We also don't really have these kind of core sources of truth for data in most enterprises, like, you know, where is the contract for that one customer stored? It might be in five different places. That's five different five different chances for the agent to get the wrong one. That is heavy, heavy human work. There's no way to automate it. You have to have people go in, understand the workflow, understand the data, build out evals.
Aaron Levie:I think we need to not be waiting around for the one agent to rule them all.
Christoph Magnussen:Wir haben wieder etwas Besonderes für euch, nämlich einen der Top-Gäste aus dem Silicon Valley bei AI to the DNA. Wir sind hier im Remote Setup, denn diese Folge war nicht möglich vor Ort aufzuzeichnen, aber wir sind hier in der Nacht, das heißt, wir haben für euch wirklich, was ist es jetzt, 2 Uhr, ich würde sagen Serious Night, diese Folge aufgenommen. Es geht um Aaron Levie, den Co-Founder und heute CEO von Box. Die haben 2005 Box gegründet, eine der longest serving companies in the Valley, er als quasi CEO und Aaron ist einer der Stimmen, wenn es um KI und vor allem das Thema Agents geht und die Frage, wie müssen die Daten dahinter organisiert sein und zwar ganz konkret, eure ganzen Excel-Dateien, eure Powerpoints, die Word-Dateien, die E-Mails, was machen wir damit? So und insofern gibt es hier eine super intensive Folge High Speed Get Ready mit Aaron Levie bei AI to the DNA, hier wird es richtig nerdig, viel Spaß.
Christoph Magnussen:Aaron, nice to meet you in almost person finally. And it's it's been a while that I followed you from the early days of being a founder and I'm looking really forward to our conversation and I actually have a question that I'm thinking of for a long time. So organizing company knowledge in order to make it workable for agents seems to me as the key question if I compare it to the industrial revolution when you had a steam engine change it to electrical engine and people were playing around with it but not really realizing how much you can reorganize and how much you have to reorganize and I know you have a pretty strong opinion on that and I would like to start right with that.
Aaron Levie:Yeah. I mean, think I I I think this actually cuts to the one of the core questions of our kind of agentic era that we're entering, which is how do you get the agents to participate in the workflows that you're doing? And how do you get them access to the knowledge that they need to be effective in any business process you throw them into. And if you just kind of keep doing things the way that you've always been doing, and you just kind of drop in agents into your existing processes, oftentimes you're not going to get the upside that you would expect because because there's just a bunch of, you know, kind of challenges inherent in agents like they're very forgetful. They have to be re prompted with exactly the right context to be effective.
Aaron Levie:They need access to a whole bunch of resources in your organization to be useful. They need to have very, very kind of well documented sort of understanding of your process. So these are a bunch of things that people don't have to have, you know, people, you can kind of throw them into a new problem. And very quickly, they can kind of intuit the behaviors of other, you know, people in the organization, what's the style of work that we do, they can, you know, relatively easily find what they're looking for by asking a colleague or kind of spending a bit of time kind of navigating the corporate systems. Agents don't have a lot of those benefits, they're just going to kind of go try and find what they're looking for instantaneously come back maybe with the wrong thing and run with that.
Aaron Levie:And so then the big question is, how do we get agents, the most important kind of, you know, data and knowledge that they need to be able to be effective. And so I think knowledge management and information management in the twenty-first century century will become, you know, one of the kind of core raw ingredients of this industrial revolution that we're we're entering.
Christoph Magnussen:Who do you think gets it right at the moment? I am thinking of the first ChatGPT moment and Microsoft came out with Copilot right afterwards and saying, we got your SharePoint. We know what we're doing. And then we were trying like, okay, guys, this is not how it works. Who would you say three years into the ChatGPT game?
Christoph Magnussen:Yeah. Who who got that?
Aaron Levie:Well, there's there's sort of different kind of parts of the ecosystem. I think there's sort of companies that are trying to deliver on that vision as closely as possible. So for instance, what we're working on at Box is make it really easy and secure to get your right the right knowledge in your company to agents, again, in a very kind of secure and well governed way. So we've been working on a bunch of technology to really kind of help with this problem. Then on the other side, there's this sort of, let's say, the practitioners are the users of the tools.
Aaron Levie:And, and I would actually say, like, generally, you can go and look at small startups, to sort of see what are the new practices that those startups are running with or using to be able to go in and work in this in this kind of more agentic or AI forward way. And what they're doing is, is the small startups are kind of running their businesses entirely differently from how we would have run a company five or ten years ago. So first of all, they're, they're, they're sort of running the business with agents as first class citizens in their company, which means that they're documenting everything, they are making sure that all of their decisions, all of their corporate knowledge is built out in such a way that agents can can on day one have access to that information. They're constructing their workflows in a way that agents can natively participate in them, as opposed to them feeling like they're sort of an afterthought or add on to the workflow after it's already been kind of designed for people. They're, they're sort of incorporating the right ways that humans now need to be participating in processes where they have to review the work that the agent is doing where the agent has to be kind of effectively, you know, making sure that that that you're you're, you know, again, like looking through all of the output that the agent produced, and then incorporating it into the rest of the workflow.
Aaron Levie:So they're designing their workflows in a sort of an AI first way. So most of the lessons are coming from small startups that are kind of operating in this way. And then the big question is, how do big companies end up, you know, starting start to how do big companies end up operating like these startups are, given how much change management is required, how much scale these companies have, that's a much harder problem than I think a lot are anticipating. But but, know, we're we're I think everybody's kind of collectively on this journey together to go and do that.
Christoph Magnussen:I I can pretty much remember the day when you were a start scale up of like 70,000,000 revenue. Now you're 1.2, 1,300,000,000 in revenue. So you're
Aaron Levie:a
Christoph Magnussen:huge company. And as I know you from the outside, I know I would assume you play around on the weekend, you test stuff, you sit there at night and you go into these things. But the scale is much different. And how do you how how would you start in transferring and what did you do in playing around making Box an AI native company? Or did you shy away and say like we will never become an AI native company.
Christoph Magnussen:There are other companies out What's that state at the moment?
Aaron Levie:Yeah, no, no, we're, I'm betting everything on us being AI native. And so that means that every process in our company has to be positioned for being AI first as an organization. So everything about how we run our company, how we operate, you know, has to be delivered in AI first way, you know, that takes on different characteristics. So in engineering, it's relatively straightforward, because we are all kind of collectively learning the best practices, where you're going to have maybe Cursor or Claude Code or Codex, and your every engineer is going to be deploying agents. And then some of your agents will be local, and some of them will be in the cloud.
Aaron Levie:And you need a kind of modern kind of AI first product development lifecycle, which means you need a workflow engine for these agents to participate in that have access to your product resource documents and your design files and your code base. So we've had to do a lot of work to kind of design a modern agentic kind of coding system for the company. Then we have to make sure that all of our kind of the more knowledge worker tasks in product management, or go to market or sales or solution engineering, how do we make sure that all of our Box employees have access to agents that help them work as effectively as possible. So agents that can access all of the kind of corporate data in our enterprise, agents that can help you automate account research for customers you're going after agents that help us accelerate our product campaigns, or our marketing campaigns. So that's, that's a lot of the rest of the work.
Aaron Levie:And I think mostly what we're doing is we want to encourage everybody to use AI as much as possible. But as long as it's kind of practical in the use case, we want to make sure that we keep people in the loop with everything. So one thing that we've been very clear on is that you still are accountable for the work that the agent is producing, you can't kind of relieve that accountability and say, no, the agent did it. We're also very, I think, biased toward the sort of human plus agent collaboration modality as opposed to just move to everything being agents and kind of get rid of the people. We want to make sure that we find a way to augment the work that people are doing as opposed to kind of really think about replacing that work with agents.
Aaron Levie:So we really think about it as a force multiplier of the people side. So we haven't looked at like, how do we cut jobs with agents? We kind of do the opposite. We sort of say, how do we get more output per person? And and how do we just make sure that we're we're getting more done as a as a company as a result of adopting agents?
Christoph Magnussen:How do you like I mean, this is much much more as I also see it as a compute revolution instead of like doing communication and a little bit of efficiency. What you're describing is how do I leverage the intelligence of a person through an agent on top. And this is much more difficult in my experience for people to adapt to because it's so different regarding the job. So let's take an example on the way here of course I prepared with Codex. This is my usually go to spot because I like to code, but I also like to do knowledge work.
Christoph Magnussen:And I find it a very, very reliable agent. And there are certain things I can delegate to it and know, that'll happen. And then there are certain things I have, like, feedback when I really want to relieve it Codex can say, okay, you double checked it. Most of the time, the problem is finding the right source in the vast tool set that we also have, that many other companies have. And I know that you're always saying you're the security and governance layer.
Christoph Magnussen:How does it practically work? Because what tools and on what level the rules can clean up the mess that many companies have out there, all the Excel files, all these dead artifacts, like they were so important in order to get a good outcome. And this is to me one of the key question marks that they usually have.
Aaron Levie:Yeah, I mean, this is, you know, this is sort of one of the big problems for an enterprise is a large amount of their data, as an example, is fragmented across, you know, ten, twenty, 30 different systems inside their enterprise. Agents don't easily have access to that data. The protocols don't work well. The systems are on prem, so they don't really kind of connect to MCP server. I mean, they can't really kind of show up as an MCP server very easily.
Aaron Levie:They're not built for you know, agents to go and kind of navigate very effectively. So our data is fragmented. We also don't really have these kind of core sources of truth for data in most enterprises, like, you know, where is the contract for that one customer stored? It might be in five different places. And so if it's in five different places, that's five different chances for the agent to get the wrong one, the wrong version, you know, go off of the wrong, you know, answer to a question.
Aaron Levie:So we need our information to be, you know, we need better sources of truth, we need more authoritative sources of documents, we need to move this data to the cloud effectively. So that's a lot of the work that has to be done if we're going be able to use agents across our workflows as in knowledge work. That's that's actually why we're seeing so much demand right now is more and more companies are realizing they have to go and modernize their systems and be able to move that to the cloud effectively.
Christoph Magnussen:Do you have also, I mean, the the famously forward deployed engineers for, like, doing this for Box customers. So they they are with clients and doing cleanup of the data and stuff like that. How did that job then change completely?
Aaron Levie:Yeah. So we we do have our equivalent of FDEs and what they're basically doing is, they're going into an organization, they're saying, okay, let's understand where the data is in this workflow. We go to a bank, and they have a bunch of customer documents for getting loan process loans processed or onboarded in a wealth management, you know, kind of relationship. There's a bunch of personal documents, a bunch of loan documents, a bunch of bank statements that they needed to go in and get into the cloud. Well, that data might be fragmented everywhere.
Aaron Levie:So we have to first help the customer understand where it is, we have to find a way to kind of map it from the prior sort of architecture and move it to a modern architecture. So we have to kind of get it into the cloud, we have to get it in a data ready, in an agent ready data format. Then we need to actually figure out what models are going to work well on this data. How should we configure agents to work well? We have to do a lot of evals for that customer.
Aaron Levie:We to help them sort of figure out like is this workflow actually going to work for you inside your organization? So all of that work has to be done. That is heavy, heavy human work. There's no way to automate it. You have to have people go in, understand the workflow, understand the data, build out evals.
Aaron Levie:That just takes in a tremendous amount of work. And that's why you're gonna have humans go in and get utilized for just all of these processes in the future.
Christoph Magnussen:But how do you see it when I have many questions regarding that. Let's start with one. That is: a new model comes out. In that moment you need someone and you need the skill set in order to evaluate again the new model against the data against the system prompts you might run-in your system. So who would do that then? Customer side or do you see that as a service in the future that will be continuously there?
Aaron Levie:I think that vendors will have that kind of talent and they'll help support that. Then they will certainly you know, kind of vendors will lean in with with their version of FDEs, you'll probably have systems integrators that have a role for that as well. So they'll have a bunch of talent that is sort of FDE talent, as well. And, and then individual customers will also probably have some of this homegrown within their IT organization. So the IT organization of the future will often be really kind of these FDEs in the organization, and their job will be to constantly kind of maintain and curate these agents for each of the lines of business to make sure that all of the workflows and the agents keep running effectively across the enterprise.
Aaron Levie:So I think you're going to see a lot of interesting roles emerge inside the enterprise to help deploy and manage and, and observe and improve the quality of agents inside of organizations.
Christoph Magnussen:And what about the little bit juicy topic of building, let's say relationships to the providers that have models that are maybe locked off for the public but available to a certain amount of people. Is this the new kind of PR relationships that has been built? Do we see here? What do you think? I know you have a strong opinion on that one.
Aaron Levie:Well, this is a very it's a very interesting question and it's because I I think there is a little bit of debate is is, you know, will there be these kind of trusted access kind of layers to get access to Fable or to get access to GPT 5.6? I think there's an argument that that's one scenario that plays out. Alternatively, there's a lot of value in this idea of kind of being a model router where sometimes maybe it's frontier model access that you want. And other times it might just be an open source model that you want to make sure that is kind of tuned to your particular workload, and who's going to be in the best position to be able to give you access to an open, an open weights model, it's going to be the abstraction layer between the workflow and the model, it's not going to be one of the model providers, it's going to be that kind of routing layer between the actual business process. And so that's the layer that we provide.
Aaron Levie:That's the layer that Cursor provides or Factory or Cognition or Salesforce or ServiceNow. So that's gonna be a lot of the value that that kind applied AI layer will will deliver.
Christoph Magnussen:Yeah. No. I I I get that and I I can see that and it's still new. It was definitely new to me that in model has been rolled back and hold back. I was like, this is a bit awkward and I told our clients in the end the stronger your layers in between are
Aaron Levie:Yes.
Christoph Magnussen:Like the contacts, the prompts and so on, the more independent you are. But still, it has a little bit of a taste that this happened and it also dawned to me, okay, there is a new age of models because there is something to I always see it as a triangle. You have the model as the engine, you have the context and you have the prompt. And it all needs to be in balance in order to provide something. But there are certain moments when the model becomes so strong that you sit in front and you might have had this too in December when everyone thought like, oh, wow, Opus 4.5.
Christoph Magnussen:Woah.
Aaron Levie:Right.
Christoph Magnussen:This works finally now. And this is different to technology we had in the cloud age and if there are now companies having a big advantage, do you still see that as startups running behind you can still catch up or would you say this is a mode that is so strong it's hard to catch up? Because these companies are partners on the same side, right? Like they're partners, providers and mix it up.
Aaron Levie:Well, sorry, the moat being so strong on which side?
Christoph Magnussen:The the moat being so strong on the provider side, let's say, OpenAI with Yeah. 5.6 or or Anthropic with Mythos having such a strong, not only responsibility, but also a moat, saying having a stronger model and providing that to certain players, but not everyone.
Aaron Levie:Well, I actually think that the model providers probably would like to make sure their models are broadly accessible. Right. If you were to talk to Sam Altman right now, my guess is he wants to make sure that GPT 5.6 is massively available. Because anytime that they constrain access to their model, that's just less demand that they can go and have driven. I would say the models want to be freely available, And but the routing layer is what can actually understand the workflow the best to be able to figure out like what model works best with, you know, with the specific business process.
Aaron Levie:And so, so the reason why the routing layer adds a lot of value is because there's a lot of scenarios where maybe you want Fable 5 for really kind of being a planning agent or an orchestration, you know, agent in a workflow, but then you want, you know, you want glm 5.2 for doing heavy token processing of a lot of the unstructured data in the workflow. So in this case, Anthropic still going to generate, you know, quite a bit of revenue, because Fable will be involved. But by token volume, more of the tokens will flow through an open weights model. And the routing layer will be the reason why you can choose different models for those different tasks. So I think this is going to, this this idea that, you know, what's going to happen is as an enterprise, you probably want to have some some kind of abstraction between your knowledge, your workflow, your data from the models so that way you can always swap in the different models and that and the routing layer basically will provide you that abstraction.
Christoph Magnussen:If you but still with with this router, let's let's think that through and I think you were the one when Uber had the headline out, we cut back on Claude Code because it became too expensive and I think you mentioned but don't forget they're levering up on Codex now so you cannot give up these models but you need to budget it. You need to make a decision of what value do you get out and how do you teach people on the one hand to make this decision to have an idea for the value and how do you budget it next to the IT budget, for example, saying, okay, we have to see it in the controlling shape at least and plan it ahead a little bit. This is totally new.
Aaron Levie:Yeah, know the ROI of kind of agents is this really fun topic because you know, how do you measure? How do measure the ROI when, when it's a brand new technology, it's changing very quickly, it's being deployed in like a sometimes a diffuse way throughout an enterprise where I, you know, you can't measure every single person's kind of use of AI at all times, it's in a lot of it is sort of obfuscated, and not not sort of, you know, something that you can kind of just go and, and deeply understand. So in general, what you have to do what that leads to is you probably need to have different org owners have budgets for AI, like they have budgets for anything else in the business. And at the end of the day, you have to give your head of engineering a budget, or your head of marketing a budget, or your head of sales a budget, and they have to use AI, they have to think about AI as one of the budget line items. And they kind of have to be responsible for how how effectively do they want to deploy AI in their organization.
Aaron Levie:There's almost no methodology in the history of management science, other than having accountability, be driven by the team responsible for the delivery of the thing that you're trying to deliver. So if you want AI to accelerate your sales team, you want your head of sales to be the one that owns how much should they be spending on AI. You don't want a different organization owning that decision, because then you kind of remove the accountability from that org owner. So I think the reality is what's going to happen is the IT team is going to own managing the technology, they're going to own kind of bringing in the latest version of it, best best in class version of it. That's their job.
Aaron Levie:And then the you know, the sort of management and effective use of the technology will go out into the organization. So the head of engineering has to figure out how are they deciding on the ROI and the productivity that they're getting from AI. That's not the CIO's responsibility. Because the CIO also can't be responsible for how effective is your sales team, you know, on the on the people side, and AI right now is just an augmentation of the people's, kind of ability to get work done. So it's going to be this kind of joint responsibility, the tech has to be owned somewhere.
Aaron Levie:But then the accountability of the ROI really has to be coming down to the business owners who ultimately own the outcomes of those job functions.
Christoph Magnussen:Do you already see fortune five hundred changing in that direction? Or is this something more close to you and anyone in the valley at the moment?
Aaron Levie:I think it's, I think it's early for the fortune five hundred. I think the fortune five hundred has generally seen this as an it technology initially. Yeah. But now is starting to figure out that no, no, no, this is a this is really a technology that augments people. So that requires that kind of joint tied at the hip dynamic.
Aaron Levie:Yeah, you can't hold a department responsible for something that it can't see the entire workflow through. This is going to I mean, interestingly, it's going to put the IT organization in a more powerful position. Because the IT organization is now responsible for the outputs of and the sort of the leverage that the entire company has. Because if the IT department brings in technology, that's not good, then then obviously, you know, then the whole AI strategy is going to fail. Conversely, if they bring in technology, that's amazing, then the company could run way faster and way more effectively, but still the budget decisions, the ROI decisions have to come from each individual organization.
Christoph Magnussen:Which makes perfect sense. And let's say let's let's let's stick with that example and say the IT has a decision saying, we also have to come with something that is secure. And then you come from the departments and they, let's say, started with Claude Code or Claude Cowork or something but from the IT side it has been the tool of choice and now even further you come with Claude Tag and it even enters your Slack account and for normal people at the moment it's not easy to overlook what is happening. You definitely know what is happening and how strategically important this can be. How do you balance these two sides that are racing up?
Christoph Magnussen:Yeah.
Aaron Levie:No, no, sorry. Go. Yeah.
Christoph Magnussen:Yeah. The security side and at the same time people driving these new trends very fast.
Aaron Levie:I think it's very important that so I mean, unfortunately the answer is to try and find the best the point on the intersection where you're not trading off either. And so what does that mean? So let's extreme take ends of polls. So the most kind of like a locked down governed system, let's say, would probably not remain very modern or move very quickly, because you would just implement a system and not ever update it. And so thus your users would always be unhappy.
Aaron Levie:And they would because they'd always be getting bad technology, but it would be like theoretically up, they would be theoretically like locked down and governed. Conversely, on the other end, you could have a version where users can bring in any tool they want. That means you'll always be using the most modern tool for any given team that adopted something new. But obviously, then there's no governance of it. So that doesn't work either.
Aaron Levie:So the only real solution is like, what is the intersection where you get the most amount of modern, you know, technology that works really well, and the best amount of governance. And that does mean you need an IT team that is moving way faster than ever before, It has to change its its sort of adoption patterns more than ever before. We've already had to change our internal engineering toolkit probably three or four times in the past two years. Because of the amount of changes happening in the best practices in the space. So we had GitHub Copilot, then we had cursor, then we had Cursor plus Claude Code, then we had Ccursor plus Claude Code plus Codex.
Aaron Levie:Now we have cloud agents that also can be deployed. That's just a two year period, we've had to change or update our practices three or four times. That is basically the pace that the IT organization has to keep up with in 2026. There's simply no way to not do that. If you want to be able to actually have your users have the modern technology that's available.
Aaron Levie:So I do think that it requires it teams to be moving a lot faster to be very forward thinking, to be paying attention to what's going on in the AI space much more, you know, much more intently than ever before. Those are the IT organizations that are gonna be able to kind of stay ahead.
Christoph Magnussen:And user interface design wise and product design wise going to headless, meaning going back from the typical SaaS, you have GUI, you can click through in the interface and I'm not talking about only the CLI phase that we've been running through in high speed now, all going back to like GUIs when we work with our agents, but at the same time agents also use these tools headless and how do you then make sure that you still are the tool of choice also strategically for a company because this is a totally new kind of product like do you go then all in on the headless design which is pretty new for many companies I guess or do you have to do both?
Aaron Levie:Definitely have to do both. So there's a lot of ways where agents need access to some kind of underlying APIs. They're gonna have to do a bunch of work on a system and access data and do workflows. But then there's a lot of times where I as a user still wanna go and see like the general kind of like terrain that is related to that particular task. Want to see the whole Jira ticket, I want to see all of the documents related to this one topic.
Aaron Levie:And that's not something that I want to necessarily chat back and forth. Want to be able to just see like the underlying data just as I still go to lots of websites. I use ChatGPT a ton, I use Gemini a ton, I use Claude a ton, but I have more browser tabs open right now than at any point in my history. Because what happens is, is I go and click around and I go and see stuff after the agent has sort of answered something. And that's no different from effectively using GUIs of software products.
Aaron Levie:So you're going to have agents want to use data in a very headless way, you're going to want to be able to pop out into well designed systems as well to look at that information or to be able to kind of execute a task in a workflow. And I think those things will just work hand in hand together. I think they are, they actually have a very symbiotic relationship. Just as there's a lot of times where I'll go and work on my mobile device, and I'll load up something on my phone. And then, you know, two minutes later, hop on a computer to complete that task with the same application.
Aaron Levie:Like, sometimes I'll start writing a Slack message on my phone and be like, it's it's gonna be long enough, I'm gonna go and hop on a keyboard, or I'm looking at a document on my device, but then I have to go into a meeting. And so then I want to take the document with me on my phone. So I think we have we know how to have these kind of fluid relationships with technology, and agents plus deterministic GUI-based systems, they will they will be very symbiotic in how they interact with one another.
Christoph Magnussen:What is your go to agent harness or organization tool? Like where are you shooting off agents or are you pretty good in deciding this is good for agents, I do this directly, How do you do that?
Aaron Levie:Yeah, I mean, my workflows don't always approximate, you know, all of the work that's being done. But for me personally, like my, you know, most of my time is either like reading, you know, a document, a presentation, a strategy. So I haven't been able to really replace that with agents. I can use agents to go find the surrounding set of information that I'm looking for. I do that quite frequently with agents.
Aaron Levie:And that's often done within Box, or it's a lot of external research. So I'll use something like Claude Claude or CoWork, or Codex or Perplexity to go and do that. And so that's a lot of external research or I'll connect to the Salesforce MCP server, or I'll have Perplexity Computer go do a bunch of analysis on the internet. So those are those would be like a lot of the common tasks. And then, you know, I'm often prototyping product ideas, I'll use cursor, I'll use Claude for that.
Aaron Levie:And, and so playing with agents from a kind of more coding or product development standpoint as well.
Christoph Magnussen:What's your take on the current development around I mean, Codex has been a product from inside because the development team was using it. Same with Claude Code came coming from the inside and now huge hits in the market in a very very short amount of time and it seems like no one yet has found the let's say iPhone of agents like how do we do it what is really AI native now Claude came out with Claude Tag so it's integrated in Slack So Slack would stay the layer with Claude, but there's a huge discussion like what's the entering layer? And I don't know if there is an answer to that yet, but what's your guess and what's your view on that? Because you usually have broader ideas than most of the ones I'm listening to, and I would be curious to to hear some of those.
Aaron Levie:I I actually have a maybe of a more pragmatic view on this one. I I think it's I think we're about to enter an era of of actually all the above. Everything you just said. I don't I think we need to not be waiting around for the one agent to rule them all. I think the ChatGPT moment was a very special moment in, in, you know, kind of AI, because it kind of was a unifying sort of paradigm that that, you know, made sense for all of the, you know, kind of broad based consumer use cases for AI.
Aaron Levie:I think as we get into more knowledge work, I think you're going see just a full plethora of different patterns. The way you work with an agentic coding system inside of Factory or Devin will be different from how you'll work with an agentic legal system inside of Legora or Harvey, or how you work with an agentic document or knowledge system like you do in Box. All of these things can be stitched together. So you'll connect to it via MCP inside of Codex or inside of Claude Cowork. But I think we more have to imagine a future where there's not one agentic system that rules them all, there's going to be, you know, 20, 30, 50 different systems in an enterprise, all for the different data types or different job roles or use cases that the organization has around AI.
Christoph Magnussen:So how do we price that? I mean, this is so challenging also for customers saying like, this is what we're gonna buy or not. I tell you, typical German company says at the moment, oh, we have Microsoft, Let's use Copilot and now we use AI. I'm like, you're not. You're far away from that.
Christoph Magnussen:It's one step. It's not everything. And they shy away from from leaning into that and investing into that which is then difficult to catch up later. Yeah. How to fix that?
Aaron Levie:Well, I think it's tough. I mean, you do wanna have, a few maybe best practices. One, you probably want people that are really, really deep in the technology space in the company. Often these are maybe maybe people coming right out of college, they've been playing with these tools, they have more of a kind of an AI native approach to work and to coding and to research. So that tends to be a pretty good sort of way of getting information from the outside world constantly into your company is like, just like always listen to the young people, because they're the ones that are using the modern set of tools, because they didn't grow up with the old ones.
Aaron Levie:They only discovered AI at the same moment they were having to do these things for the first time anyway. So they were getting the modern technology into their workflows on day one. I think to some extent, competition kind of forces this also. If your company is not using the right technology stack, the right AI stack, you'll feel it your, your company will just move more slowly, or you'll be spending more on technology that you need, or you'll be getting kind of slightly worse decisions. So to some extent, the market kind of like, like, ends up solving a lot of this for you, because you eventually need better tools to stay competitive, or you end up spending your resources on things that are not differentiating, and the rest of your competition is able to reapply its resources into more competitive areas.
Aaron Levie:You know, there was a time in the early days of the cloud, where it just become very obvious, if you kept running your data centers, you would be spending more money than your peers on undifferentiated activities. Like, if you go in and spend time putting a CRM system into a data center, instead of using a SaaS based CRM system, then all of those dollars and time and energy that go into that work, your customer doesn't care about. You get no benefit from your customer base by you doing that. But that is still resources that you have to go expend on that problem. So conversely, if you use an agentic system that is not you know, giving you the right upside, but your your competition is getting that upside, you'll just very quickly, you know, realize that that actually you're falling behind in the market.
Aaron Levie:So I do think the market kind of, you know, ends up making a lot of this stuff get get solved for you a lot of the time.
Christoph Magnussen:And regarding that, since you also mentioned how the cloud works and what worked back then, I mean, US Box or other companies like Salesforce or even SAP in Germany, you have totally different challenges when you build a product much broader than most people think that code on the weekend with Claude Code and say, great, I build my own software. Yeah. But what is your take on this additional software layer where it's not regarding millions of people or not even hundreds of people, but maybe one person saying, okay, I built this for myself like I do with an Excel spreadsheet. And with that, you still enter a software product even if it's one person that needs to have data, that needs to have governance and rolled out and so on. So this famously quote unquote software of one for one person.
Christoph Magnussen:Is this something you see really as sustainable because you can now build things that were just too expensive to build in the past? Mean, the early days of the cloud, it was crazy to me that you then could rent with AWS and you said you build Before that, you had to have your own servers. It was just too expensive. Now it gets even cheaper. Or do you think this will be over soon because this will be provided by big companies?
Aaron Levie:Well, I think that I think I like the idea for certain use cases. I think that for most of your software activities, you're not gonna be generating software on demand for yourself just because most people are not, their needs are not that unique, and nor do they have the kind of like appetite or energy to kind of think through how do I go prompt the system to have it build something. You know, probably for my entire life, if I had gotten good at Excel, I probably could have done amazing things, but I never got good at it. So I just never ended up doing anything interesting with Excel my entire life. I think that's kind of like how a lot of people will treat N of one software because they're like, yeah, like I just don't want to have to learn how to agentically code or run a piece of software.
Aaron Levie:And the upside for that person is just not high enough to make it worth doing that. Conversely, there's actually a lot of people that it is, you know, this incredible new technology, you know, there's going to be a lot of people in it teams that can go out and now build custom software for their company. And I think there's going to be a lot more custom software than ever before, there's no question. I just don't think that will be the dominant use of software in the future. Think it'll be 30% of software or 20% of software.
Aaron Levie:But what's going happen is we're gonna have more software in general, we're going to have way more applications, we're gonna have way more more software for all of these, you know, tiny little niches that we could have never had software before, because some entrepreneur will come up with this idea, and it'll be far faster for them to go build it. And they'll be able to actually go and deliver in a market that they could not have gone and pursued previously, because it wouldn't have been justifiable building an entire engineering team just to go after this one niche market. So I think we're gonna have more software, we're gonna have more commercial software. And I think as a result of, of agents making it easier to code, I think we're gonna have more n of one software as well.
Christoph Magnussen:How important became regarding that the topic of relationship when it comes to clients and internally? Because people always think like the AI will replace, but I know that you're a company of course of with a with a huge sales force internally having relationships with your clients. And how did that change in the past three years when you look at that? How important became that?
Aaron Levie:Well, I think relationship with customers is one of the most important and sustaining moats that exists. So the ability to have an ongoing relationship and learn your customers environments and update your technology to better support them is critical. I think that will be one of the sustaining moats in the future of software.
Christoph Magnussen:And regarding that, the topic of the I think Sequoia put it out saying that the next trillion dollar market is a technology company but wrapped as a service company. So Yeah. A little bit of both. Is this something you're you're heading to with the FDE is at Box or are you still more the tech company in the classical way?
Aaron Levie:I think we're probably more of the tech company in the classical way. I think they're they're they're you know, I think the outline that they are are focused on is probably more of the new companies opportunity. Like if you're going to do like an agentic law firm, or an agentic insurance company, or an agentic financial services company, you kind of have to both both compress the human services work, and the you know, software work into one company, that is that is actually a pretty compelling new opportunity for lots and lots of startups, we probably aren't don't make as much sense there, because we've always been very horizontal. So what we would probably do is actually help power many of those companies. So you would use Box as a platform for delivering an agentic law firm or an agentic insurance company.
Aaron Levie:And so we would be more in that ecosystem as an enabler as opposed to delivering one of those types of propositions.
Christoph Magnussen:Let's for something that goes through my mind, and I have your timing in my mind as well, I know that you are tight on schedule, but this is something that really interests me. Let's say you go back twenty years and restart, but you restart as an AI native company. Would you where would you start and why? Like how would you rebuild something like Box or something today if you would not have to restructure or rebuild it, but say you start ground up as an AI native company, what would you do?
Aaron Levie:Well, you know, one of the things that we're benefiting from is that that most of what you do with agents, you need documents, and you need enterprise content for these agents to be effective. So actually, there's a lot of things that that I wouldn't necessarily reset, I would actually kind of take forward into the into the future. There's just more ways about running the company that we're constantly learning. So again, like the, the, we are constantly kind of raising our expectations and our goals around projects, because we know that that we can use AI to better deliver technology far faster. So what we're having to do is not accept the same, the same goals or planning cycle that we did a year ago, or two years ago, we have to like constantly ratchet up our goals and our expectations.
Aaron Levie:We have to make sure that again, we have the right documentation for agents. So when a new employee on boards in the company, and they ask a question, there's an agent that can instantly answer that question for them. So we can get employees to be much more productive far faster. We obviously want to constantly make sure that the right that all of our employees have access to the leading tool set to be able to help them work effectively with agents. So those are a couple of the areas that we're constantly thinking about how do we run-in a more and more modern way.
Christoph Magnussen:So in a twenty year timeframe, I mean, you're one of the longest serving CEOs in the Valley when it comes to that, and you've seen a lot of things. Where would you say in these twenty past years, what kept you that driven and that motivated that you keep on going like this today?
Aaron Levie:I think it's mostly just how exciting the technology is. This is one of the fastest moving markets of all time. There's constantly, you know, change that's happening. We get to, play with new AI models as they come out. We see these breakthroughs happen constantly.
Aaron Levie:So I think it's really about the sheer just velocity of the space, and then being able to bring that innovation to customers. That's what keeps me excited. And yeah, so we're having the best time that I think we've ever had in building the company.
Christoph Magnussen:Definitely also number wise, I see, and as I said, follow you for a while and one final thought, what is something of an area just comes to my mind that we didn't touch but we definitely should have a look at when we regard this as the new industrial revolution that is happening. What did we didn't touch and why do you think it would be important? Anything
Aaron Levie:I you can think think mostly, maybe the only thing is I have a pretty optimistic view on the future of jobs. Just because I think people are gonna use companies are gonna use AI to really be able to get more output and get more done in their company. There, you know, they might use it to save money initially, but what's going to happen is some other company will emerge. And, and that that company will, will use AI to do more. And the company that does more and better serves our customers will be the ones that get more customers.
Aaron Levie:So you have a very virtuous flywheel that will cause that will basically cause the companies can drive more output, that might be in the form of software development that might be in the form of selling to customers that might be in the form of better marketing campaigns. And all of that work still takes people. So so I think you're just going to see this, this dynamic where the companies that use AI to do and better serve their customers to do more and better serve their customers will be the ones that keep hiring. And that will actually make the case for why you're actually gonna have more jobs or at least as many jobs in the future for people as as we do now. And AI doesn't actually end up kind of automating all these jobs out of existence.
Christoph Magnussen:So more like a computer revolution that really leverages through also you can shoot off more agents. You can use more intelligence instead of just being more efficient, what many people claim
Aaron Levie:That's right. Meaning.
Christoph Magnussen:Yeah. Exactly. Got it. Aaron, thank you so much, and I appreciate your time. I always wanna make sure that we that we keep in that time frame.
Christoph Magnussen:So thank you for that session, and I hope that wasn't the last one because we we need more of these insight also here in Germany, and I always try to connect that. And you're one of the very open minded leaders that is long serving. So thank you so much for for these thoughts and ideas.
Aaron Levie:Awesome. Thanks, Christoph. Appreciate it.