24 May 2025 · 36m · Bob Safian, Aaron Levie · 21 quoted facts
Box CEO Aaron Levie argues that AI agents, which are AI programs that work through a task over several minutes using tools, are the next big shift for businesses. He says the real impact is not replacing workers but making affordable the kinds of work companies never funded before. He also argues that being 'AI first' is about speed and reinvestment, not cutting headcount.
The tension is between the breakneck speed of AI change and the slow pace of enterprise adoption. Levie says he has never seen technology move this fast, with staff working nights and weekends to keep up, yet he expects company-wide adoption to take a decade, as cloud computing has. He ties this to policy, criticizing tariffs and immigration limits while praising the administration's open-source and light-regulation stance on AI.
Levie says about 80% of his customer AI conversations in one quarter were about work the company never did before. His example is an agent scanning 50,000 contracts for upsell prospects for roughly $5,000, a job nobody would have paid staff to do.
GitHub Copilot launched a year or two before ChatGPT, so coders had their 'ChatGPT moment' 12 to 18 months early. Levie expects tools like Cursor to show ordinary workers what agents will do for them within one to two years.
Levie argues that a sales team that sells 5% more will usually be given more money to grow, not just be counted as extra profit. He says firms that grab the profit instead risk being outflanked by competitors.
Levie contrasts Box with the Duolingo-style rule that teams must prove AI can't do a job before hiring. He says that rule creates a perverse incentive, so Box gives more resources to teams that show they use AI well.
Levie points out that cloud computing is still growing at its usual quarterly rate 20 years on, and banks still announce new cloud deployments. He expects AI to move faster but still need long change management, which is why he pushes AI-first habits early.
Levie describes agents as AI models that think for longer, use tools and double-check their own answers, instead of giving a quick answer in a few seconds. He says deep research features, which take 5 to 10 minutes, are the first agent-style tool most office workers will meet.
Levie says most of what customers describe is work they never did before, not work moved from people to AI, and that the resulting insights tend to create more work for humans. He admits this is based on his own customer conversations, not on a study.
For Levie, it means using AI to speed up customer outcomes, decisions and product building, treating work as a timeline to move faster along rather than a pile to shrink. At Box, teams that prove they use AI well are the ones that get extra resources.
He says both artificially raise costs and slow companies down against competitors, calling himself 'anti cost increase'. He called tariffs 'an economic disaster' but also said he gives the administration fairly high marks on AI policy.
Levie says the next 12 months are when businesses get their first real proofs of concept with agents, but full enterprise adoption will be a decade-long journey. He compares it to cloud computing, which is still being rolled out after 20 years.
Stated by the speakers. Each quote matches the transcript, which shows what was said, not that it is true.
Aaron Levie says GitHub Copilot was introduced about one to two years before ChatGPT, so developers reached their breakthrough moment with AI earlier than the general public.
“GitHub Copilot was introduced about a year or two before ChatGPT, so developers actually had their ChatGPT moment”
Aaron Levie says that, by every statistic he has seen, Cursor is the fastest-growing software company in history, at least in an enterprise context.
“As far as every stat I've seen, cursor is the fastest growing software company in history, at least in an enterprise context”
Aaron Levie says deep research features in ChatGPT, Grok, Perplexity and Anthropic's Claude are the first agent-style use case that regular knowledge workers will meet.
“something like deep research inside of ChatGPT or inside of Grok or inside of Perplexity”
Aaron Levie says deep research tools take about five to ten minutes of thinking time per answer, compared with a few seconds for a typical chatbot reply.
“takes about 5 or 10 minutes of kind of thinking time for these AI models”
Bob Safian says Salesforce CEO Mark Benioff told him that use of Salesforce's AI agents had not taken off as hoped.
“I talked with Mark Benioff at Salesforce several months ago about his embrace of AI agents, but the use of his agents hasn't quite taken off the way he hoped.”
Box launched Box AI Studio, a product that helps organizations build their own custom AI agents.
“I know you launched Box AI Studio to help organizations build their own custom AI agents, and I'm curious how how that's going.”
Aaron Levie estimates he had at least 100 customer interactions in the first quarter of the year, with about 80% of AI discussions focused on work the company did not do before.
“I've had at least 100 interactions with customers in in the first quarter of this year. The vast majority, you know, 80%, I'm guessing”
Aaron Levie gives a hypothetical in which an AI agent reviews 50,000 customer contracts for about $5,000, a task companies would not previously have paid people to do.
“But now, if it only costs them $5,000 for an AI agent to go do that, they would do that all day long.”
Bob Safian says Shopify and Duolingo have announced that staff must justify work that is not produced with AI.
“Are you like Shopify and Duolingo, who've announced that staffers have to justify anything that's not AI produced?”
Aaron Levie describes the Duolingo memo as an emerging idea that teams must prove AI cannot do a job before they are given new headcount.
“There's some emerging idea, which is sort of you have to prove that AI can't do this thing for for you to get then headcount.”
Aaron Levie says Box's approach is the opposite of the Duolingo-style rule: teams that show they can use AI get additional headcount.
“If you can prove that you can use AI, then that's actually when you will get headcount.”
Aaron Levie says there are arguably five to six companies competing in AI models, any of which may briefly hold the leading model.
“there's arguably 5 to 6 players, all of which you could throw a dart on a dartboard and they will have the leading AI model”
Aaron Levie says earlier platform contests, such as operating system and device wars, usually had only two to three credible competitors.
“And even then, there's usually only 2 to 3 credible players that that kinda take the attention of the market.”
Aaron Levie says he has worked in technology for 20 years, beginning at the start of cloud computing and through the iPhone and iPad eras.
“Having done this now for 20 years, uh, starting right at the at the start of the cloud, seeing the iPhone, seeing the iPad”
Bob Safian put Aaron Levie on the cover of Fast Company about a decade before this interview.
“I put Aaron on the cover of Fast Company a decade ago to represent how a generation of new leaders were embracing adaptability.”
Aaron Levie argues that a punitive tariff above roughly 30% raises the cost of serving an industry for that country's exporters, which no competitor faces.
“take any kind of punitive tariff above, I don't know, 30% on a country. You've now inflated your cost of serving”
Aaron Levie says the Trump administration's AI approach has some positives over the Biden administration's, including a bias toward supporting open-source AI.
“there are some incremental positives relative to the Biden administration. So so there there's a general bias toward supporting things like open source AI.”
Aaron Levie proposes giving graduates studying AI or computer science in the U.S. a green card with their diploma, and says neither the Trump nor Biden administration did this.
“What you would do is you would just do the whole green card stamped on the diploma, uh, after college”
Aaron Levie says cloud computing is about 20 years old and its quarterly adoption rate is still growing at the same pace as before.
“We're 20 years into cloud computing, and and the and the quarterly rate of cloud computing sort of adoption is still growing at the same rates as ever before.”
Aaron Levie says banks still announce new cloud deployments, for example for trading platforms, 20 years into the cloud era.
“When I go to a bank, they'll still say we just deployed cloud for this trading platform.”
Bob Safian says his Fast Company cover story featuring Aaron Levie described 'generation flux', people defined by their willingness to build new skills amid uncertainty rather than by age.
“it talked about a concept I call generation flux, a group of people not defined by chronological age, but by the willingness to jump in and build new skills in an environment of uncertainty.”
Levie explains how Box moved from asking questions of company files to agents that think longer and use tools. This sets up the episode's main theme.
“there was this sort of emerging concept of agents”
Levie rejects the idea that only infrastructure sellers profit, pointing to fast-growing AI app companies and enterprise uses like contract review.
“some of the fastest growing companies in history get built on this”
Coding tools got their breakthrough moment before ChatGPT, and deep research tools are the first agents ordinary workers will use.
“developers actually had their ChatGPT moment”
Levie argues most enterprise AI use is work that was unaffordable before, so it expands what firms do rather than replacing people.
“spent on things that the company didn't do before AI”
Levie defines AI first as speed and says Box rewards teams that prove AI use with more headcount, unlike the Duolingo-style approach.
“think about work as a timeline and not a mass”
Levie describes five or six rival model makers leapfrogging each other and the late-night scrambles needed to keep up.
“absolutely the fastest I've ever seen tech move by at least 5 to 10x”
Levie explains why he opposes tariffs and regulation alike, and how he separates his personal views from Box's.
“Tariffs are a way of of increasing the cost.”
Levie is less afraid of China than many, rates the administration's AI policy fairly well, and argues that keeping foreign talent helps U.S. jobs.
“I'm weirdly less afraid of China.”
He argues that skilled graduates will build AI somewhere, and the country that keeps them gains the jobs over time.
“You want the Sergey Brins to to to build that company here”
Looking 12 months ahead, Levie expects first proofs of concept with agents but a long road to full adoption, as with cloud computing.
“this is gonna be a decade-long journey”
Stripe is built for whatever comes next because AI isn't just changing business, it's evolving fraud. Stripe Radar detects fraud with the fast scale of Stripe's 1.9 trillion dollar annual transaction volume. In fact, when a card is used, there's a 92% chance the Stripe network has seen it before. That means emerging fraud patterns are detected sooner, blocking bad actors, welcoming good customers. Learn more about fighting fraud with Stripe at stripe. Com radar.
On Masters of Scale, iconic leaders reveal how they've beaten the odds.
Asking really strong questions is a superpower.
You want to show up with something radically different and how they've grown companies to incredible heights.
The greatest rewards always come from the greatest risks.
Let's hit the gas.
Airbnb, Zillow, Microsoft, Liquid Death, and more. Hear from the founders who've changed the game. It's anything but business as usual. Find Masters of Scale on Apple Podcasts, Spotify, YouTube, or wherever else you get podcasts.
The business world is moving faster than ever, and when change hits, we need to learn in real time. On Rapid Response, you'll hear candid conversations with CEOs and leaders making tough calls about AI, human talent, responsibility, and the bottom line, how they navigate uncertainty, pressure, and high-stakes moments. I'm Bob Safian, former editor in chief of Fast Company, and I'll be your host as each episode breaks down what you need to know right now. You can find Rapid Response wherever you get your podcasts.
Hey listeners, we wanted to share an episode with you from our sister podcast, Rapid Response. Host Bob Safian recently sat down with the CEO of Box.com, Aaron Levie, who's evolved Box from a cloud storage company to an AI forward enterprise. The conversation dives deep into how all types and sizes of organizations can approach adopting AI and get more value from it across departments. Leaders of all types will wanna hear these insights on transforming their enterprise. We hope you enjoy it.
Having done this now for 20 years, starting right at the at the start of the cloud, seeing the iPhone, seeing the iPad, this is absolutely the fastest I've ever seen tech move by at least 5 to 10x. Every week, something is happening that is causing this category to shift in some direction. Something will happen at 10 pM and there'll be like a a group chat with with AI friends, everybody's blowing up like, Oh my god, I can't believe this just happened.
And then it'll be an internal Slack with, you know, 15 people on the AI team of, Oh god, oh gosh, we have to jump on this. And it's just like, hope everybody has, you know, had their family dinner and that's all behind them, cause now we're we're back wired in on, uh, on what's happening.
That's Aaron Levie, founder and CEO of Box. I put Aaron on the cover of Fast Company a decade ago to represent how a generation of new leaders were embracing adaptability. I wanted to go back to Aaron now because his cloud storage firm is increasingly an AI firm, putting him at the heart of a new wave of chaotic change. Aaron gives us an insider's roadmap on where AI is moving for enterprises, why there's so much buzz about AI agents, and what it means to be an AI first organization.
Aaron isn't shy about calling things as he sees them, whether that's mistakes others are making with AI or critiques of both Trump and Biden. So let's get to it. I'm Bob Safian, and this is Rapid Response. I'm Bob Safian. I'm here with Aaron Levie, CEO of Cloud Storage and Tools Platform Box. Aaron, great to see you. Thanks for being here.
Thanks, Bob. Good to, uh, good to be here.
So I went to your LinkedIn page just earlier and I scrolled down your posts over the last few months, and it's like, AI, AI, AI, then AI agents, AI agents, AI agents. You are you are all in, aren't you?
All in, all in. So, we, w- we when we saw the the you know, everybody kinda had their chatGPT moment and ours was like, wow, all of a sudden, we're gonna be able to use AI on top of all of this unstructured data that we have. So, in Box, it's contracts and invoices and marketing assets and research data. And so, the the initial breakthrough use case was, okay, we can now talk to our data. We can ask it questions, we can get insights from all this information.
And then, somewhere about last year, there was this sort of emerging concept of agents, which is, you know, what if you actually had these AI models think longer? Uh, they they they could use tools, they could go back and and check it, double check an answer, or kind of apply more judgment later later in the in the stream of work. And so, we jumped full on on that, and so, we've we're we're kinda all in on, uh, on AI agents right now.
Now I got I gotta ask you, some folks murmur that AI's business impact isn't quite living up to the promise yet. You know, the sellers of the picks and shovels of AI are making money, but few prospectors have struck gold with AI tools, practically, for real. Where like, where are we on this effectiveness curve for for AI?
Yeah, I I would take the other side of that that argument, the prospectors are, let's say, cursor or windsor for these AI products that that automatically generate marketing, you know, content for you, or Figma generating, you know, design assets and and web pages. I think you certainly have the picks and shovels, all of the the hyperscalers, you know, that that's kind of always gonna be the the first phase of the rollout. But the prospectors, you know, in this analogy of of as the app layer, I think we're seeing some of the fastest growing companies in history get built on this this AI foundation.
As far as every stat I've seen, cursor is the fastest growing software company in history, at least in an enterprise context, and that is 100% due to having AI agents right next to you while you're coding, that can, you know, accelerate your code development. On the enterprise front, you know, in the customers we talk to, they are doing everything from saying, Hey, I wanna take, uh, you know, 100,000 contracts and automatically review them with AI and extract the most relevant data from those contracts.
Or I wanna take all of my sales presentations that I ever created, and I want new employees to be able to ask them questions so they can get the level of expertise of of an existing employee that's been at that company for a number of years. And so, this is already starting to to roll out in in large enterprises. You know, the amount of time it takes for these AI products to get incorporated into people's workflows naturally takes some time.
I remember having a conversation with a CEO who, uh, like you, is is a a a tech person. And and they were like, I spent two weeks doing nothing but coding, and it just blew my mind about what was possible. And I was like, Well, I'm not a coder. How do I actually feel that? And I think it's it's sometimes hard for folks to sort of see that leverage because that layer isn't quite here yet.
I think that's exactly right. GitHub Copilot was introduced about a year or two before ChatGPT, so developers actually had their ChatGPT moment at at least 12 to 18 month months before everybody had their ChatGPT moment on the consumer side. And so, very similarly, Cursor, Windsurf, Replit, these products are now having their AI agent moments, and that is happening a year or a year and a half or 2 years before then the the general population is gonna have their AI agent moment.
Now, you can already see the the very initial versions of this. So, if you take something like deep research inside of ChatGPT or inside of Grok or inside of Perplexity, um, or or Anthropic, or or Claude from Anthropic, you know, this is the first agentic use case that I think the the kind of regular knowledge worker will run into, which is okay. I don't just want a very quick answer to a question about, you know, some historical event.
I really want the AI to go out and research, you know, 100 different companies and give me all of the best practices in my particular field for for a specific, you know, type of topic. I wanna get the marketing, you know, strategies of my, uh, my the the 100 companies in my in my industry. So, this is sort of deep research, and what's happening is the AI isn't just kind of very quickly, in 3 seconds, coming up with an answer where there's the risk of hallucination or where it hasn't had enough time to kind of, uh, go across the entire internet and find up-to-date sources for all of the questions.
The ChatGPT moment was was these very quick answers to moderately complex questions. The deep research moment of of this paradigm is is much longer answers, i in terms of the amount of content you get, takes about 5 or 10 minutes of kind of thinking time for these AI models. And they're and they're going across, you know, the entire up-to-date, you know, corpus of the world, which is the internet, to go get the answer. And and so I'm using this, you know, almost on a daily basis for, you know, every kind of question imaginable in the company.
I want I wanna figure out the pricing strategy for everybody's AI products. And so late at night, I'll be, you know, kind of alone working. There's not, there's not somebody I can just go ping that says, Hey, can you go do this research for me? And, you know, 15 minutes later, it it has it's in my queue, and it has come up with the a full research document of every single company in our space and what they're all doing for pricing.
And that just moves the project that I'm working on one step forward faster than than, you know, by days, faster than it would've been in a kind of non-AI world. And so, I think what what will happen is as these types of products become more pervasive, everybody will have versions of that in their daily work, where you're just gonna be sending tasks off to AI agents that are just working on your behalf as as you would've if you if everybody had a research assistant that was fully at their disposal for any kind of project work that they wanted.
I talked with Mark Benioff at Salesforce several months ago about his embrace of AI agents, but the use of his agents hasn't quite taken off the way he hoped. I know you launched Box AI Studio to help organizations build their own custom AI agents, and I'm curious how how that's going.
Yeah, uh, s so far, it's it's either at or exceeding our expectations on all the use cases that that customers are coming up with. So, we're we're pretty blown ab- away about the what what we're starting to see. We're still very early days, to be clear, but the but the rate of adoption is is, um, is going fairly exponential. And the the imagination that customers, uh, uh, now have on this is is blowing us away. So,
You mean uses you didn't anticipate for for the way they these these agents work?
yeah, I I've ne- I have I've rarely been in a customer conversation, either 1 on 1 or in a dinner, where I'm not hearing about a new idea that the customer has for Box AI that we did not have on a whiteboard. And what's what's exciting, and this is counterintuitive, I think, to a lot of folks outside of AI, you know, you you initially sort of see AI in sci-fi and and sometimes in the, you know, news headlines, New York Times or whatever, as like, okay, it's going after jobs, it's gonna replace these types of work.
From my anecdotes, I've had at least 100 interactions with customers in in the first quarter of this year. The vast majority, you know, 80%, I'm guessing, the bulk of the time of AI use case, kind of, you know, conversation was spent on things that the company didn't do before AI. So, it wasn't, Hey, I want to take this type of work and I want AI to go replace it. It was, there's a s there's a type of work that we never get around to in our company.
I want AI to go and do that, finally, because finally, it's affordable for me to deploy AI agents at the kind of work that we just could not, we could not fund before. It's opening up people's imagination to, Hey, I'm like sitting on, you know, 50,000 customer contracts. What if I could have an AI agent go around all those customer contracts and figure out which customers have the highest propensity to buy this next product from me?
And this is not something that they would have people ever do. So, it's not replacing anybody's job. They never they never said, Oh, let's have 50 people go read all the contracts again. It it just never happened. But now, if it only costs them $5,000 for an AI agent to go do that, they would do that all day long. And so, and then, guess what? When they get those insights, they're probably gonna have them more work for the humans in their business to go and do as a result of this.
That hopefully, if it's effective, drives more growth in their business, which then causes even more productivity and then ultimately hiring and growth. And so so so I it's not, it's not kind of everybody's first instinct, but most of the use cases that we're hearing about are things where because it is now affordable to deploy AI at a problem, I'm actually expanding the set of things my company can go do and and then the work that we can now execute on.
And that's that's not only very, you know, I think, exciting, but I think it's gonna be the default case for most AI adoption in the enterprise.
It's interesting cause in some of the conversations that I have, it feels almost like some of the businesses and leaders, they don't really know what they're looking for from AI. And and it's it's and hearing you, it sounds a little bit like you have to think about your mindset on it a little differently to to open up and find those things that are most valuable to you.
Yes, yeah, every business is gonna be different because because, uh, some of the upside is a virtue of your business model. What are the core, you know, parts of your business model that, as a result of access to information, can change or be modified or improved? If I am a law firm, I could either reduce my my my cost because now AI is gonna do more of the, let's say, paralegal work, or I could expand my service offerings because now all of a sudden, my my team can can, you know, venture into more domains because they can take their expertise and use AI to augment that.
The default assumption is, Oh, no, it's gonna go after the hours of a law firm. But but once this technology hits an individual business, they can actually decide to expand their customer base. They can go after they can go after previously customers that would have been unprofitable for them to serve. So, these these industries are not a static and zero-sum, the software industry. Y- you know, on one hand, everybody says, okay, if AI can do coding, then will we hire fewer engineers?
And in general, my argument is that we'll probably hire as many, if not more engineers, if AI can get really good at coding. Because what will happen is the productivity rate of our engineer goes up, which means that, uh, that that we can then, you know, ascribe a higher degree of value per engineer in the company.
So your ROI is even better on each of those each of those positions.
and e- exactly, and and and take something like sales. If we can make a sales rep able to sell 5% more because we give them better data and they can prepare for a customer meeting that much better, or they can they can, you know, understand exactly the best pitch because because they have access to all of Box's data and they can ask it questions. I'm not gonna just bank that as 5% more profit because what will happen is we're gonna internally, in some in some planning session, we're gonna get greedy, and we're gonna say, wait a second, that 5% gain that we just got in sales productivity, what if we reinvested that back into the sales team to grow even
faster and get that much more market share? And so, you have an entire economy of company making the uh, uh, of companies making those individual decisions of, do you bank the profit or do you use it to go and accelerate growth? And what we what we tend to know from history is that the companies that get too greedy on the profit side, you just end up leaving yourself vulnerable to being outflanked by competitors. So, you know, capitalism has a pretty pretty, you know, convenient way of almost driving the the sort of productivity gains of these types of innovations to get reinvested back into the business.
You've been talking about running box in an AI first way and and encouraging other leaders to do it. Are you like Shopify and Duolingo, who've announced that staffers have to justify anything that's not AI produced? Like, what what does AI first mean?
Yeah, so for us, AI first means that we wanna use AI as a means of driving an acceleration of the customer outcome, an acceleration of decision making, acceleration of building new features. So just think about it as mostly a metric of speed. On one hand, you could think about AI as going after, like, a mass of work, and you could say, AI is gonna, you know, remove some part of that mass of work and do it instantly, so the mass of work goes down.
Or think about work as a timeline and not a mass. All we're doing is trying to get through each step so that way we can get to the next step, and then and and so on, and And everything's everything's all the and faster. everything's faster. So I wanna have us use AI to move faster down the timeline, not just purely to reduce the total mass of work that we're doing. There's probably one pronounced difference versus, let's say, the Duolingo memo.
There's some emerging idea, which is sort of you have to prove that AI can't do this thing for for you to get then headcount. And and my my general our general instinct is actually the opposite. If you can prove that you can use AI, then that's actually when you will get headcount. Because what we want is we want the dollars of the business to go back into the areas that are the increasing areas of productivity gain, because those areas will then be har higher ROI for us over time.
Hm. I worked for a while, the place where our our whole job was to create the piles of cash for us to bring over to someone else. And it it it was a frustrating way to be.
What you don't want is you don't want you don't wanna develop any, um, kind of accidental, uh, disincentive, right, for for AI adoption. If you kind of make it so when AI can't do something for you, that's the you you kind of get this kind of benefit of more budget. That's obviously, you know, would create, you know, a perverse incentive. You'd actually rather just say, okay, we want every org to be AI first, and the ones that are even more AI first will be the areas that we probably prioritize, you know, with with additional resources.
I love how Aaron meshes his understanding of AI development with business realities, incentives, investments, how greed can propel both missteps and opportunities. I also love his personal stories about how he uses AI and how specific conversations with clients are shifting his priorities. So what kind of best practices does he offer for keeping up to date on AI shifts, and is China really an AI threat or not? We'll talk about that and more after the break. Stay with us.
The companies winning right now aren't just building great AI, they're building the infrastructure to actually run a business on top of it, which is where Stripe comes in. Stripe is built for whatever comes next, with planetary scale financial infrastructure to help you power any business model. As a matter of fact, 86% of the Forbes AI 50 run on Stripe, along with industry leaders like Salesforce and Pepsi. Last year, Stripe users processed $1.9 1 trillion combined. See how Stripe can power what's next for your business at stripe.
Before the break, Aaron Levie of Box talked about how AI and AI agents are poised to change enterprises. Now he talks about what it takes to keep up with the constant flood of AI advances, the risks posed by China and by tariffs, and how AI will change our society in the next 12 months. Let's dive back in. You guys work with with everyone, with all the models, OpenAI, Google, Amazon, new models, new products come out like every day, like constantly.
How how do you personally stay caught up?
This part has exceeded my, uh, my highest expectations of the amount of of of just d dynamism in the space. Our only historical context for this many platforms competing are like in these early stages of OS wars, device wars, and so on. And even then, there's usually only 2 to 3 credible players that that kinda take the attention of the market. We have somehow ended up in a category where there's arguably 5 to 6 players, all of which you could throw a dart on a dartboard and they will have the leading AI model for some at least brief kind of period of time.
And and the the leapfrogging is sort of happening at no greater pace than than on a monthly basis. So so that means that you have every week something is happening that is causing, you know, this category to shift in some direction. You could possibly just say, you know what? Then I'm gonna just like kinda not pay attention because it's moving so fast. We don't have that, we don't have that, um, uh, that benefit, cause because our customers rely on us to know what what the heck is going on.
So then you have to do the inverse, which is we're just, you know, drinking from the fire hose of constant information, uh, coming at us. I mean, we'll we'll be on at Saturday at 10:00 PM. Preparing for a Monday release of an AI model because we're we're doing the finishing touches of our, you know, evaluation of that model, and we wanna get it in the hands of customers, and we wanna make sure that everybody knows the benchmarks of it.
Having done this now for 20 years, uh, starting right at the at the start of the cloud, seeing the iPhone, seeing the iPad, this is absolutely the fastest I've ever seen tech move by at least 5 to 10x. The craziest I've ever seen this industry and the hardest I think anybody's had to work just to stay up to speed with everything going on.
Hm. And d and do you have a a strategy about how you personally do that, or you just don't sleep?
It is possible to get 8 hours of sleep in this environment, but it's not possible to have any hobbies. So so so you you you can either choose hobbies or sleep, and most most people are probably choosing sleep over hobbies. But, like, something will happen at 10:00 PM and there'll be like a a group chat with with AI friends, everybody's blowing up like, Oh my god, I can't believe this just happened. And then it'll be an internal Slack with, you know, 15 people on the AI team of, Oh god, oh gosh, we have to jump on this.
And it's just like, you know, hope everybody has, you know, had their family dinner and that's all behind them, cause now we're we're back, we're we're back wired in on, uh, on what's happening.
And and y- you're not gonna, uh, spend 2 hours working out that, uh, that next morning, you're just you just gotta get right back to it.
No, I think if you have like 12 minutes to lift some weights, you're good. So,
Um, a amid all of this rapid and radical technology change, of course, w- we're seeing a lot of other cultural change. Another topic I saw a lot in your social feeds, I guess, on X, is tariffs. Now y- you're in digital products, so theoretically, you'd feel less whiplash about tariffs, but you've gone right at the Trump administration about them. Why is is that because of what you hear from clients?
the reason why I was outspoken a against them is I I just, like, fundamentally don't like tariffs, you know, b- box or not box. You uh, I I think that the the the market does better, uh, with as few kind of market distortions as possible, or economic distortions as possible. Unfortunately, things are seem to be trending in in a in a kinda winding down direction. But take any kind of punitive tariff above, I don't know, 30% on a country.
You've now inflated your cost of serving in a particular category or industry. So all of a sudden, the goods and services that you're building, that coming out of your country now have this this built-in cost increase that no other competitor deals with. Does that increase the odds that you win in a particular industry or decrease the odds? Anything that creates a a an artificial increase in cost, by definition, you know, causes you to slow down relative to your alternative competitors, which is, you know, e- equally, I'm very much on the other side of the spectrum of normally my my Democrat views on things like regulation, because regulation is equally an artificial increase in the cost of executing in a particular
industry. Tariffs are a way of of increasing the cost. Regulations are a way of increasing the cost. I'm anti cost increase. I I I certainly have found myself in this environment being more libertarian than the people that I thought were libertarian. And, uh, I guess that's one one sort of virtue of, uh, of of the Trump administration so far.
Yeah, I mean, there there's been talk about tech leaders sort of bending the knee to Trump, starting with the inauguration, maybe maybe bending their values, too. How do you calibrate your sort of your personal views with, like, what the company's values are? It's a it's a it's a question that some other CEOs are struggle with.
Yeah, I think I think maybe by being a founder, I I I have less of the struggle cause I I can get away probably with more, or maybe even more than I should probably be able to get away with. In general, the the default, um, disclaimer is that these are my views, these are not like the box official, like put our logo on it views. These are just Aaron's views about a thing. And, um, uh, and and And you're not worried so
about the administration making trouble for you because of that?
I'm not worried in terms of like I lose sleep over that, um, but I also I also have I've evolved my personal ability to get kind of, like, like, uh, aggravated quickly about about everything happening. I'm speaking out on a handful of topics that that I think matter, uh, to the country economically, and and then I'm, but but, like, I'm I'm trying to be balanced. I do think that we could spend less money in the government and get more out of it.
Some components of Doge, uh, I I've I felt made sense. And then on areas like tariffs, where I just think it's an economic disaster, I'm gonna be clear that, like, I think that's a really bad idea.
And do do you have a a conviction yet about what Trump's impact on AI is gonna be, whether that's regulatory competition with China? I know you've been pretty positive about DeepSeek. Some s some people see that as evidence of China's ascension, something to be to be afraid of.
I'm weirdly less afraid of China. I have not gotten bitten by the by the sort of be deeply afraid of China bug. People can definitely call me an idiot if things go really, really, you know, you know, poorly with them. I think there's some huge, hugely bad problems with China. The hacking issues are a huge problem. IP issues in China are a huge problem. I don't have the oh my god, if they win AI, it's all over.
I have the reverse, though, which is which is we should win AI because economically, more of the profits a are gonna accrue to your economy. When I think about the, you know, this administration's approach to AI, there are some incremental positives relative to the Biden administration. So so there there's a general bias toward supporting things like open source AI. There's a general bias toward not trying to overly lock down AI, but actually let let AI get more diffuse because you you eventually innovate faster when there's more more kinda angles testing your your systems.
I give, um, this administration decently high marks on AI policy thus far. That could change, of course, but but but, you know, I I feel more comfortable at the moment on AI, you know, than than than a number of other topics.
Winning the AI war, like, how much of that is reliant on, you know, continuing to attract high-skilled, uh, immigration? I mean, the the the Trump administration's immigration policy is sort of at odds with the the vision of of this being a destination.
So if you want to win in AI, you know, my my wish list is support open source AI. Check. Um, uh, rapidly accelerate the development of of chips in the US. and and, and make sure we can control that that infrastructure as much as possible over time. Sort of check. Uh, h- tariffs are actually a headwind to that. Don't over-regulate AI. So so also check. But then you need the best talent in the world, and we need a lot of it.
And then that would be that would be kind of neutral to negative so far. And to be clear, Biden didn't didn't do this either. What you would do is you would just do the whole green card stamped on the diploma, uh, after college, and you would get anybody studying AI in the US. or even, you know, computer science in the US. You would just be like, we want you here. We're gonna build the best environment in the world for you.
No need for you to go back to to your, you know, kinda country of origin.
And the fear that those people are gonna take American jobs, y- your point is it's gonna grow American jobs by having that kind of talent.
By having those those amazing people not being America, you over a ten-year period will lose vastly more jobs than the incremental jobs that they are theoretically taking in the near term. Because those people are gonna be smart in our country or a different country, and they're gonna build AI in our country or a different country. And the country that has more of those, that flywheel of the better better computer scientists doing the better algorithms that need more compute, that build better products, the flywheels that happen in other countries will mean over time more jobs go to those countries.
You want the Sergey Brins to to to build that company here, um, you you want the Jensens to build that company here. And so you want the talent to stay here and and grow here as much as possible.
You've been through so many waves of change running box, and I'm curious if you have lessons or advice for other leaders today who may be struggling with the pace, you know, the sort of feeling of chaos that that a lot of folks are are experiencing.
I mean, I think I probably credit to you, you kind of had this initial insight, right? The world is is is increasingly, um, uh, eh, you know, a world of acceleration. It's highly dynamic. I do believe we're in a globally interconnected world at this point. I think it's a good thing. I think it's healthy for society. I'm firmly, like, a 99 999 percentile optimist of of the benefit of all of this. It means we'll get better cancer treatments.
It means that we're, you know, the people that that aren't mobile today will become mobile. We're entering this incredible era of abundance and acceleration. And and so I think the reason why I can stay up with it is because I'm excited by it all. And so I think if it was if it was, uh, if it was draining and I was I felt like, like, wow, like, this is all bad, I I think I would be, you know, fatigued and and stressed out and and burned out.
But but it to me, it's kind of the opposite. Like, I grew up with technology, I fell in love with technology. I I, you know, I think it's this amazing gift, and so I'm only basically having fun. I mean, like, it's stressful, it's stressful in the sense of, like, I need to make sure that Box stays ahead of everybody else. And so, like, that has anxiety provoking feelings, but the excitement of the space and the field is what is what propels, uh, you know, I think so much of my energy.
And so so, to some extent, I j- I think everybody has to kinda find their version of that to stay enthusiastic, to stay energized about what's what's going on. I think for anybody in technology, that should almost be coming for free because we're in the we're in the the period of the greatest amount of change in technology history, and you're you're gonna be s, you know, able to say that you were at the center of it.
If we talk again in 12 months, or when we talk again in 12 months, w- what will have changed most about AI and the way we use it?
For consumers, um, I think our daily lives will still incrementally kind of evolve. Your AI assistant, in the form of ChatGPT or Perplexity or Gemini or Grok or whatever you use, is just gonna be ex- more powerful. You know, you'll take an image of something, and it'll instantly give you all of the analysis you'd ever want on it. Um, it'll generate images that are perfect. It'll give you the perfect answer to your question. The consumer world kind of have the playbook.
On the enterprise, I think I think the next 12 months, we should think about this as the the period where we have our ChatGPT moments with AI agents. This is the 1st year where you could see the initial proof of concepts. But I think it's worth noting that for the enterprise adoption of AI, this is gonna be a decade-long journey. Um, you know, we we we thought that, like, cloud computing was was, like, you know, gonna happen in 3 years or 5 years.
We're 20 years into cloud computing, and and the and the quarterly rate of cloud computing sort of adoption is still growing at the same rates as ever before. I I think this will be much faster with AI, to be very clear. But just so we like, shock ourselves, imagine we're in 2045 and we're still talking about, you know, wow, AI, you know, was adopted in this enterprise for this thing. Like, like, think about how unbelievably crazy that, as a concept, sounds.
That's how we are still talking about cloud today. When I go to a bank, they'll still say we just deployed cloud for this trading platform. 20 years, 20 years into this. So these things take a while. The change management takes a while. Partly, this is why you're you're seeing us and Shopify and others go in so hard on AI. You actually do have to kind of nudge and accelerate and create the permission to move faster. That's why that's why moving AI first now is a very relevant exercise because because you do want those learning cycles to happen in your company sooner and faster than it does, you know, outside your company.
Um, so you can, you know, be ahead of the curve. Society in in, you know, a decade, decade and a half will look just completely different than it does today.
Well, Aaron, thanks so much for doing this. This was great.
Thanks, Bob. Appreciate it. Good to see you.
Aaron sees disruption as opportunity, not that he thinks every disruption is good, tariffs are a case in point, but tech-driven disruption just gets him excited. And I think he's right. We all have to find ways to be excited about AI, even if it scares us or feels alien, because it's advancing at a pace that's hard to get your mind around. The challenges for each of us are emotional, every bit as they are technological or intellectual. We may not want to sacrifice our sleep or our hobbies to stay up to date in these new areas, but those of us who do, who can make peace with the chaos and make it our friend, those people will be in the best position
to thrive. That Fast Company cover story I referenced that included Aaron, it talked about a concept I call generation flux, a group of people not defined by chronological age, but by the willingness to jump in and build new skills in an environment of uncertainty. Today, because of AI and more, we all need to become part of generation flux. I'm Bob Safian. Thanks for listening. RapidResponse is a wait what? original. I'm Bob Safian. Our executive producer is Eve Troeh.
Our producer is Alex Morris. Associate producer is Masha Maku Tonina. Mixing and mastering by Aaron Bastinelli. Our theme music is by Ryan Holiday. Our head of podcasts is Lital Malad. For more, visit rapidresponse show dot com.