The New Playbook for Agility During Disruption

At Fast Company's 13th annual Grill during SXSW, I joined Prashanth Chandrasekar, CEO of Stack Overflow, and Sam Jordan, who leads computing and emerging technology at Future Today Strategy Group, for a panel on how leaders can stay agile while AI rewrites the rules around them. KC Ifeanyi from Fast Company moderated.

What struck me about this conversation was how honest it was. There's so much noise around AI right now, so many sweeping claims about what it will do to companies and industries and work itself. This panel stayed grounded in the real challenge leaders are facing: how to create clarity and coherence and forward momentum while the ground is shifting in real time. We talked about the missteps companies make in the AI race, what's actually worth paying attention to versus what you can safely ignore, and how to move from reactive to resilient.

We live in existential times across every company and industry. The organizations that navigate this moment well will be the ones that build coherence fast enough to act on it.

KC Ifeanyi: Good morning, everyone. Welcome to the 13th annual Fast Company Grill. My name is KC Ifeanyi. I lead the programming here at Fast Company and our other live events. We’re kicking things off this year by talking about how leaders can stay agile during mass disruption. I’ve got a wonderful panel here, so to start, I want everyone to go down the line, introduce yourself, and tell us a little bit about your company.

Prashanth Chandrasekar: My name is Prashanth Chandrasekar, and I’m the CEO of Stack Overflow. Stack Overflow is one of the world’s largest software developer communities and platforms. We’ve been around for about 18 years. We have a large enterprise presence, where our knowledge base is now being used by thousands of companies for internal agent use, which I’m happy to talk about.

Sam Jordan: Hi, my name is Sam Jordan. I lead computing and emerging technology for the Future Today Strategy Group, or FTSG. We are the firm that companies come to when they cannot afford to be surprised. We specialize in strategic foresight, which is basically a disciplined way of navigating the future, creating a vision, and creating a plan.

Peter Smart: My name is Peter Smart. That is my real name. I lead Fantasy. We are a pinnacle digital product team. We work with the world’s biggest companies to ship their flagship digital products. In this moment of change, where value is being rewritten, we help companies figure out where the new focus of value is moving and then help them ship it today. That’s our job.

KC Ifeanyi: We’re here to talk about how to stay agile during disruption, and of course AI is one of the main forces of disruption at the moment. I’d love to start with everyone’s point of view on the current landscape of AI. What have you seen that is BS, or worth ignoring? And what have you seen that’s actually disruptive and worth paying attention to? Prashanth, we can start with you.

Prashanth Chandrasekar: I’m happy to go. We have a fairly decent view of what’s going on because on our public platform, we’ve had 18 years’ worth of knowledge created on every possible technology question. That’s something like 70 billion tokens of information and around 90 million questions and answers on everything technology. All that data is now being used by the LLMs for pretraining and for AI answers. If you’re a technologist, you’re likely using our data under the hood.

In terms of hype versus what’s real, this phenomenon is absolutely real. We have to accept that this is here to stay. It’s a tectonic shift, a once-in-a-generation type of change. Calling it a platform shift is an understatement. It’s more like a completely new ability to actually do things, more akin to the internet or even electricity back in the day. That’s how I think about it, versus a cloud platform shift, which is much smaller.

It’s going to have massive ramifications, but it will take some time to work its way through different functions and different industries. When I think about hype, it’s where people are overstating where the capability is today. Can you trust the answers? Can you really trust it in your workflow if you’re a professional in a particular part of the organization? If you’re in finance or legal, do you feel 100% confident that you could use AI to go a lot faster? The answer is maybe in certain circumstances and use cases, but definitely not in all circumstances, because there are hallucinations and all those kinds of things you have to deal with. Trust is a very important topic.

On the other end of the spectrum, something like coding absolutely is being used extensively. Ninety percent of our users on Stack Overflow use AI in some fashion for code generation. It just depends on the function and the use case. But in the grander scheme of things, this is definitely not hype.

Sam Jordan: I think there are two ways I’m seeing value emerge that are exciting to me. The first, which is certainly not hype, is the individual empowerment story. I’m sure everyone in this room has used AI tools to do things you could not have done a year ago. I could not code one year ago, and today I can make an app in 45 minutes. It’s not going to be a good one. I’m going to have to spend a whole day on the app, but I can make it. I think of my fiancé, who is getting his MD-PhD. He’s a wet-lab guy. He is a biologist, and now he doesn’t really need a computational biologist. He can do a lot of that work himself. Individual empowerment is absolutely value that we are seeing now.

In the very near future, where I see the value emerging is in the coordination problem. Now I am a principal and I have agents. I’m not talking about AI agents. I’m talking about agents that represent me in Congress and agents in my financial institution that are supposed to represent my interests. Today they can’t perfectly represent those interests, but with agentic systems, we’re going to get much closer to being able to capture that coordination. If there are any econ nerds in the room, Ronald Coase is smiling up at us now for reducing those transaction costs.

Where I see the hype, or where I tend to roll my eyes, is when I hear people talk about AI as one thing arriving all at once. When people talk about AGI as the moment where it happens, I don’t think it’s going to happen all at once. I think it’s going to be something that emerges, and emerges differently in different places in the economy.

Peter Smart: Similar answer, but if we zoom out and think about what happened in the Industrial Revolution, it was the exact same thing. Machines came along and we had the exact same conversation as a human species: Is this hype? Are these things really going to replace us? Hype is maybe the wrong way of thinking about it. It’s actually about zooming out and saying this is a fundamentally transformative technology. Do we really know how best to apply it just yet? I think the answer is that we’re still learning how to use this technology incredibly well.

A very practical example of the problem I see is that AI is treated as a panacea, like it’s going to solve all of our problems for us all at once. If we look at Hollywood, for example, and distinguish the real value of AI: AI is silicon thinking, but there’s also biological thinking, and that’s us. You can use an incredible video model to produce incredible visuals on a screen. Can it tell an amazing human story that resonates in your soul? That’s biological thinking.

There’s a difference between intelligence, insight, and instinct. AI is really good at intelligence. Where people see it as hype is where they ascribe to AI the hope that it can be both insightful and instinctual, and it can’t. AI makes mistakes and doesn’t live up to our expectations. But if you zoom out and raise your altitude to what’s happening to our ability to be productive, to orchestrate, to define novel answers, to radically create efficiency in our lives, this is transformative. This is the most transformative technology that we as human beings have ever made. We have to figure out how to best use it and not overascribe to it what is uniquely human.

KC Ifeanyi: Peter, I want to stick with you for a second because I know at Fantasy you work with massive companies like Apple, Google, Disney, Nike, and Microsoft. What’s the biggest mindset shift leaders have to make when they’re looking to adopt new technologies?

Peter Smart: I have a lot of strong opinions about this, and I’m super keen to hear from these guys as well. There are two types of company in the world right now. There are companies where there is an absence of leadership in this moment, where there is an overly democratic response to the question, What does AI mean for our company? It’s up to the team to try and figure it out, and leadership is just bringing up those answers and aggregating them. In those environments, there’s a lack of coherence.

For companies that are accelerating in this moment, the key word is coherence, and that coherence starts with great leadership. Ultimately, it’s up to leaders to determine what this technology means for their organization. Without that, you aren’t really getting to the fundamental root of what this technology can mean for you.

AI wasn’t created to take better meeting notes. That is not the purpose of AI. If we think again about massive technology companies or Hollywood, it can create fundamentally new forms of value as quickly as it’s replacing the value that you produce today. I think it’s up to leaders in these environments to create the conditions for coherence, where they have a really clear vision for what this technology means, and then create the frame for their organization to start accelerating toward that. In my day-to-day lived experience, there’s a very marked difference between those who have sought external support and brought that expertise in to help them figure it out, and those who are still trying to aggregate it from different perspectives across the company.

Sam Jordan: If I may, what I’m going to say kind of piggybacks off this. How many of you are familiar with the Pirates of the Caribbean franchise? I promise this is relevant. I don’t know if it’s the second or third movie, but Jack Sparrow has this compass, and the compass is supposed to point him toward the thing that he wants. In the movie, all of a sudden he doesn’t know what he wants, and the compass is spinning around. He’s aimlessly going in different directions, really going wherever the wind takes him. I think of this with a lot of companies right now. I think a lot of companies are quite aimless. They don’t really know what they want to become.

When I think of resilient organizations, the ones most positioned for change, this may sound counterintuitive, but I think of companies like Berkshire Hathaway, the Catholic Church, and the U.S. military. These are organizations you wouldn’t describe as agile, but they are opinionated and they do have a vision. I think that’s the most important thing right now.

Peter Smart: Building on that, if you are very clear on your purpose, then how you deliver that value can change. I’m a designer. The purpose of a designer is to imbue value into a product. How a designer has shown up in the world 100 years ago, 50 years ago, or 10 years ago has fundamentally changed. It’s always going to change. But if you’re very clear on your purpose, you can figure out how to apply this technology to do that purpose in a new way. Those examples are good because the fundamental document doesn’t change, but how we do it changes.

KC Ifeanyi: Prashanth, I have a question for you on this topic, because in a previous interview you said that the moment inside Stack Overflow when ChatGPT was emerging was “code red,” because it was hitting your industry and your business really hard. As a leader of this company, what did that disruption teach you? How did you pivot?

Prashanth Chandrasekar: Absolutely. We’ve been around for 18 years and were definitely the most dominant knowledge base for technology throughout that period. Then in 2022, when ChatGPT came out, we realized that all the data we had generated through human contributions on our community of 100 million people was now being used by them and other AI companies. So it was very much a code red moment. It was absolutely jarring, because you have to figure out what the mission and purpose of your organization is. Why do you exist? What value do you provide to society, to your users, and to the companies you serve? It was very much a refounding moment for the company.

I had been through this a couple of other times. I used to work at Rackspace, and back in 2012 Amazon Web Services disrupted the traditional managed hosting industry. That was a similar moment where you had to figure out how to get out of that as an incumbent. One of the ways you do that is to carve out a very autonomous team that is purpose-built. Think of it as a brand-new company inside your company, with none of the pressures of the existing business, going to disrupt yourself.

That’s effectively what we did at Stack Overflow when ChatGPT came out. We carved out a team of about 10% of the company and gave that team complete autonomy to go after it. We measured them completely differently compared to the rest of the organization, with different incentives, and told them to think about how we were going to use AI to solve customer problems and user problems, not just use AI for the sake of using it. We also put a stake in the ground and said that by the summer of 2023, we were going to announce to our user base what we were actually doing with AI.

So we had a sprint from January to July where this team of about 10% of the company was cranking away every week, and we said, “Don’t come out until you actually figure it out.” I personally ran product at that time also, because I think it is very difficult to delegate something as fundamental as this. You have to be in the trenches with your people. So it was a lot of change and a lot of disruption. It’s not for everybody. It’s not for the faint of heart. If you’re not going to embrace doing things differently and being adaptive, then you’re going to be in trouble.

Since then, we announced our AI products. We also went and partnered with every single AI lab that had leveraged our data and wanted to leverage our data. We did partnerships with OpenAI, Google, and all the big cloud hyperscalers. We now have data licensing partnerships with them where they pay us for access to our corpus of data on the public platform for historical and ongoing access. Financially, that has put us in excellent shape.

Then in our enterprise business, over the past 18 months, we started noticing that the APIs of Stack Overflow for Teams, which is a private Stack Overflow inside the largest organizations and companies, were red-hot inside companies. When we went to talk to them, they said they were using it to power their AI agent experiments and assistants that were answering Slack channels and support tickets. That gave us the insight that we should be building the next-generation version of that product meant for AI agents.

So over the past 18 months we built that, and we launched it this past year as something called Stack Internal, which is the knowledge intelligence layer that companies use to hoover up relevant context from agent interactions, Slack interactions, documents, and other sources, and produce valuable context that can then be reused by other agents. We believe we can be the authoritative context layer for enterprises to use AI agents.

And we’re not done yet. This world is continuing to accelerate. It is only the early innings. I think we’re talking less than 10% adoption across the world. This technology is here to stay, and there are going to be multiple more waves on this topic.

KC Ifeanyi: Building off that, I want to hear from Sam and Peter. Off the example that Prashanth just laid out, how can leaders make confident decisions when the ground beneath them is always moving?

Peter Smart: Leaders are as human as everyone in this room. They are as fallible. They don’t know definitively where the future is going. Prashanth doesn’t, Sam doesn’t, I don’t. But it is your job to figure it out. As a leader, the first thing you have to recognize is that you cannot defer that weight of responsibility. It is your job to be the individual, or one of the individuals, creating the conditions for clarity and coherence for your company. That is job number one.

Job number two is to lead with empathy and say that in figuring this out, we’re going to be answering some very hard problems and difficult challenges, which might fundamentally rethink the way that we deliver output today, the value that we produce today, what we’re putting out into the world today, and the jobs required to do that. At the same time, you have to be committed to developing a new vision for your organization that serves human beings with the same purpose as before, but in a brand-new way.

The advice I always give to leaders, and Prashanth is a great example of this, is to start small. At Fantasy, we’ve done a very similar thing. We’ve siphoned off a portion of the organization where there are no rules and no relying on past ways of doing things. The question is: Here is the value that we have as this pinnacle digital product team, but what is that going to mean two years from now? How will we deliver it? Let’s go figure it out together.

Starting with a really small group of people is important. Number one, they need awareness of what’s actually happening with the technology. Sometimes it’s very difficult to source that inside your organization, and sometimes you actually want to bring in experts from the outside to help you figure out what’s happening and to understand the adjacencies of what’s happening. I give the example that Nike and One Medical in the future may be way closer to each other than they are today. Have they realized it yet? Maybe not. But the locus of value and their purposes in the world are so related.

Then you need a group of individuals for whom, if the answer is that in this new future my job doesn’t exist anymore, they’re okay with that, because they have the resilience to understand that the value I produce today isn’t who I am. What I actually output is not who I am. It’s my ability to creatively problem-solve, think differently, and bring the inimitable human ability to conceive of the new. I’m just going to apply that mindset to new problems.

If you can create that really small group of people, then you create the conditions where there are no wrong answers, and you can come up with hypotheses. I think there’s a big difference between hypotheses and vision. Vision, where it’s unverified, is dangerous. There are a lot of executives in the media right now talking about how their industry is going to be transformed with nothing to back it up, and they’re talking about pipe dreams in which they have imbued AI with this infallible ability to do everything. That’s very dangerous.

What you need as a leader are hypotheses for where your company is going, so that you can validate those hypotheses very quickly and then confidently go to your organization. And you don’t do it all at the same time. You do it in cascading increments of organizational change. I talk about something called the three Rs: reorganize, rally, and realize. You start with that really small team, let them figure something out, validate that it’s actually a viable path, treat it like a pilot, and then apply it to the next wave of your organization. You don’t do it all at one time. Your organization will experience ontological shock, and you don’t have all the answers yet. You have to do it incrementally.

I see leaders really struggling to create these conditions, and I would bring it back to one word: coherence. Bring it back to clarity and coherence for your team. That starts with you as a leader, and with bringing in external perspectives and people within your organization who might look very different today but have the mindset to work on these novel challenges.

Sam Jordan: To do that is easier said than done. One practical thing is that you have to change the incentive structure, because right now leaders are not incentivized to do that. They are not incentivized to have hypotheses and test them. They are incentivized to meet what their shareholders expect of them. So the incentives have to change. You have to figure out ways to reward experimentation and testing.

The other thing I would say is that when things feel really chaotic, it usually means that something much more structural and fundamental is shifting. I think of the decade of the 1960s. It was so chaotic. We had Vietnam, multiple wars, the Cuban missile crisis, presidents being assassinated, and we went to the moon. There were so many things happening on the surface, it was easy to get distracted.

But the companies that won realized that it wasn’t those headlines they had to be paying attention to. It was the fundamental order of work and identity that was changing. To give one example, the public reaction to Vietnam was very different from the public reaction to Korea, and one of the reasons was the introduction of television. Television changed how people consumed media, so that pipeline of trust between institutions and the public, governments, and news organizations started to change.

I feel like now feels a bit like the 1960s, in that there is so much chaos and so much happening on the surface. So in addition to doing what Peter just said, and in addition to changing the incentive structures, you have to get really disciplined about asking: What is actually changing? What is the thing that is actually shifting?

KC Ifeanyi: Sam, I want to stick with you for a second because I know later at SXSW you’re giving a talk about how leaders are kind of running backward. Historically, industries have followed a very linear process of research, building, and testing, and now we have AI modeling and digital prototyping, so that sequence is inverted. How can leaders rethink their workflows when experimentation is happening in a completely different way?

Sam Jordan: Thank you for the plug. I hope you will all come to my speech.

One of the things I’ll be talking about is that I’ve identified three pipelines that I think are shifting now. One is the way that we build things. The order in which we build things is changing because of technologies like our ability to simulate in virtual reality. We’re getting really good at that, which means learning can happen so much earlier. That changes the build pipeline, which changes the skills and the people that are required.

Another pipeline that’s changing is the discovery pipeline. How we are discovering things about the world is changing.

And the last pipeline, which is probably the most important, is the talent pipeline. There has been a very specific order through most of human history for how you cultivate talent. You have a junior person do the grunt work. They make mistakes, those mistakes get corrected, and eventually they build character and skills. Once you build character and skills, then you can be a leader.

But as we are seeing right now, so much of this work is being automated. I’m not just worried that we’re going to lose some of these skills. We’re hollowing out the skills for later leadership, but we’re also hollowing out some friction points that were important to creating character for leaders.

The argument is not to forego the new pipeline. The argument is not to slow down and not adopt the new technology. You absolutely should adopt the technology. The question I would ask leaders to start thinking about is: What are some of the byproducts of those old pipelines that we’re going to lose with technology? Sometimes friction is good. Sometimes friction creates character. Sometimes efficiency is not always optimal. So how do you put those unintended byproducts back into the new technology pipelines?

KC Ifeanyi: In the two minutes we have left, I want each of you to give me the top lesson in how leaders can stay agile. We’ve covered a lot of ground, but I’d love for everyone to leave with a solid takeaway. What is the top lesson, the top tenet, on how leaders can stay agile during disruption?

Prashanth Chandrasekar: Generally speaking, the rate of change is extremely high. I can’t remember a time when it’s been this dynamic. So staying very agile means having a constant learner mindset and being flexible in your point of view. Loosely held points of view are good, because things could change. Fundamental assumptions could massively change very quickly because the rate of innovation with chips, the rate of innovation around AI models, and hallucination rates getting lower are all moving fast. You’d have to be some kind of soothsayer to know where the world is going to be even a year from now. That’s extremely unnerving, and the ground is constantly shifting.

The best way to do that is to be rooted in your mission and your purpose. Why are you here on this earth as a company? How do you serve your user base and your customer base? And make sure that mission is still relevant. You may have to refound that mission and not be scared to do it. It’s going to be noisy. It’s going to be messy. You might say, “We’ve been that way for 30 years.” But do you want another chapter or not? I think being brave and fearless enough to do that is very important.

We had a mantra at Stack Overflow this past year around being fast, focused, and fearless. Fast because things are constantly changing, so how do you incorporate those daily learnings into your thesis and your hypothesis? Focused because you are constantly updating your assumptions. And fearless because you may have to change completely and redefine your company while you’re learning about all these new things.

It’s a mentality shift. Carol Dweck’s work on growth mindset is a very important book to read in this moment, because it tells you it’s nothing personal. It’s not about you. You have to be okay with being wrong because a lot of assumptions are going to be proven untrue. You need to roll with the punches, keep learning, and keep moving forward with resilience.

Sam Jordan: Do not forego judgment for efficiency.

Peter Smart: We live in existential times across every single company and industry. It is incredible to me that a team of three people using AI video models can compete with a major Hollywood studio and have their content bought by Netflix. The entire order of companies, the way in which they deliver value, and the business models they’ve been built on for the last few decades or longer are being rewritten in front of us.

The number one thing that separates organizations that are going to navigate this moment well and those who won’t will be coherence. The size of your team is no longer a benefit. How well-resourced you are is no longer a benefit. The new agility is coherence. Can you create the conditions by which it is very clear what we do, what the value is that we produce, and how we’re going to get there?

That’s why teams of three people are superseding teams of 3,000 people, because coherence is much easier with a smaller group of people. As a leader, there are playbooks for this, but creating the coherence of your organization, the clarity of vision, and the frame by which you invite your organization to start accelerating toward that future are what matter. I think probably all of us in this room are in one of those two companies. Look for the companies, like Stack Overflow and others, where there is that coherence. And if you don’t have it in your organization today, fight for it. Bring in help to support you in creating that coherence. It doesn’t take a long time, but it’s going to be the difference between existential crisis and success, and between crisis and collapse.

KC Ifeanyi: Brilliant. Thank you so much, everyone. I really appreciate it. Give it up for this amazing panel.