What we talked about
Charlene Li joins Federico Ramallo for a special 100th-episode conversation on disruptive leadership, business transformation, and her upcoming book Winning with AI. Charlene shares how she “fell into” being an analyst and author:starting at Forrester in 1999 after stepping back from running businesses to focus on family:then building a career at the front edge of major tech disruptions: the internet, search, social media, and now generative AI.
Show notes
Charlene Li dropped the book she was working on the moment ChatGPT launched, because she recognized it immediately as the same kind of interface breakthrough that she had watched transform the internet, social media, and search, and she knew that whoever understood it first would have a real advantage. That pattern-recognition instinct, sharpened across twenty-five years of watching disruption unfold, is what shapes her framework for winning with AI.
What we covered
- “Winning” is not a universal definition, it means whatever your organization’s definition of success actually is, whether that’s beating competitors, serving customers better, or achieving something specific to your mission as a school, hospital, bank, or nonprofit. The critical first step is knowing what winning means before asking what AI should do.
- Speed is the new moat. When everyone has access to the same technology and data, the only sustainable competitive advantage is how fast an organization can adopt AI and, more importantly, adapt its people, culture, and processes to the changes AI brings.
- The 90-day plan is not a replacement for long-term strategy, it’s a way to build a rolling 18-month AI roadmap in manageable increments, reviewing and adjusting each quarter, because the strategy is written in ink but the plan is written in pencil.
- The hyperscalers, Google, Meta, Microsoft, OpenAI, Anthropic, will invest nearly $400 billion in AI this year, while enterprise spending is roughly $37 billion, a 10x gap that signals an AI bubble of expectations even if the underlying technology transformation is real and irreversible.
- The hardest people change with AI is identity disruption: a bank went from 90 people in a check-reading group to five, and the remaining challenge is not the technology but helping people answer “who am I now?” when the role they built their identity around no longer exists at scale.
- AI fluency in 2026 requires four levels: understanding what AI can and cannot do, using it responsibly, applying it to actual job tasks, and being able to teach it to someone else, because teaching crystallizes knowledge and forces the kind of organized thinking that makes skills stick.
About Charlene Li
Charlene Li is a New York Times bestselling author of six books and the founder of Quantum Networks Group, an advisory firm focused on business transformation and disruptive leadership. She has advised 49 of the Fortune 100 and previously founded Altimeter Group and served as Chief Research Officer at PA Consulting.
- LinkedIn: https://www.linkedin.com/in/charleneli
- Website: https://charleneli.com
Episode 100 of the PreVetted Podcast.
Full transcript
Federico Ramallo (00:00) Welcome back to the pre-vetted podcast where we spotlight extraordinary people and remarkable talent reshaping our world. Today we’re joined by Charlene Lee. She’s an expert in business transformation strategy and disruptive leadership and a New York Times bestselling author of six books, including The Disruption Mindset. Charlene is next book, Winning with AI.
She’s worked with hundreds of leading companies from Adobe to Southwest Airlines and has advised 49 of the Fortune 100. Charlene has also spoken at the World Economic Forum, TED, and South by Southwest. Early in her career, she founded and led the disruptive analyst firm Altimeter Group, served as chief research officer at PA Consulting, and today she’s back at her entrepreneur roots, running her firm.
Quantum Networks Group. She’s a Harvard College and Harvard Business School graduate and was named one of the most creative people in business by Fast Company. Charlene, welcome to the show.
Charlene Li (01:06) Thank for having me.
Federico Ramallo (01:07) I am honored to have you here today, and particularly because we are recording our 100th episode, which is an interesting milestone.
Charlene Li (01:17) Yes, congratulations to you on that.
Federico Ramallo (01:19) Yeah, thank you, thank you. There is something with the wrong numbers that find interesting. So let’s start with how did you get into this work in the first place?
Charlene Li (01:30) Well, I became an analyst at Forrester Reacher’s in 1999. And I did that because I thought I’d take two years off from running businesses. I was running a newspaper group, putting newspapers up online. And I was just tired of managing people because I just had my first child. I was about to probably have a second child. So I’m like, I think I need to take some time off of managing people and just manage my family. So I’ll be an analyst for a little while.
And I found that I loved it. I was at Forrester for 10 years, and then I started my own firm and did that for a number of years. And I just really love doing this type of work. So I kind of, you never kind of grow up and say, I want to become an analyst. I want to be an author and go around and give speeches. That’s not on anybody’s, quote, a career roadmap in there in school. So this is, I kind of fell into this and I just feel so incredibly lucky that I can do the work.
that I do.
Federico Ramallo (02:29) Amazing amazing. Yeah, I mean it’s it’s a It doesn’t doesn’t have a lot of appeal for kids right when you don’t think I want to grow up to become an analyst right but there’s so many rewarding jobs that once you get into it and And you appreciate everything that entails then you fell in love you fall in love of the job
That’s amazing. And what made you decide to write about AI on your book Winning with AI Now?
Charlene Li (02:54) Well, I’ve made a career of looking at the latest disruption that’s happening, especially if it’s driven by technology. So there was the internet, a big disruptive technology that came around, did search engine and search engine marketing, email marketing, again, online publishing, as I mentioned before. And then of course, social media, I wrote my first book about how companies could use social media. And so just have been very much at the leading edge of
all these different changes. So looking at where disruption is coming from, and when AI came on the scene and in particular chat GPT, which made AI so much more accessible to anybody with a browser, I went, this is going to change everything. And so I dropped the book that I was working on at the time and picked up writing about AI instead. And so the things that came out of that was the book winning with AI. I just felt like people needed
to have this explained, something very complex, simplified and explained in a way so that they could understand it and when they could understand it, they could do something about
Federico Ramallo (03:58) Right. On one of the earlier episodes, we talked about the importance of interface, how ChatGPT brought AI to the masses because of the interface. And then a lot of the… It was before generative AI, right? It was just ⁓ LLMs at that point. And then a lot of other engines…
start building their own interface which allow people to use them, right?
Charlene Li (04:24) Right. And again, we talk about how important data and the algorithms are, the LLMs are, but it was the interface that was the breakthrough moment. And it continues to be the interface in the same way that as you are using AI and incorporating into your workflows, the way you interact with that AI. And again, it may not necessarily be through a chat interface. It may be through a different dashboard. It may be through your CRM.
or another platform that you’re using. could be something that looks completely different than specialized app on your phone. It could be a voice interface. It could be a robot. So the interface of how you interact with AI has a lot to do with your adoption rates.
Federico Ramallo (05:09) Yes, yes, the adoption rate started with this first interface, but now the amount of interfaces that has been opened up has increased a lot because people started to find other use cases. Actually, I recently heard an entrepreneur said that he feels much more productive using the voice of AI than the chat.
So he would have a conversation with the agent and he felt it’s more productive than himself just typing stuff into it.
Charlene Li (05:39) It feels a lot more natural too as well.
Federico Ramallo (05:41) Right, right. I come from the technical side. So for me, the keyboard becomes the most efficient way to provide input to the computer. But his argument was, well, when I’m talking about ideas and being creative and being able to summarize and expand different ideas, then I found.
a conversation much more efficient, right?
Charlene Li (06:07) And again, I think it depends on the task and also the person, which interface actually makes the most sense.
Federico Ramallo (06:15) Yes, yes. Yeah, because we realized we’re both right, I guess. It’s different perspectives, right? It wasn’t like one is right, the other one is wrong, So what does winning with AI mean to you?
Charlene Li (06:27) Well again a lot of people use AI and the biggest complaint that I hear from people is that I’m not seeing the value I don’t understand the ROI and and I go is that really what you want AI to do with you wanted to make sure that you get some sort of return on investment that you’re investing in AI whatever amount it is and that you’re getting return on that it’s and I like but what do you really need the value to do for you and I think.
In the end, what you really want to do with AI is to help you create or maintain or extend your competitive advantage that allows you to win. And what I mean by winning is whatever your definition of that is, is that beating the competition? Is it helping your customers? What are the things that really matter to you? What’s your definition of success? Whatever that is, how can you use AI to support you in that definition of success?
in that creation of value. Because it could look very different if you’re a school or a hospital, a government, a bank, a nonprofit. They could all look very, very different. Your definition of what winning means. but again, in the end, we all have this desire to win. And the book is all about how AI can support you in that effort.
Federico Ramallo (07:48) Right, right. The goal has a personal meaning. The winning, the fiction of winning has a personal meaning.
Charlene Li (07:58) Yeah and it’s not just personal it’s personal to you as a person but also to your organization and when organizations say I don’t know how we’re creating value with AI my first question back to them is. What is what are your biggest strategic goals what are your biggest problems that you have to overcome now apply AI to those goals or those problems. And that will guarantee you.
that you’re creating a lot of value that it’s you being used in the way that’s helping you so if you just sit there and go well what can I do that’s not helpful because they do so many things but if you ask instead. What can I do for me how can it help me achieve. My goals and my objectives as a very different question. You would answer them in two very different ways and that the second way is much more productive in terms of helping you figure out what you can do with AI.
Federico Ramallo (08:51) Right, right. I think we’re talking a little bit about this before, where we become the leading of a team of agents or even if a team of one, right? But this idea of we have a lot of eager agents wanting to please and help. We just need to point towards where we want them to go, right? And then they’re going to…
facilitate us, even if we don’t know the path, we know the direction we want to go, the agent can help us find the path and then move it forward, right?
Charlene Li (09:32) And I want to simplify it even further. It’s not even just agents. This is called AI. Again, it may be in the form of a agent that is optimized for a particular workflow, but it may also just be this exploratory AI that’s saying, look, let’s figure out this. There’s these opportunities here, which are the best ones for us to do right now, which one would give us the biggest strategic advantage, which are these.
Can we do right now? So that may be something that you do as a quick win to establish a foundation on which you can build. mean, there’s many different ways to approach it, but knowing what you want to accomplish, what are your top strategic goals? Not all of them, but your top strategic goals, ones that are meaningful to you and your organization, then focus on using AI to create value against those things. And it won’t be easy because it’s a big
huge problems that you’re trying to solve. It won’t be fast, but again, it’s something worth the fight, worth the effort to go and figure out how to use AI. It won’t be obvious, so you’ve got to put the work in.
Federico Ramallo (10:42) your book, you talk about a 90-day plan instead of a long road roadmap. Right? Why did you choose a short term instead of a long
Charlene Li (10:53) Well, again, the biggest question that we get from people is where do I start? Like I’m doing all these different things with AI. None of them are really creating value. Where do I start? What is the first thing you have to do? The second thing I have to do a third thing we have to do. So we said, look, just focus on the first 90 days. And part of that 90 days is figuring out what your roadmap is. So you’re laying the foundation. So of course you have a longer strategic roadmap of what you want to do with your company.
Federico Ramallo (10:58) Right.
Charlene Li (11:21) We’re not addressing that. We’re assuming that you have a strategic roadmap that goes off into the future. You have that. If you don’t have that, there’s a different problem that you should be solving for first. But you have a strategic roadmap. You know where you’re headed. And what we’re doing is how do you take AI and for 90 days, make sure it aligns against that roadmap, against that strategic plan. And so we’re going to help you create a shorter term roadmap. We’re going to give you
exactly what to do for 90 days a week at a time. And that will actually create a roadmap for 18 months of how you’re going to create value every single quarter. And it seems insane in some ways to plan out an AI roadmap for 18 months when everything is constantly changing. But I do believe that it’s so important to have a roadmap to say what
where we want to be 18 months from now in terms of capabilities and value we’re creating, because then you know what to do today to create that value 18 months from now. Otherwise, you’re just going to do what’s the easiest thing, and it may or may not be creating the value that you want to create down the road.
Federico Ramallo (12:23) Right.
The value is on building the plan, not necessarily the plan per se, right? Because the plan could change.
Charlene Li (12:35) Right, and we also
like to say that the plan is that the strategy is written in ink, but the plan is written in pencil. So that you can change the plan as the technology changes as your capabilities change as the competition and the landscape and your industry is changing. So you should be able to revise that road map and at the end we advise you to we recommend doing that once a quarter.
So at the end of the first quarter, you say, where are we on the roadmap? What has changed? What did we accomplish? Are we ahead? Are we behind? And then you make adjustments for the other five quarters and you add another quarter at the end. It’s a rolling 18 month plan so that you have something longer than just the next month or just the next quarter to be looking forward to. So you’re looking and saying, this is where we want to end up. So we’re going to have to do these things today.
make these choices in technology, make these investments in terms of our people or in our data, and then that’s going to pay off over the next 18 months.
Federico Ramallo (13:38) Interesting.
So if everyone can use AI with a browser, what makes one company win versus another company?
Charlene Li (13:47) Well, this is the thing is that everyone has access to the same technology, the same data. And people will say, no, but I have amazing data inside of my company that’s unique to everyone else. So does everyone else. They have their own unique data too as well. And so in the end, the only thing that will help you is speed. We believe that speed is the new moat. M-O-A-T, moat.
And that’s the only thing that keeps you in a competitive advantage compared to other people in your industry, or keeps the people who are trying to enter your industry or redefine it out of your space. So you can defend yourself and, and speed. And what I mean by that is not just the adoption of AI, but the adaption of your organization. How quickly can you change your organization?
Because AI is just a technology. AI is not the strategy. If you’re not talking about AI coming in and technology being changed, it’s also about the organization changing because of that technology coming in. So if you’re not planning for a transformation and preparing your organization for the hard work ahead that is required to transform, then you’re not prepared.
You have to prepare yourself and your organization for that kind of disruption.
Federico Ramallo (15:12) Right. And on the book you describe more detail how people leaders can prepare. Right.
Charlene Li (15:18) Yeah, and the key thing here, I think, is that if you can think about AI again, not as a technology, and this is the biggest mistake I think that people do, is they think about AI as a technology, so they give it to their technology people and say, implement this technology. They’re not treating it like a business issue, like a transformation opportunity. And if they see it as a transformation opportunity, an opportunity to
really advance their organizations to get to their strategic objectives better, faster, cheaper, safer, then they’re not really looking at AI the right way. They’re looking at it as a technology, like a calculator that you can pick up and use when you need to. It’s much more than a calculator. It’s much more than a tool. It’s one of the reasons why we don’t like to use the word AI strategy. And we use AI roadmap because you already have a strategy. You have a business strategy.
This is an AI roadmap to support your business strategy. And what will you do? What won’t you do on that roadmap? You can change it. You can call it a strategy if you want, but just be very clear. AI is in service of your business strategy.
Federico Ramallo (16:30) Right, right. I I’ve seen similar things with creative people saying that AI is going to replace their jobs, but I think that creative people can use AI to assist them. But AI is not going to be, it’s not going to replace the creative people. It’s not going to replace the business strategy. It’s not going to provide the vision of the company, right? It’s going to help on assisting towards.
that vision, but we need to provide that vision to the AI in order to give us value, right?
Charlene Li (17:02) I like to say that AI will give you the answers that are as good as the questions that you ask it. So if you ask it better questions, it will give you better answers. That’s it’s a reason why people who are experts in their fields do so well with AI, they can do so much more than somebody who is just starting out in the field.
Federico Ramallo (17:24) right. The specificity of the questions provides much more value. I’ve seen that happening as well. And that’s why I’ve been thinking of this idea of people with core skills, core knowledge, or as you mentioned, experts in their fields, they’re going to have a much more demand in the market because they will be able to leverage
AI and all the technology behind it much more than than somebody that start fresh with no without that experience, right?
So what does it mean connect AI to your business strategy? Can you elaborate a little bit more on that?
Charlene Li (18:04) Sure, again, I talked a little bit about it before, which is when you have your business strategy, it has a vision of where you want to be at some point in the future. And it says, these are the things we’re going to do. These are our strategic initiatives to be able to achieve that strategy. And so when I say connect AI to strategy, OK, it’s that. It’s like, how does AI then support those strategic initiatives?
Or you could be looking at your strategic plan and saying, we have these plans, but these are the biggest roadblocks in the way. And every company has roadblocks. are things that you have to do, new markets you want to open up. These are just challenges and opportunities for you to use. And the question becomes, well, how can AI help us address those problems, overcome them, or help us take advantage of those opportunities? But it’s always in service again of your strategy.
I if I see so many times people like yeah we’re doing this and that with AI and like okay great how does that support your strategy haven’t really thought about it that’s not a good sign it’s not a good sign because then why are you doing this
Federico Ramallo (19:12) Right, right.
That’s what you see happening when the technology people are implementing AI disconnected with the business strategy.
Charlene Li (19:23) Exactly, exactly. It’s not to say that the technology people don’t have a role. They have an incredibly important role. But to do that in absence of that business strategy is a big mistake. So again, the really good CIOs now are strategic in nature. They are thinking about the implications and the advantages that AI and technology in general can give them.
They’re not just looking at it only from a technology point of view, they’re looking at it very holistically.
Federico Ramallo (19:55) Very interesting. So you talk about the CIOs. What do you think the CEOs should do related to AI?
Charlene Li (20:04) Well, in the very first week of the book, the first chapter of the book, we talk about getting your hands in and using it. Right? There’s no substitute to you using it. And if you already are pretty comfortable with it, then figure out what is your edge? What are the things that you don’t know about AI that you are curious about? What are the things that you want to learn about so that you can be better prepared to lead your organization? So
There’s always something, but people will ask me like, what’s the best course I can take? What’s the best podcast to listen to or newsletters I can read. And so what I say to the CEOs and any leader who wants to do this to get up to speed, so to speak with AI is you’ll never find it in a course. You will never find it in a newsletter. The first thing you should do is to figure out.
what are the things that you need to figure out? What are the questions and the problems you want to tackle? Because you know your business better than anyone else. You know yourself better than anyone else. So get up to speed on how to use AI, but then apply it against one of the biggest questions or problems that you have. It’s always in context of something that you’re trying to do. That’s really important and strategic to you.
Federico Ramallo (21:24) Right, right. I used to think of AI as workflow automation. That was kind of my first approach to how AI could provide value to
Then recently we did this exercise with my family where we would ask our AI prompts based on what you know about me, generate a drawing of how you see me, right? It was something like that, right? But then we got really interesting, very different responses from each other, right? And we shared it. And then we kind of see
in one picture puts together out of the history and the context of everything we kind of discuss with the prompt. So it was a very enlightening exercise, which comes back to this idea of if you have a conversation based on the problems that you want to solve and you keep…
pushing forward that conversation, then you’re going to get much more interesting results because then the agents or AI can truly understand what you’re trying to achieve.
Charlene Li (22:33) Agreed on that.
Federico Ramallo (22:34) Yeah, I used to think it was more of an input and output and that’s it. But now I’m realizing it’s more of this ongoing conversation, right?
Charlene Li (22:43) All right.
Federico Ramallo (22:44) So why do you say it’s about people and not technology?
Charlene Li (22:51) Yeah, this is something that I’ve been saying for quite a while, I’ve been through multiple phases of digital transformations. And the thing that I’ve seen over and over again is that it’s never about the technology, it’s always about the people. Because it’s digital transformation, there’s a digital technology part, but the transformation part is about people.
It’s about organizations and cultures and process. So I think it’s so important to understand this isn’t just a technology that you put in place and then push a button and it works. The organizations that think that’s the case of deploying AI, how come it’s not working? Well, you haven’t trained your people on how to use it. and like, yeah, I guess we should do that. I’m like, yes, absolutely. And so it’s just overlooked over and over again, where they provide the training.
Federico Ramallo (23:34) you
Charlene Li (23:38) But then give people no time to experiment with it to figure out how they can use it in their jobs or they give them the time to train the things, but you can’t make any mistakes, right? If it’s something that’s going to be important, then use it perfectly every time. And any of these big moonshot ideas that we have, they’re just too risky. They take too much time investment. Don’t use AI for those things because we don’t know what the value is. Well, if you don’t try it, how will you know there’s no value?
There might be value here. it’s, it’s a lot again. It comes back to people every single time. It consistently always comes back to people and AI even more so because you could just look at it as a replacement tool. And some companies do, I can just replace all these people with AI and I don’t need the people anymore. And you realize that oftentimes if you don’t do this the right way, if you do it too quickly, you do it too much. It’s the only focus, cetera.
then you could really shoot yourself in the foot. And again, not have a workforce that has some sort of leadership ladder. You could be not investing in critical thinking and then people get just not as diligent about making sure that the results are within they could most likely be. Again, the ways to win are oftentimes not just low cost, but differentiation. And if you clean out all the creative thinkers,
in your organization all the people who are bringing in new ideas then you’re not going to be very innovative you’re not going to have that friction that allows for innovation.
Federico Ramallo (25:12) Right, right, and we cannot expect AI to innovate for us.
Charlene Li (25:20) Again, AI won’t do the innovation. It’s you asking AI the questions that allows for the innovation to happen. But the innovation starts with the people using AI.
Federico Ramallo (25:32) I haven’t been able to figure out how to trust the decision making of AI. I mean, how can I delegate to AI more complex tasks where decisions needs to be made and be able to trust it given that it’s non-deterministic, right?
Charlene Li (25:57) Yeah. So here’s the question is how do you learn to trust a person when you’re bringing them into a new job and new person position where they don’t know how to do it. You train them. So you show them how to do it. Then you watch them do it along with you. Then you watch them do it on their own. And when you trust that they know how to do this complex task, then you let them do it on their own without maybe a little bit supervision from far. And then you just let them do it completely because they got it. They understand it.
A is exactly the same. The thing is that when you’re training it, you may notice that there’s some variability in these things. And so the question becomes, how do we train it? So there isn’t that variability, or do you actually want that variability? Because it reflects the fact that there is no ever not ever a single right answer. And that it really depends on the inputs that you have. So you can then figure out if you’re getting still what we call hallucinations or variabilities that are unacceptable.
Is it the data? Is it the query and the task that we set it up to? Is it the QA? You can build quality assurance directly into the agents, where it’s another agent that says, you delivered this result and like a human that doesn’t match what our standards of success and quality are, go back and do it again. And I suspect that it’s the data inputs that aren’t good. So go choose a different data source or your reasoning is not right.
you fell apart on this. And so we can build AI to check AI in much the same way. So we can move the human out of the loop. So it’s right now the standard, the gold standard is I have a human in the loop, but the question becomes when can we get the human out of the loop? And every organization is going to have a different threshold for when that’s going to be.
Federico Ramallo (27:30) Right.
Right, right. Because as you said, if there’s no margin for errors, then how can we trust them? How can we trust AI, right? We need some, and as you describe it, we need to treat it as a new hire or as a person that it’s going to be, it’s non-deterministic, it’s going to make mistakes, and it’s going to learn based on the training and the experience, right?
Charlene Li (28:13) Yeah, and people make mistakes too. And we have the safeguards in place to help catch that and to do quality control and to do improvements. And AI can do exactly the same.
Federico Ramallo (28:16) Yes. Yes.
I think that’s an interesting take that we should think it more of as a person. And I can understand that leads to the replacement idea, which I don’t think that should be the approach for AI, right? It should be more as an assistant, right?
Charlene Li (28:41) No, mean, there two different questions.
Yeah, there two different issues. One is how do I train AI effectively? We decided that we want to do it because it’s going to do things that nobody’s ever done before, like do analysis across 50,000 documents. No one ever did that before. But you also want to ensure that the quality of the analysis is at a certain level and that’s acceptable. So you would again train, evaluate, delegate.
And then the other part, the question, I think is not so much will AI replace us as we know that if we do not stay up to date, if we do not constantly improve our skills and our knowledge and expertise, that we will at some point be outdated and it will be either another person or a technology that replaces us. So I do believe that AI will take jobs.
Hopefully that the jobs that nobody really wants to do or that it can do much better. And that allows us to do the things that are really interesting and challenged to us rather than these mind numbing routine high volume things that AI this is perfect for doing.
Federico Ramallo (29:52) for those repeatable tasks, I think it’s great because we as humans get tired of doing the same task over and over again. And then we get fatigued and then we lose focus and then the quality drops. So I can see how AI would support us on that.
Yeah, and as you said, think it’s the replacement theory is going to happen on the organization level, right? A company not implementing AI is going to be replaced by the company that is using AI, right?
Charlene Li (30:24) right.
Federico Ramallo (30:27) Yeah, but then on the expert level or the individual level within an organization, I think it’s more about adapting to work with AI rather than trying to fight it, right? Because then they’re going to get out of a right?
Charlene Li (30:41) Right.
Yeah.
Federico Ramallo (30:44) And how should leaders help teams feel more confident about using AI?
Charlene Li (30:52) Well, a couple things. First of all, look at the fluency levels of your people. Do they feel like they’re fluent with AI? And what I mean by that is one, they understand what AI can do what it can’t do. They are comfortable and confident about the way they are using AI in a responsible and ethical way.
Third, that they can use AI to help them with their job. They have the beginning and the foundations to be able to say, I can use AI in this situation and actually apply it to help them do their jobs better. And then the fourth area, and this is kind of a bit tongue in cheek, you can teach it to other people. Because when you can teach something, it says that you have enough fluency with it to be able to explain it to somebody. And also, it’s just.
crystallizes your learning when you can explain it to people because it has to be organized in a way that you remember it. So the soonest that you can teach something you know to somebody, the sooner you will have it ingrained in you too as well. So I do believe that 2026 is going to be the year of AI fluency when people are investing heavily and not only the top down training, which is really good for the first two things, like what does AI do? What doesn’t it do well?
and responsible and ethical AI. Those things can be from the top down. But it’s also from the bottom up. How do I use AI for my job? And it’s talking to somebody else who does a similar thing to you, either in your department or maybe in a completely department. But we’re still using AI to do many of the things that we want to do. So, and do believe that it’s going to be very important for companies to figure out how they can get their organizations up to speed as quickly as possible.
Federico Ramallo (32:38) Right, interesting. And what do you think is the hardest people change you see with AI?
Charlene Li (32:47) I think it’s giving up your sense of identity. If your identity has always been, I remember this bank talking to me and saying, we had 90 people in our check reading group. These are people who would read the checks that the machine couldn’t. The machine just couldn’t figure it out. So like some small percentage are actually processed by humans who can look at it, do research into it, figure out whether it’s real, fake, whatever.
And then the process that that has gone from 90 people down to five people in their bank. So that is the hardest change is to say, if your entire identity is about being a check reader or about being a certain thing, and that is what it is. And it’s fairly brittle and rigid about what your definition of what you do and how you add value and AI comes along and blows the whole thing up. So your job doesn’t exist anymore. If it does, we need a fraction of the people to do that.
Now who are you? What value do you create? I like to say that disruption happens when you leave the place where you were comfortable, you knew where everything fit and how everything worked together. And you have moved into a future state where you know things are going to have to work in a different way, but you still haven’t figured out how it all works together. So you don’t have this new normal yet. And you’re just disrupted. don’t, it’s not a comfortable place to be in.
And that’s where we are right now. We’re still all figuring this out. And in the meantime, it’s incredibly uncomfortable. And so the biggest mistake and the hardest people change to do when it comes to AI is to prepare people for this change so that they can get ready for it. They understand what it means that this is going to go on for a while, that we’re just going to be going through a constant series of transformations.
and we need to go fast. So we need to invest in ourselves, have the time to breathe, to recollect ourselves, to figure out where we are, and then go fast again. So it’s important to understand that this is about dealing with disruption and identities. And these are really hard, hard things that we’re asking people to do. So give them the support that they need.
In organizations, I sit in there, yeah, this is going to be hard, know, tough your way out of it. And like, no, you can make it a little bit easier here. I’m not asking you to take away all the friction, but you can reduce it a lot. And the organizations that do a really good job of changing in a disruptive way, pace themselves. They’re moving super fast. But they’re doing it in a way that they are preserving the humans. They’re making sure that people are rested and can think fully.
Federico Ramallo (35:12) Yes.
Charlene Li (35:41) They’re not completely running to the ends of their capacity so that there was this room in this space to experiment and learn and grow.
Federico Ramallo (35:50) Right, right. Yes, because the…
I’ve seen this happen when internet started and so many other technology disruptions where it was a little bit chaotic, but then some of the projections in the future were accurate, most were not, right? So the future was uncertain, but then some people lost their jobs because of it. But then eventually they were able to adapt and find new jobs to…
to start working on the new market,
And then there are new jobs that we may not see coming today that are going to come in the future.
Charlene Li (36:26) Absolutely. I think of some research that shows that 40 % of the jobs today didn’t exist even 40 years ago. And you see this constantly, who had content marketing specialists in their organizations even 10 years ago? So again, things are constantly changing. New skills. Again, I think the best skill you can have is the ability to learn. That is the most important skill.
Federico Ramallo (36:43) Right.
Charlene Li (36:54) And so if you keep that up to date, you are trying to learn. And one of the things that you have to get comfortable with AI is that you’re never going to learn all of it. So it’s one of the hardest things that leaders have a problem with is that they don’t feel like they can do anything with AI until they can feel like they completely understand it. And I’m like, you’re not going to, you’re never going to get to that point. So.
Assume that you can’t do that then what’s keeping you from diving into AI and using it and leading it. In your organization.
Federico Ramallo (37:27) I love your analogy that it’s like a human because you never fully know a person. There’s always something new that can surprise you. So it’s non-deterministic in a way, a human, right? And if we think the AI like that, that analogy holds.
I agree, yeah, the amount of information that is processing, that it’s learning, it’s more than we can fathom.
Charlene Li (37:54) Yeah.
Federico Ramallo (37:56) So, you know, let’s talk a little about AI hype. What should leaders ignore and focus related to AI hype?
Charlene Li (37:56) based.
Well, again, I think there’s just so much news and you could spend all day reading just news reports and you will not be better off than you are today. So what I’d like to say is focus on, again, it sounds like a bit of a broken record, but really focus on these are the problems I need to solve with AI. These are the opportunities we need to explore.
be really focused on your business and because you understand your business better than anyone. And so really to help you deal with AI hype, start with those things that you know about yourself and your organization. And then from there, keep asking, okay, here’s a new technology in AI. How can it help us achieve one of these things that we have on our goals? How can it help us address these problems?
And so that digs to the AI hype very quickly because when you talk to a vendor, he goes, just stop, stop telling it. Stop telling me about the features. Just tell me about the value it creates for me and how can it help me solve one of these five problems that I have. So you’re helping the vendor understand your problems rather than you trying to understand their technology. And it is the number one thing that I think people have a problem with is it’s so easy to fall into the hype.
For example, are we in an AI bubble? Probably, because we’re spending the hyperscalers, the big companies like Google, Meta, Microsoft, Chatch, EBT, Anthropic. The hyperscalers are going to invest almost $400 billion this year. And the actual expenditures from companies is going to be a 10th of that, 37 billion. So there’s a big disconnect of 100X that
Federico Ramallo (39:50) you
Charlene Li (39:51) It’s not there, right? So I, again, I’m not a financial analyst. I can’t tell you whether the stock prices are worth it or not. I can tell you that the P-E ratios from the dot com bubbles were much higher than what we’re seeing in the market today. So I do think we have an AI bubble of expectations about how fast it’s going to come. But make no mistake, this technology is absolutely real. The transformations are absolutely real.
Federico Ramallo (40:07) interesting
Charlene Li (40:19) And so don’t use a hype in the AI bubble as an excuse not to do anything. It’s the worst mistake you can do is to sit here and go like, well, I got time. No, you don’t. You need to get at it. And there’s a famous quote from this great philosopher called Yoda from Star Wars. He said, do or do not, there is no try. And so I feel that that applies to AI. You just have to go do it.
or decide not to do it and have good reasons for it. But don’t just like, I’ll try a little bit here or try there because that’s not committing to the possibility that AI is actually going to make a difference. You have to commit, you have to do it.
Federico Ramallo (41:00) Right, interesting, very interesting. and I didn’t, I mean, when I talk with the community, startup community, most people agree on this idea that there is so many greenfield opportunities because of now AI allow us to do so many things that we couldn’t do before, right? Or new business opportunities that didn’t exist before. But it’s interesting that you’re mentioning that the P-Ray shows
back in the dot com bubble was not as high as, was higher than today. So, but I think that there’s more potential now than during the dot com. So that is very interesting.
Charlene Li (41:31) was higher than we are now.
Federico Ramallo (41:40) So what is one habit you want leaders to build to stay relevant in the world of AI?
Charlene Li (41:48) That’s a great question. I think it’s the ability to learn. I keep coming back to this. Are you setting aside time to learn? And what I mean by that isn’t just to read about a technology or a process. For example, vibe coding. It’s one thing to read about vibe coding. It’s another thing to actually try it, to use it. And again, I’m very excited about vibe coding because as a non-technical person,
I want to experience that joy of making something, creating something. And I’ve created an app or two and they’re pretty decent. It’s great fun. And it’s also really hard to do that with the tools today, but I think they’re going to get much better, much faster. And if you think about every single employee having the ability to use these agentic tools to modify the way they do their work.
and everybody has their own customized personalized agents because everybody does their work differently. You can choose from a library of all these agents and put them together in a way that just makes sense for you and then have have clearance to use your data. They’re already connected to your data sets. They have the permissions of the governance all built in, so you’re removing a lot of the back end headaches that it takes to create apps and you’re just focused on using these agents to customize themselves to the way that you do your work.
That is going to be amazing. That’s the future of vibe coding. Right. And the ability for any person, you democratize the ability to go and create these apps. Every company has this big long list for IT department to go fix these things, these bugs, cetera. What if you could do all those things yourself with your little vibe coding agent that you didn’t have to go to IT to ask them to build this for you, that you could do it yourself.
Federico Ramallo (43:35) You
Charlene Li (43:40) and that you do within a permission and security structure that is approved by the enterprise, that would solve a lot of problems. It would reduce a lot of friction. People would feel like work is these tools are working for them rather than against them to get the work done. I mean, it’s one of those things where I’m very excited about just what the potential is. So CEOs need to get their hands dirty and they need to apply the technology to their business every single day.
And they need to be public about this so that other people can see it and follow them, that they are modeling how to do this, that you’re experimenting, you’re learning every single day in everything that you do. So learn and learn publicly.
Federico Ramallo (44:25) Interesting, very interesting. I have a… I don’t know how to ask this question, but it’s about shared experiences, right? I I have mixed feelings about the democratization of bytecoding or any other… You know, when we lower the entry bar to…
build something, right? Because on one side, I agree that now we have more people that could provide value, solve problems that couldn’t do it before, right? A non-technical person now could build an application, right? But then on the other hand, that barrier would, well…
maybe it’s wishful thinking, it would allow us to plan better, to build better tools, right? Although sometimes it was bad planning, which led us to frustrating tools, right?
Charlene Li (45:20) Well,
I like what the potential is. You have a workflow. I mean, you’re in the workflow as a person, as an individual contributor, and you’re looking at the workflow, like this could be done in a better way. We could use AI to change this part of the workflow. Let’s try and see if that works.
Federico Ramallo (45:46) Right.
Charlene Li (45:48) And it may or may not work, but you give it a try and see if it works. then, oh, okay, that worked or it didn’t work. I’ll try something else next time. That kind of learning is incredibly valuable to an organization. It’s not like you’re going to redesign everything in the backend for the entire company. You’re just taking your workflow of, for example, processing one type of claim in the company. But if every single employee does that, a revolution happens.
So this is what speed looks like. It’s when you can enable somebody at the front lines, that they have the confidence to be able to make these changes, know that they have the permission to do that, know how they can do it safely, and know that they can improve this to be better, faster, cheaper, again, safer. Those are the things that are really important and requires that we trust our people to do these things, but also that we put the guard whales.
to make sure they know what the boundaries are. know, innovation takes a lot of work. It isn’t like just go out there and innovate. It takes a lot of work to structure things. And so I do believe in having what we call go-de-lux governance. Not too much, not too little, just right to keep everyone safe, but also to keep everyone moving fast.
Federico Ramallo (47:10) Right, freedom to be innovative but also controls and checks and balances.
they work within a safety net, right?
So if every organization has their own different personalized approaches, and probably this is maybe this not a good analogy, but something similar happens on the social networks. The algorithms show us what each of us wants to see, which kind of reinforces, and then we have completely different experiences. So I’m wondering what would happen in the future if instead of shared experiences where
most of organizations have the same similar workflows or similar ways of doing things with their own flavors, then now going from one organization to another would be a completely different way of working, right? That would provide their own challenges and also their own benefits, right? What are your thoughts about how would that develop?
Charlene Li (48:07) Well, first of all, you will be very unlikely to be able to take your agent from one workplace to another workplace. And this is an interesting situation where you might have a different set of handcuffs in the future, being that you’re used to doing work in a certain way with all these agents and knowing that you have to leave them all behind and start over from scratch again. You’re like, do I really want to do that? Cause I just invested so much time and energy.
into these agents that are just fine tuned to the way I work and I’m going to have to leave them all behind and start over again. So serious barrier to exit. similarly, if you’re coming in, how do I enable a new employee to get up to speed and to create those agents as fast as possible so that they can be up to speed and doing things? So it’s, an interesting conundrum. But I think
your question of differentiation. I think it’s a big danger when everyone has the same technology, everyone has using the same underlying algorithms. They all have their own unique data. What then allows them to be truly differentiated? And I think it’s partly the way that you use the tools. If you depend on AI to come in norm to give you the answer to something that’s safe, because that’s what AI is going to do. They’re going to give you the most likely answer to be right.
So there are some prompting tools called probabilistic prompting and verbalized sampling that says, give me five options, but go across a range of possibilities or probabilities. Give me a score against all of them. But I don’t want you to just come up with one answer. I want you to come up with all the ranges of possible answers and rank them. Give me a numerical scoring that says like high is a hundred percent. Like this is the right answer, but it’s also the safest answer and most unoriginal answer that you can think of.
to like 50 where it’s on the edges here, but it’s not crazy yet, but it’s on the edge and you will get very different results because AI has been trained to give us one answer and that one answer may or may not be right. But if it’s a spectrum of right, then you’re much better off seeing the spectrum and all the probabilistic potential that’s in between. it’s again, knowing what AI can do and what it can’t do.
Federico Ramallo (50:12) Right.
Charlene Li (50:26) knowing what it’s good for and then how to use it in a way that’s going to meet your needs. But I think this whole idea of differentiation is going to be something that’s really important because it’s the people again who can use AI in a smart way, use it to ask the most original questions that no one else is asking. It’s testing a hypothesis of what you believe the future is going to look like. And then if your future looks remarkably, remarkably different than anybody else’s, that will be a very creative answer that you get.
AI because no one will have asked these questions.
Federico Ramallo (51:00) Right. And you can push the AI to give you not the average answer, but also the spectrum of possible answers. You can push the AI to do much more. just, it all starts with you as a user knowing what you can and cannot expect from AI, right?
Amazing. So Charlene, we’re running out of time, but I could keep talking with you because it’s a very interesting topic. But I truly appreciate you being here today.
Charlene Li (51:29) Yeah. ⁓
thank you very much for your time.
Federico Ramallo (51:35) Thank you, Charlene.