390 Who should be in your transformation team to make all of this happen? 00:09:55.710 --> 00:09:59.790 Well, I absolutely believe we believe that the CEO should be involved. 00:09:59.790 --> 00:10:02.030 And that's the

last thing that the CEO wants to hear. 00:10:02.030 --> 00:10:04.350 It's like, I can I just delegate this to somebody? 00:10:04.350 --> 00:10:12.190 And the reason is because if it is strategic and it is core to your business strategy, you have to be involved. 00:10:12.190 --> 00:10:20.270 Now, you don't have to be involved day to day, but you need to know what's going on because the CEO is where the book stops. 00:10:20.270 --> 00:10:24.750 That is a person who decides go or no-go on strategic initiatives. 00:10:24.750 --> 00:10:37.870 They're also the ones that are central to making sure that it gets implemented, that breaks down all the departmental silos that inevitably come up, that gets people to understand this is the bigger picture, this is why we have to do things. 00:10:37.870 --> 00:10:41.150 So the CEO involvement is absolutely crucial. 00:10:41.150 --> 00:10:44.590 Again, it doesn't have to be full-time on their part. 00:10:44.590 --> 00:10:46.270 They just have to know what's going on. 00:10:46.270 --> 00:10:51.070 You definitely need somebody who is digital or AI, somebody who understands the technology. 00:10:51.070 --> 00:10:54.670 And I would caution that this is not typically your CIO. 00:10:54.670 --> 00:11:07.470 Again, many, many CIOs are fantastic at what they do, but they are uh really administrating technologies, they are maintaining security, all very important jobs. 00:11:07.470 --> 00:11:10.750 They're not necessarily creating and transforming businesses. 00:11:10.750 --> 00:11:15.150 There are few CIOs out there who are transformational CIOs. 00:11:15.150 --> 00:11:20.030 If you are so lucky to have one of those people, you are fantastic. 00:11:20.030 --> 00:11:25.710 Some people also bring in somebody like an AI expert that has its own tensions. 00:11:25.710 --> 00:11:31.070 So the the again, that technology person is extremely important to have, but to have the right person. 00:11:31.070 --> 00:11:34.670 You need to have somebody who understands the customer, to your point. 00:11:34.670 --> 00:11:37.310 How is this going to impact our customers? 00:11:37.310 --> 00:11:39.710 How is that going to change the experience? 00:11:39.710 --> 00:11:48.110 Somebody who understands the business, the um, again, the revenues and profits, not necessarily financial, but who understands the business side of it. 00:11:48.110 --> 00:11:50.910 Very importantly, somebody in HR. 00:11:50.910 --> 00:12:00.910 This is the overlooked position because if you are going to transform the way people work, you need somebody at the table who is absolutely crucial to that. 00:12:00.910 --> 00:12:15.070 And again, this requires somebody who's stepping up and thinking about HR beyond the administrative and the risk mitigation aspects of HR, who looks at it strategically and understands the impact that AI will have on the future of work. 00:12:15.070 --> 00:12:28.110 Finally, you need somebody on the communications side who can help you figure out

how to align all the different points of view into a consistent message, both internally and externally. 00:12:28.110 --> 00:12:39.870 So that's your small team, but eventually you will need to have one person, one person who is responsible and accountable for driving AI adoption and also adaptation of the organization. 00:12:39.870 --> 00:12:46.030 One person whose job every day is to wake up and going, how are you using AI strategically in our organization? 00:12:46.590 --> 00:12:51.230 I mean, you're absolutely right to just pick up on your comment about external customers. 00:12:51.230 --> 00:13:00.270 I might also add internal customers because one of the issues that I've come up when talking to clients is, you know, the language of sales is different to the language of supply chain. 00:13:00.270 --> 00:13:14.910 And if you have IT people coming in who are more supply chain-oriented, then you get disconnects because the flow of value isn't properly managed in the organization because different teams are talked in different languages. 00:13:15.390 --> 00:13:19.870 Yeah, most change management efforts are focused on departments. 00:13:19.870 --> 00:13:22.990 So it's it's it's one thing to do change at that level. 00:13:22.990 --> 00:13:28.030 When you're talking about this type of transformation, it's a very different type of management. 00:13:28.030 --> 00:13:32.350 I think of it as transformation management, which is very different than change management. 00:13:32.350 --> 00:13:38.750 And if you approach it with the traditional change management toolbox that you have, it's not going to work. 00:13:38.750 --> 00:13:48.750 Because it requires working across all these departmental silos that you have, norming against the higher order strategic goals that you have as an organization. 00:13:48.750 --> 00:13:53.629 And when you remind people like this is what we're trying to do, the people typically say, Oh, yeah, that's right. 00:13:53.629 --> 00:13:57.390 They sort of forgot that, oh yeah, we're here to do that, right? 00:13:57.390 --> 00:14:00.030 Not to just produce marketing materials. 00:14:00.430 --> 00:14:03.950 I mean, if you're trying to change your business model, you're going to change your business model. 00:14:03.950 --> 00:14:08.350 That's a that's your flow of value between you and the customer and everyone in between. 00:14:08.350 --> 00:14:10.910 It's it's a non-trivial activity. 00:14:10.910 --> 00:14:14.510 And you also picked up on you know the human factor of change management. 00:14:14.510 --> 00:14:20.590 And I loved a couple of examples that you gave about the sequencing of projects in organizations. 00:14:20.590 --> 00:14:32.590 I can just came back to another popular podcast we had was with Hillary Scarlett, who talked about our default as human beings is to see change as threats. 00:14:32.590 --> 00:14:38.350 I mean, that's what we were two million years ago on the savannah, seeing yellow things running across the plains. 00:14:38.350 --> 00:14:39.390 We see default that. 00:14:39.390 --> 00:14:45.470 And there was a lovely case study by Paul Bezick at Marsh and the way he organized the work. 00:14:45.470 --> 00:14:47.150 Can you tell me about that? 00:14:47.790 --> 00:14:59.790 Yes, one of the things that we found is that the biggest resistance to AI isn't to the technology, it's because of the fear, and in particular, our need to protect our identity and our sense of purpose. 00:14:59.790 --> 00:15:10.110 And so I think Hillary's frameworks really talk about how when your brain senses danger approaching, it all shuts down and it's all about survivor mode. 00:15:10.110 --> 00:15:15.150 And so what Paul Beswick understood is that he made it just low stakes. 00:15:15.150 --> 00:15:18.910 He has 95,000 people worldwide in the organization. 00:15:18.910 --> 00:15:21.790 And he didn't mandate the use of AI. 00:15:21.790 --> 00:15:25.790 He didn't frame it as something that was transformational. 00:15:25.790 --> 00:15:27.710 He just said, it's here, it's a cool tool. 00:15:27.710 --> 00:15:28.910 It's not great for everything. 00:15:28.910 --> 00:15:29.870 Just use it. 00:15:29.870 --> 00:15:39.950 And he made it so that it was easily accessible, but he also put a lot of fences around how you could use it so responsible and ethical AI, put training in place. 00:15:39.950 --> 00:15:43.550 So the key part is that this came in with no threat. 00:15:43.550 --> 00:15:45.710 It came in as something that would be useful. 00:15:45.710 --> 00:15:46.830 If you picked it up, great. 00:15:46.830 --> 00:15:48.750 If you didn't use it, that's fine. 00:15:48.750 --> 00:15:57.550 And that's what the ideal situation was, because then what happened was over 300 employee-built AI applications were created in the first year. 00:15:57.550 --> 00:16:02.750 25 million requests, over 1 million hours saved. 00:16:02.750 --> 00:16:07.550 And they spent just a couple hundred thousand dollars in that first year to enable this. 00:16:07.550 --> 00:16:11.150 So they didn't have to worry about ROI of AI because it was so low cost. 00:16:11.150 --> 00:16:14.110 He just paid for it out of the extra budget he had in his thing. 00:16:14.110 --> 00:16:16.430 He's just like, I can get an enterprise version of this. 00:16:16.430 --> 00:16:22.830 And he he literally wrote the interface for it, created the interface over it in one night and just on his own because he could do that. 00:16:22.830 --> 00:16:27.470 And then he just kind of deployed it, made sure there were security parameters around it, and just gave it to everybody. 00:16:27.470 --> 00:16:28.510 Go ahead, use it. 00:16:28.510 --> 00:16:30.750 He just took the stakes down. 00:16:30.750 --> 00:16:40.190 And when people around you are using it and saying, this is the greatest thing since sliced bread, you believe them because you know Joe, you know Mary, you trust them. 00:16:40.190 --> 00:16:43.310 And then you want to look over their

shoulder and say, what are you doing? 00:16:43.310 --> 00:16:48.190 So he did this from a top-down, but also very bottom-up organic way. 00:16:48.190 --> 00:16:51.629 And I think another example is from Ally Bank. 00:16:51.629 --> 00:16:57.710 They understood right from the very beginning that customer service agents were going to be rebellious against this. 00:16:57.710 --> 00:16:59.390 They could see the writing on the wall. 00:16:59.390 --> 00:17:01.070 This is coming from my job. 00:17:01.070 --> 00:17:07.470 And they, they, they, they did training right from the very beginning to address those concerns. 00:17:07.470 --> 00:17:17.710 They went right after the most skeptical agents that they had in the call centers and gave them this experience that was very personal to show them how value was being created. 00:17:17.710 --> 00:17:23.309 And so when we talk about training for people, we don't talk about just get in there and show them how it works. 00:17:23.309 --> 00:17:26.670 The first thing you have to do is address what are you afraid of? 00:17:26.670 --> 00:17:28.269 What are you concerned about? 00:17:28.269 --> 00:17:34.670 And because unless people, unless you can address those fears and concerns, like, is it coming from my job? 00:17:34.670 --> 00:17:36.349 Can it be more creative than me? 00:17:36.349 --> 00:17:42.190 I've defined my entire career from being able to write beautiful prose and content for my marketing. 00:17:42.190 --> 00:17:44.750 Now you're telling me AI can do that for me? 00:17:44.750 --> 00:17:45.390 No. 00:17:45.390 --> 00:17:51.149 That that puts the value that I create as a person completely, it liberates that. 00:17:51.149 --> 00:17:53.629 So I'm not signing on for that. 00:17:53.629 --> 00:18:05.950 Versus if you say AI creates the first drafts, it creates multiple approaches, and then we tap into your expertise to say which of these are the best ones to use, or maybe combine them into a completely different approach. 00:18:05.950 --> 00:18:10.029 Your value now isn't to take a blank page and create value from that. 00:18:10.029 --> 00:18:18.829 Your now value is from taking all of these resources in beam to make sense of them and apply your judgment and expertise against that. 00:18:18.829 --> 00:18:24.509 So it's it requires a bit of unlearning on our part to be able to learn how to use AI. 00:18:25.149 --> 00:18:31.789 What you're saying just resonates with another data point we had from Matt Topalter when he was the author of Governing the Machine. 00:18:31.789 --> 00:18:44.269 And they went from low single digits or 15% confidence in AI up to 18, 90% confidence just by giving people a couple of hours training before the transformation happened. 00:18:44.269 --> 00:18:47.309 And you know, obviously the speed was so much more successful with that. 00:18:47.629 --> 00:18:51.069 But one thing I want to say is when I ask people, like, how come you're not doing training? 00:18:51.069 --> 00:18:53.230 He goes,

Well, our plates are completely full. 00:18:53.230 --> 00:18:54.669 We have no time for training. 00:18:54.669 --> 00:18:58.990 Or how why don't you give people some hour a few hours here and there to experiment? 00:18:58.990 --> 00:19:00.589 No, their plates are completely full. 00:19:00.589 --> 00:19:02.669 I'm like, do you really truly believe that? 00:19:02.669 --> 00:19:03.629 First of all. 00:19:03.629 --> 00:19:20.990 And second of all, if this is strategic again, not some frivolous AI thing, but if it's strategic to your future as an organization, also if it's strategic to you as an employee to gain these skills, you're absolutely going to invest the time in doing this. 00:19:20.990 --> 00:19:28.750 And just as you were saying before, that the the company found that they got so much more from just a few hours. 00:19:28.750 --> 00:19:31.710 We're not talking about taking two weeks off. 00:19:31.710 --> 00:19:35.069 We're talking about, I've got this task here. 00:19:35.069 --> 00:19:40.349 I wonder how I could use AI to make that task easier, better, faster. 00:19:40.349 --> 00:19:42.269 And let me experiment with that. 00:19:42.269 --> 00:19:45.869 I might fail, but I might learn something too along the way. 00:19:45.869 --> 00:19:54.669 And then, very importantly, you share that with the person sitting next to you or the person sitting across the organization who is doing something similar and you build on that learning. 00:19:54.669 --> 00:19:58.269 So that is where I think the investment needs to come from. 00:19:58.269 --> 00:20:03.549 And the person who needs to do that most and demonstrate that most is the CEO. 00:20:03.549 --> 00:20:16.429 If they see the CEO learning, failing, experimenting, figuring it out, not having all the answers, it gives them the knowledge that, oh, I can do this too as well. 00:20:16.429 --> 00:20:23.389 It is the most powerful signal you can send as a CEO to show people how you are learning with AI too as well. 00:20:24.269 --> 00:20:24.909 So, right. 00:20:24.909 --> 00:20:33.549 This the concept of psychological safety in the organization is being one of the essential ingredients in all of these transformations. 00:20:33.549 --> 00:20:40.990 And in our journey so far, we've come across and referenced several times something called the playground paradox. 00:20:40.990 --> 00:20:45.629 That in my research, if I'm right, this may have actually originated with yourself, Charlene. 00:20:46.109 --> 00:20:47.069 I think so. 00:20:47.069 --> 00:20:50.189 I've been using it for quite a few years. 00:20:54.669 --> 00:20:56.029 We've talked about this already. 00:20:56.029 --> 00:21:04.109 Yeah, can we just remind yourselves why this is a relevant thing to think about in creating psychological safety in organizations? 00:21:05.230 --> 00:21:13.470 It's some of the research that's out there that talks about how when you put more constraints around people, it doesn't necessarily slow them down. 00:21:13.470 --> 00:21:18.829 In fact, constraints are necessary so that people know what they can do and what they cannot do. 00:21:18.829 --> 00:21:25.789 And when you have those guardrails in place, then you can go up to the very edges of those guardrails, go up to the limits. 00:21:25.789 --> 00:21:32.829 So the paradox of the playground is that when you constrain where people can play, guess where people are? 00:21:32.829 --> 00:21:38.189 They're right up against the fence because they know they can go right to the edges of that fence and be safe. 00:21:38.189 --> 00:21:46.509 But what's really important is that you also have a gate to that playground, that you'd be able to come into it and to play, that it's welcoming and it's important. 00:21:46.509 --> 00:22:00.269 So so much of AI governance, so much of governance in general is about what you can't do versus saying, do all of this, just stay within these guardrails, but go, go, go as fast and far as you can up to these guardrails. 00:22:00.269 --> 00:22:09.470 And what I particularly like is when organizations really double down on this and they go together as a team to the very edge. 00:22:09.470 --> 00:22:14.909 And it's scary there, they don't know what they're going to find, but they know they're safe. 00:22:14.909 --> 00:22:18.669 And in particular, if they should stumble and fall, they're there together. 00:22:18.669 --> 00:22:21.470 And they're there to pick each other and support each other. 00:22:21.470 --> 00:22:23.230 And to just yeah, do it. 00:22:23.629 --> 00:22:38.909 Yeah, and but the point is it's important for the leader just to define where the edge of the playground is, because the reality is if there isn't a fence around the playground, children taste tend to spend most of their time on the swings and the roundabouts and not venture outside. 00:22:38.909 --> 00:22:46.669 And so it's not that people are going to venture off, it's people often put their self imposed constraints, which are much narrower than the organization really. 00:22:46.669 --> 00:22:48.029 Wants them to go. 00:22:48.029 --> 00:22:51.230 And again, leadership is key in defining that stat. 00:22:51.710 --> 00:22:54.509 Because you don't know how far you can go before it's unsafe. 00:22:54.589 --> 00:22:54.829 Yeah. 00:22:55.069 --> 00:23:02.750 So it again, we are trained as good corporate citizens to not put ourselves and our companies at risk. 00:23:02.750 --> 00:23:07.629 We do not want to end up on the front page of some website, news website. 00:23:07.629 --> 00:23:09.629 So we play it safe. 00:23:09.629 --> 00:23:11.230 And safe is slow. 00:23:11.230 --> 00:23:15.869 So if you again, I we believe that speed is the new moat. 00:23:15.869 --> 00:23:20.429 And you it there's the only differentiation because everybody has access to the same tools. 00:23:20.429 --> 00:23:23.230 Everybody has reams of data inside the organizations. 00:23:23.230 --> 00:23:26.349 They all understand their customers potentially at the

same levels. 00:23:26.349 --> 00:23:31.869 And so the only differentiation that's going to make a difference in the future is speed. 00:23:31.869 --> 00:23:36.909 Speed to change and adapt and transform your organization to create value. 00:23:37.869 --> 00:23:44.669 But of course, one of the big pushbacks you get is but we can't start just yet because our data is dirty. 00:23:44.669 --> 00:23:46.269 We're not ready to go. 00:23:46.269 --> 00:23:50.750 So give me three months to clean the data and then we'll have another discussion. 00:23:50.750 --> 00:23:52.750 What would you say, Charlene, to that? 00:23:53.069 --> 00:23:53.789 It's a myth. 00:23:53.789 --> 00:23:56.829 It's a that you can ever have clean data. 00:23:56.829 --> 00:23:58.589 And so we hear this all the time. 00:23:58.589 --> 00:24:01.309 You can't start AI because our data isn't ready. 00:24:01.309 --> 00:24:05.149 And so it and the truth is your data will never be ready. 00:24:05.149 --> 00:24:06.589 It will never be perfect. 00:24:06.589 --> 00:24:13.149 So waiting for that day to come is you ceding your competitive advantage to other people who are willing to start with messy data. 00:24:13.149 --> 00:24:17.389 The irony here is that AI can actually help you clean up your messy data. 00:24:17.389 --> 00:24:22.909 And so we think about it like the analogy is uh flying a plane while cleaning it. 00:24:22.909 --> 00:24:35.710 So one thing that I think is so important is to define what's the minimum data quality you need to deliver value against your highest priority AI applications. 00:24:35.710 --> 00:24:47.069 If you identify that these are the top ways that we can use AI to deliver value against our strategic objectives, then what's the minimum data that you need to have ready? 00:24:47.069 --> 00:24:53.789 What's the minimum quality, quantity of data that you need to have to start executing on that AI application? 00:24:53.789 --> 00:24:56.669 So not perfect, but just minimum. 00:24:56.669 --> 00:24:58.750 So it's a very different order. 00:24:58.750 --> 00:25:03.869 And if you look at it that way, you go, oh, I have enough here now to actually begin working on it. 00:25:03.869 --> 00:25:07.789 So I can actually test some of the other things I have to do. 00:25:07.789 --> 00:25:13.230 Just connecting to your legacy data is a chore and of itself. 00:25:13.230 --> 00:25:18.990 And again, AI can oftentimes take it out, be able to do things with it, put it someplace else. 00:25:18.990 --> 00:25:26.429 There's so many things, but begin using your data against high value applications in your organization. 00:25:26.429 --> 00:25:30.909 Don't pick the easiest, pick the most high value ones that are worth doing. 00:25:30.909 --> 00:25:36.189 So many organizations go, well, we need to do a feasibility study, or we need to do a readiness assessment. 00:25:36.189 --> 00:25:38.189 And we thought long and hard about this. 00:25:38.189 --> 00:25:42.269 We created an entire readiness assessment and we threw it out the window. 00:25:42.589 --> 00:25:43.230 Really? 00:25:44.109 --> 00:25:45.629 Speed is a new moat. 00:25:45.629 --> 00:25:48.349 What are you going to find out in the readiness assessment? 00:25:48.349 --> 00:25:49.950 That you're not ready. 00:25:49.950 --> 00:25:52.829 What are you going to find out in the feasibility assessment? 00:25:52.829 --> 00:25:54.189 That it's not feasible. 00:25:54.189 --> 00:25:56.429 You don't have the skills and capabilities. 00:25:56.429 --> 00:26:02.669 But if it is strategic to your organization, you will find a way. 00:26:02.669 --> 00:26:09.230 You will find the resources, you will make the investments, you will hire the people or hire the consultants, you will buy the technology. 00:26:09.230 --> 00:26:13.950 If it is strategically important to you, you will find a way to make it happen. 00:26:13.950 --> 00:26:20.189 Therefore, feasibility and readiness are not a consideration if it is strategic. 00:26:20.750 --> 00:26:26.669 So I, you know, I'm loving this talk about speed as an emote and just use AI to help you clean the data. 00:26:26.669 --> 00:26:30.429 It's the solution, not the problem in many ways. 00:26:30.429 --> 00:26:38.109 Another thing about speed is companies are still stuck in the 12-month annual budget mentality. 00:26:38.109 --> 00:26:41.950 One of the podcasts I'd love to have is the author of Beyond Budgeting. 00:26:41.950 --> 00:26:46.349 But I think you've got a view on this as well that you may want to share. 00:26:46.909 --> 00:26:47.309 Yes. 00:26:47.309 --> 00:26:57.309 And in particular, I feel that this linear traditional budgeting, it was designed for a world that doesn't exist anymore. 00:26:57.309 --> 00:27:02.349 You would plan in the fall to start executing January for what will happen in December. 00:27:02.349 --> 00:27:04.029 And AI just doesn't work that way. 00:27:04.029 --> 00:27:08.589 It is a technology, there is the competitive landscape, everything is changing. 00:27:08.589 --> 00:27:18.189 Just in the past two months, we've had cloud co-work come on board and then open claw, which is completely blown apart the way that Agentic AI works inside of organizations. 00:27:18.189 --> 00:27:24.349 So it's just showing how quickly things can change within just a few weeks. 00:27:24.349 --> 00:27:38.509 So what I what we really advocate is a quarterly approach, a six-quarter walk over 18 months, in that every quarter you are evaluating how are we doing against our plan to create value for the organization with AI. 00:27:38.509 --> 00:27:40.589 And it resets every quarter. 00:27:40.589 --> 00:27:41.710 And it sounds insane. 00:27:41.710 --> 00:27:43.869 Like, how do you reset a strategy? 00:27:43.869 --> 00:27:50.589 Well, it's because you know what your strategic objectives are, but you're changing is your roadmap of how you're going to get there. 00:27:50.589 --> 00:28:01.950 And it's taking you into account what has changed in terms of your

capabilities, what has changed in terms of the technology available, and also what are the competitive aspects happening. 00:28:01.950 --> 00:28:10.349 So every quarter, it's a time where you can reflect, you can revise your plan, you can extend the horizon to another quarter. 00:28:10.349 --> 00:28:17.789 So it's a rolling 18-month plan, and then you communicate it to everybody to make sure everyone understands what this is what the new plan looks like. 00:28:17.789 --> 00:28:31.470 I talked to one organization, uh and it's McKinsey, and they said, Oh, we are constantly rejuring a roadmap almost on a daily basis, because we're kind of we're fine-tuning it and it's written in pencil, not carved in stone. 00:28:31.470 --> 00:28:41.149 So this idea that you can set off a strategy and then come have it completely set and know exactly where you're gonna end up, it's just does not work. 00:28:41.869 --> 00:28:44.269 You are gonna find one of your biggest supporters here. 00:28:44.269 --> 00:28:48.990 I mean, it connects with so much of what we think about in terms of integrated business planning. 00:28:48.990 --> 00:28:55.149 And by goodness me, if we don't have the technology now to update our plans on a quarterly basis, we'll never have it. 00:28:55.149 --> 00:28:56.589 I mean, it's not that difficult. 00:28:56.589 --> 00:29:01.149 And surely this is again where AI can help upgrade our planning on that. 00:29:01.149 --> 00:29:08.269 But right at the beginning, you talked about how HR has to revolutionize and just think about it being different ones. 00:29:08.269 --> 00:29:12.109 And especially how man works with the machine. 00:29:12.109 --> 00:29:18.109 We talked about homofoba and you know, this what is it that we have to be in in the workplace in the future. 00:29:18.109 --> 00:29:21.950 Then you gave a case study in the book about Modena, the vaccine company. 00:29:21.950 --> 00:29:23.869 Can you tell me something about that? 00:29:23.869 --> 00:29:25.549 Because I was fascinated. 00:29:25.950 --> 00:29:34.750 Yeah, so what Moderna did is they took their CHRO, Tracy Franklin, and made her in charge of all of AI. 00:29:34.750 --> 00:29:38.750 So she they took their entire technology group and put it under HR. 00:29:38.750 --> 00:29:40.269 That just doesn't happen. 00:29:40.269 --> 00:29:42.029 People don't do those kinds of things. 00:29:42.029 --> 00:29:47.470 And the reason why is they understood that transformation is fundamentally about people, not the systems. 00:29:47.470 --> 00:29:52.990 Technology is the enabler, and the people strategy is at the central core of this. 00:29:52.990 --> 00:30:05.069 And so what she did was instead of HR managing headcount and org chart, she's now orchestrating how work flows between humans and AI systems. 00:30:05.069 --> 00:30:13.149 And we found that to be fascinating and very much in keeping with this whole idea that we have at the end of the book about superhumans. 00:30:13.149 --> 00:30:33.149 It is this integrated intelligence that we develop, where we use the best of AI, not to become more AI ourselves, but to become more deeply human ourselves, to spend more time on things like empathy and self-reflection and creative problem solving and wisdom and judgment. 00:30:33.149 --> 00:30:36.269 These are things that are uniquely human. 00:30:36.269 --> 00:30:40.669 And AI handles a computational, repetitive administrative loads. 00:30:40.669 --> 00:30:50.349 But it's the merger of these two things that really becomes uh this workforce of the future, the superhuman, that enables everything. 00:30:50.349 --> 00:30:55.470 And so there's a lot of news right now about the decimation of the white-collar workforce. 00:30:55.470 --> 00:31:00.909 And I think that the biggest assumption that people have is everything's going to stay the same. 00:31:00.909 --> 00:31:05.549 And we're just going to use AI to make what we do today more efficient. 00:31:05.549 --> 00:31:10.750 What they don't realize is that the human nature is to always try to improve. 00:31:10.750 --> 00:31:18.109 And that inevitably we're going to say is how do we use AI to create something new, to do things in a different way. 00:31:18.109 --> 00:31:27.869 And so we don't necessarily count into those numbers, the scenario that we also, while cutting some things, are also building new things. 00:31:27.869 --> 00:31:33.789 So I'm not I'm not saying that there's not going to be wholesale change in organizations. 00:31:33.789 --> 00:31:41.710 There's going to be a lot of transformations and technology change has always resulted in jobs being created and jobs being lost. 00:31:41.710 --> 00:31:46.669 And the jobs that are being created aren't necessarily filled by the people who are losing their jobs. 00:31:46.669 --> 00:31:56.429 So that is a very real human aspect and consequence of AI transformation that we're going to have to deal with as a society and our governments and our culture. 00:31:56.429 --> 00:31:58.509 We have to think about this holistically. 00:31:58.509 --> 00:32:08.750 But I do believe inside the organizations, we have the opportunity to develop superhumans and really actively think about ways to connect the human and AI together. 00:32:09.149 --> 00:32:12.429 And yeah, this is a theme we picked up quite often in the podcast. 00:32:12.429 --> 00:32:15.149 It's, you know, about us becoming more human. 00:32:15.149 --> 00:32:30.909 But I do think we need the lighthouse case studies from Moderna and otherwise to point the way about what that means, really, to avoid the anxiety and give us a sense of positivity about us being better, better human beings in the workplace, potentially. 00:32:31.230 --> 00:32:31.309 Yeah. 00:32:31.309 --> 00:32:32.509 I'll give you an example. 00:32:32.509 --> 00:32:37.149 I recently heard from IBM that they are doubling down on entry-level hires. 00:32:37.149 --> 00:32:39.389 And that goes against what everyone is saying. 00:32:39.389 -->

00:32:44.429 They're saying we actually are hiring people straight out of college into coding positions. 00:32:44.429 --> 00:32:48.349 And but we're also rewriting those job descriptions. 00:32:48.349 --> 00:32:58.829 So we're assuming that they know how to use Claude Code, that they are coming in using AI, that they're not writing, they're not doing the same jobs that a year ago entry-level people did. 00:32:58.829 --> 00:33:06.909 So we're redefining what these job roles are, but we need that fresh energy, that uh that perspective. 00:33:06.909 --> 00:33:08.589 And also people early in the career. 00:33:08.589 --> 00:33:12.349 So we could, again, people go to IBM and stay there for their entire careers. 00:33:12.349 --> 00:33:16.669 We need that pipeline to continually be refreshed. 00:33:16.669 --> 00:33:23.309 So char is, again, as I mentioned before, no longer just about recruitment and risk mitigation. 00:33:23.309 --> 00:33:30.909 It is really about redefining jobs, looking at career paths and deciding, designing them in a different way. 00:33:30.909 --> 00:33:33.710 It's performance metrics in a very different way. 00:33:33.710 --> 00:33:40.029 Instead of just output, it's also about the quality of the work and are you aligning that with your strategic objectives? 00:33:40.029 --> 00:33:42.990 So it's a very different world that we're trying to build here together. 00:33:43.389 --> 00:33:45.789 I mean, it's innovation in the workplace, is what you're saying. 00:33:45.789 --> 00:33:48.269 We're going to have to rethink what we're doing and redesign. 00:33:48.269 --> 00:33:51.869 I hope we haven't got too many people anxious. 00:33:51.869 --> 00:34:00.829 But for anyone who has listened to this and is thinking, maybe I'm a little bit late to the party, and seems to know what she's talking about much more than I thought she might be. 00:34:00.829 --> 00:34:03.869 What bits of advice would you give to someone? 00:34:04.189 --> 00:34:05.230 First of all, you're not too late. 00:34:05.230 --> 00:34:06.669 Too late is the wrong frame. 00:34:06.669 --> 00:34:12.670 The thing that I think is so clear is that AI adoption does not vary by age or role. 00:34:12.670 --> 00:34:14.349 It varies by disposition. 00:34:14.349 --> 00:34:19.230 Specifically, how open are you to change and how comfortable are you with uncertainty? 00:34:19.230 --> 00:34:23.309 And if you're asking these questions, your disposition is already right. 00:34:23.309 --> 00:34:26.670 You already have the curiosity to say, what do I need to do? 00:34:26.670 --> 00:34:31.070 And the second thing I would say, people constantly ask me, so how do I catch up? 00:34:31.070 --> 00:34:33.070 What are the courses I can take? 00:34:33.070 --> 00:34:36.269 What are the newsletters and the podcasts I should listen to? 00:34:36.269 --> 00:34:39.470 And I go, that is not the right way to approach it. 00:34:39.470 --> 00:34:44.990 If you feel overwhelmed by how much is out there, you're trying to

solve AI as a problem. 00:34:44.990 --> 00:34:46.590 That's not the problem. 00:34:46.590 --> 00:34:56.990 The problem is your business cycles, your sales cycles are too long, your customer churn is too high, your product launch timelines are too slow for launch. 00:34:56.990 --> 00:35:01.070 Pick one of those biggest business pains that you have. 00:35:01.070 --> 00:35:05.710 Then ask, because you know that business pain extremely well. 00:35:05.710 --> 00:35:10.670 Now, what you have to ask is how does AI help us address that problem? 00:35:10.670 --> 00:35:14.590 And that's what that's why the book is focused on that particular problem. 00:35:14.590 --> 00:35:16.830 90 days you can figure that out easily. 00:35:16.830 --> 00:35:21.470 90 days you can figure out what is my biggest problem, how can I use AI to support it. 00:35:21.470 --> 00:35:24.590 And then what are the building blocks I need to have to be able to execute on. 00:35:24.590 --> 00:35:30.830 And you'll be in such a better place if you just focus on your biggest business problems and address them with AI. 00:35:30.830 --> 00:35:39.710 So when you don't need to go and take tons of courses, and here's the meta aspect of this: you can use AI to tell you how to use AI to solve that problem. 00:35:39.950 --> 00:35:42.670 So that's the best part about all of this. 00:35:43.070 --> 00:35:43.789 Thank you, Charlene. 00:35:43.789 --> 00:35:45.390 That's been a great full circle. 00:35:45.390 --> 00:35:49.150 Folks, the message is it's your business strategy that's your problem. 00:35:49.150 --> 00:35:51.150 It was and it is and it will be. 00:35:51.150 --> 00:35:55.470 And that's what the message I think, Charlene, you wanted us all to hear from this. 00:35:55.470 --> 00:35:58.910 And I've got it, and I hope others can take it that away from that. 00:35:58.910 --> 00:36:00.510 Thank you so much for your time. 00:36:00.510 --> 00:36:02.750 Now, this book's coming out. 00:36:02.750 --> 00:36:06.670 Tell me what launch date, where we can get it, and and so forth. 00:36:07.309 --> 00:36:14.350 It's coming out on March 24th, and you can get everything you need about it at winningwithaibook.com. 00:36:14.350 --> 00:36:19.070 We'll have pre-order links and everything on there, so you can get all the information. 00:36:19.070 --> 00:36:23.470 And in the meantime, you can sign up to be notified when the ordering is available. 00:36:23.470 --> 00:36:25.870 So winningwithaibook.com. 00:36:26.269 --> 00:36:31.230 And you're gonna keep it light and fresh, and there'll be another book, I'm sure, in due course, Charlene. 00:36:31.230 --> 00:36:31.950 You're very prolific. 00:36:32.269 --> 00:36:39.550 Yeah, one of the most interesting things we're doing is it's a book that's print on demand, so we can update it with audio and print and things like that. 00:36:39.550 --> 00:36:42.269 Because people are like, isn't it insane to write a book about AI? 00:36:42.269 --> 00:36:45.309 It's gonna be out in data as soon as it's printed, and we realize that. 00:36:45.309 --> 00:36:47.870 So because it's print on demand, we can update it. 00:36:47.870 --> 00:36:50.590 And with audio in particular, we can update it very, very easily. 00:36:50.990 --> 00:36:53.470 Get new case studies coming in and evolve it. 00:36:53.470 --> 00:36:56.430 That's gonna be such a useful resource for everyone. 00:36:56.430 --> 00:36:58.110 Is it gonna be an evergreen document? 00:36:58.110 --> 00:36:59.390 Thank you so much, Charlene. 00:36:59.390 --> 00:37:04.269 Well, now I've read it before, and I would but I'd like to see what what updates become in the future. 00:37:04.269 --> 00:37:05.230 Thank you very much. 00:37:05.550 --> 00:37:06.110 Thank you. 00:37:06.110 --> 00:37:06.830 Bye bye.