Steering AI at a Privacy-First Company with Delong Lu

Delong Lu's Career Path to Proton's Early Days Hello and welcome back to the Chief of Staff podcast. I'm your host, Scott Aminta, and today I've got a really great conversation lined up for you. My guest is Dalong Liu. He's the Chief of Staff at Proton, the Swiss privacy company. You probably know from ProtonMail what I love about the long is he's been at Proton for almost 10 years. He started out as one of their earliest product managers. Speaker 2 And it's basically. Speaker 1 Grown up with the company from 30 people to over 600, he's got this incredible ground level view of what it actually looks like to scale at an organization and then step into a chief of staff role from within. We get into a lot of ground today. How the chief of staff role emerged a proton, what it means to operate as what Delong calls a loose radical on the management team. And then we spend a good chunk of time talking about AI. This is a fascinating case study at a company whose entire identity is built around user privacy. Like, how do you move fast on AI when that's your core promise to customers? Delong has a really ground to take on it. Let's get into it. All right, Delong, welcome to the podcast. So nice to. Speaker 2 Have you here? Speaker 3 Hey, Scott, great to be here. Thanks for having me. Speaker 2 How are you feeling today? Speaker 3 Doing good. Thank you. Speaker 2 Give us the way of the land along. Tell us where you're calling in from. What are you currently working on? Maybe tell us a little bit about your kind of past career experiences. We're going to get into a lot of things today, but give us a one minute preview of kind of who you are and what you're doing. Speaker 3 So hey, my name is Long. I'm calling in from Zurich in Switzerland today. What do I do? I am currently chief of Staff at a company that's called Proton. People might ask, what is Proton? Right. Proton is a company headquartered in Switzerland. What do we do? We are a company that builds a software solution, SAS. The goal is really to build an Internet where privacy is the default. It's a very ambitious mission statement, but what does it actually mean in concrete terms? What do our customers get? You can think of us as the European company that builds an alternative to, you know, Google Workspace or Microsoft Office Suite. We are known for our products such as called ProtonMail. It's an end to end e-mail service, but by now we also offer a large variety of different services, ProtonMail, Proton calendar drive, VPN, password manager, document editor and so on and so forth. I'm chief of staff here. It's an interesting role, I'm sure. As your listeners might know, chief of staff are very different in every company. So what does the chief of staff do at Proton? For context, Proton is roughly A600700 people company by now that's 14 years old. I think my role is as part of the management team reporting to my principal, the Ceoi work very closely with our CEO, our CTO and the general and the senior management team to make sure that the organization is developing and executing in the direction we want. This year specifically, currently I'm focused also on the engineering side of the organization. Speaker 2 Tell us a little bit about. Speaker 1 What you were doing before Proton? Speaker 2 What led you to the staff role? Actually, let's start before Proton entirely. Speaker 3 By education, I am an electrical engineer with a business engineering business degree on top. So I did the bachelor in electrical engineering here also in Zurich at ETH, the day what you would call a business administration focus. On top of that, doing my masters, I then end up first going into technology consulting very, very briefly with Accenture, then ended up in the world of finance in London doing derivative sales, which is a completely different field. How do I find my way? This was very, very early in my career. How do I find my way to Proton, right. So since now I've been now this is my 10th year at Proton and by now I found my way to Proton. I think after doing finance, that experience was great. I would do it again any day, but I think something was missing there. And I think it was the feeling of building something new, building something where you know who your end customer is, what they're doing with it and what kind of impact it has on their lives. I first had a bit of a sideway typical period of hey, the original career I chose didn't seem to work out the way I wanted to, what do I do? At the time, I connected with one of my fellow colleagues from university. We spent a year trying to build our own company for a while. As part of that work, we actually discovered Proton. So we were in a security space at the time, Proton and base of Switzerland. Proton was another up and coming company. This was 2017 or 16 even in the privacy space. So we reached out and then 5 cold emails later with no answers, I think 3 months almost passed. We almost gave up on the last e-mail. Andy, Proton CEO and founder, actually answered and that's how we got talking. So we started with a collaboration there, which eventually morphed into me joining Proton full time as a product manager. Yeah, that was 2017. So it's nine years in my 10th year now. Launching Products and Scaling the Company So you, you were looking for a career change, you were looking for something with a little bit more meaning. You clearly chose it. I mean, privacy start up with a very clear mission and a very strong vision, especially in kind of a highly changing kind of dynamic environment where privacy seems to fall by the wayside for most companies. Your chief of staff role didn't start as a chief of staff when you joined Proton. It started in product. What was the path for you trying to drive this mission, this vision forward as a product manager first? What were you focused on? I know Protons built out of multiple divisions, business units. How did you get your feeling for the organization as a product manager first? Speaker 3 For context, ProtonMail started as a company focusing on encrypted emails, right? So this was product number one. Originally the company started 1014. So by the time I joined or I made contact in 2016 and officially joined 2017, we were Proton at the time was already working under the hood on product number 2. So when I joined, basically the idea was, hey, and the CEO back then company was smaller thirty pitch P pill. The question was, Hey, we've got a lot going on right now. And we're also, by the way, on the weekends building this new product. Why don't you join and take that over? All right, So that was kind of my initial first job, so to speak. I joined, helped launch the second product in the line up, which was brought on VPN. But then zooming out a little bit, right, how did I spend the next couple of years here? I helped with the launch of 1 product that was really a 0, going from nothing to a first product that afterwards had the chance on switching around a little bit within the ecosystem that Proton was building, right. So then I had a chance to work on ProtonMail, started there from first building what was one product on one particular platform, so Windows and Mac, to actually working my way to, you know, hey, leading product and this with a small product team and the engineering organization there. Then afterwards switched into a more cross organizational team, still in a product role. So that was for the accounts and payments team. So you can think of that As for people familiar with, you know, general software apps, this is a team that is a central one that builds reusable modules that all of our applications can use, right? So if you think of, hey, there's a calendar application, there's a mail application, there's maybe drive application, right? What the accounts and payments team does is it builds the common pieces such as, hey, when a first user comes around, these are the standard user journeys that people go through. This is how the one they pay, want to pay. So the things I get used across the organization. So Long story short, I had a chance to cycle through various different business units as you, as you know, also I got the chance to, you know, shape the growth trajectory of the company, right? So when I joined, we were 30 ish, now we're at 600 ish. So all of this was I think before my chief of staff role kind of gave me very good context of, hey, this is what the core of the business of building new products. These are our customers, right? Because being APM at an early stage start up with 30 people, you do everything you do, you know, from writing marketing copy to answering customer support tickets to the funniest story was doing a house call to one of our most VIP customers back then, literally renting a car, going to their place, figuring out why does their set up not work. So it's really the typical start up by both, you know, getting into the trenches of what are we actually doing, who are our customers, what are we building for them and really having a deep understanding for this is how the business operates, at least on the, you know, product building side. Speaker 2 Yeah. Scaling Challenges and New Organizational Design I mean, talk about pulling up your sleeves in some regards. So you're cutting your teeth across these different business units at some point and the company's growing or headcount is growing astronomically exponentially. At what point does the Chief of Staff role start to emerge within the company? And then at what point do you set your eyes on it or is was it a natural conversation that starts to happen between you and your CEO, the rest of the executive team? How did that kind of come about? Speaker 3 The idea of chief of staff role, I think we first came to you know as part of conversation in 2022. So I think at that time the company was maybe already, don't quote me on this, but maybe like 300 to 400 people at that time. It has scaled, you know, an order of magnitude since 2 from where we started. What has changed at the time and what made the chief Gustavo even necessary? So the company, as we went through scaling, right? There are lots of different challenges as you scale a company. And one of those is, of course, how do you manage complexity, right? There's many more people working together, there's many moving parts. How do you make sure the dependencies are clear, people are moving in the right direction at the right pace? The company took a step that was very calculated a trade off, right? A major decision for how to organize the business, which is the inside was our customers come to us because they seek private solutions for certain parts of their lives, right? They come for either for e-mail, they come for, you know, scheduling, they come for the storing the photos or the files, right? And then their realization was that, hey, people come primarily to solve one problem first, which we can address with a single product, right? And then once they've adopted this product, only then do they start to discovering all the other products that we have. From this realization basically became also the organizational design, which is to say, hey, let's structure our business in a way that serves different customer segments, right? So we have one business unit that focuses on people that come to us for communication, for intercom to e-mail and calendar. There's another division that focuses on, hey, people who come over and want to store their files with us in an encrypted way, in a private manner. This is how the organization structure itself around what we call internally business units. So you can think of them as the smallest ones are maybe 1215, largest ones are you know 50 plus today. The defining factor is they build a product that is understood by the customer. It faces directly external facing and they will have their own PNL that they will have to run right and internally they have their own teams cross functional. So this is from product managers to designers to engineers to testers to everything that it takes to basically launch and operate a product that our customers can use a lot of context. But how did the chief of staff roll then come right? So this was a conscious decision. So we created, the point was we may not be perfect at doing dependencies. So at least optimize for, you know, delivering value to our customers as quickly as possible. So we created dedicated teams that can operate on themselves. The trade off was that now of course at an organizational level, there was less layer kind of connective tissue between these business units. Of course we still have people collaborating, of course we still had shared goals and so on and so forth. But the primary organizational structure was, hey, this is a customer, this is what they need, solve it, build storage. At one point, as you scale and as each of these business units gets larger and larger, the amount of effort that is needed to cross coordinate between them grew also significantly. The one or two people at a time that had this global overview were our CEO and our CTO, and the load became kind of hard to manage with two people. So this is when the conversation started off, hey, the organization probably needs more cross collaboration between the business units. So then the conversation started from, hey, why don't we consider chief of staff? In my particular case, I was approached for this conversation. So I can't claim that, hey, I knew, you know, chief of staff is my next thing. That's not how it happened. But in this case, it was a conversation with at the time I was already reporting to the CEO and you were saying, look, as we scale, we probably need more work to be done at the cross business unit level. So I'm considering a chief of staff. Would you be up for it? And the rest is history, so to speak. Operating as a 'Loose Radical' on the Management Team And and your role generally sits kind of outside those functional lines. You play a dotted line between those business units. What for you does that mean day-to-day, especially since you're kind of spending the majority of your time, I would imagine with product and engineering? How does that look as you kind of think about the different products, different business units across Proton? Who are you spending your time with? What are some of the functions that take up the majority of your capacity? And is that are those recurring things or are there like strategic projects that come out of the blue across any one of those teams that end up being a company priority or a huge distraction that you have to say no to? Speaker 3 Reality is messy, right? So I would probably say it's somewhere in the middle. Exactly all everything. So I should just answer yes, right. But jokes aside. So again, a bit of framing, right? So Proton has a senior management team, right? That's roughly 10 people. That includes our our C level people, but also, you know, general managers for some of our business units and for very specific functions. So think of, you know, even though we have business units, we have a central marketing functions. There will be AVP of marketing, there will be, you know, AVP of people. Think of it as, you know, different functions that together steer the company. So I am part of that team. I've been as Chief of staff now three years. You can classify my work as there is a primary field and then everything else that also gets out on top. So my primary field kind of changes depending on the business needs as we grow, right throughout my last three years and has changed three times already. My current focus of this primary area is engineering organization, even though I don't have, you know, direct functional line recording to any of these business units, right? Currently, for example, I also have personal targets right together that I agree with my principle of, hey, you are responsible for delivering on, you know, certain metrics for the engineering organization, certain strategic projects that we need to get done. So it's not that I don't have targets and get to choose whatever the hell I do everyday. But indeed, there is the primary part that has clear the lovables measurable KP is achievable things, but then also the general expectations. And also how I interpret this role is look, you need to see what is happening in the organization. You have to understand the context of this is what the company is trying to achieve, be it on our annual targets, on our multi year plans, right? And you need to be able to see the context and see, hey, are the different parts of the organization moving accordingly? Is information missing? Is maybe one team pulling slightly in a different direction than another one? Hey, let's get those teams together and get everybody aligned and moving in the right direction. Speaker 2 It's an interesting concept here that you're bringing up and we're kind of skirting along the seams here, if you will. One thing that I've heard pretty consistently on this podcast and from dozens of other conversations with chiefs of staff at the more senior end of the chief of staff spectrum, you tend to find the chief of staff describe his or her role as very organization focused. And this person that is very much helping to align teams and move projects forward and is actively volunteering and working, you know, building things side by side with the rest of the company. That is quite different from a chief of staff that describes their role as, you know, more of this support function for their principal. And it's not that that's wrong necessarily. They're just two different ways of describing how the function can play a part inside the company. And so I'm curious, because we haven't talked about your principle at all. Actually, the word hasn't come up. And so I'm curious, how does that world kind of look for you as you think about what he or she needs day-to-day and how you make their priorities come to life, but acknowledging the fact that their priorities are not always totally in line with the organization's priorities, Sometimes they are different. Bridging the Gap Between Vision and Ground-Level Execution Definitely, I think this goes with the theme of, you know, chief of staff is a different role in every company, right? And therefore comes also the spectrum that you describe from, you know, very much very closely in line with your principal versus very well. I wouldn't say completely independent, but more looser kind of looking roaming around the organization, kind of looking at what's there, right. So in my case, my principal officially is I report to the Ceoi guess you would call him my principal, but in terms of, you know, really day-to-day, what does that mean, right. I don't have would you maybe could expect a, you know, daily morning call with my CEO and my principal saying like, hey, what's the priorities for today? What are we doing in that sense? Our CEO is a bit of a special person. He organizes his life the way he needs to. He does it Andy, our founder and CEO still after 14 years. Instead, I think what we do is we have a weekly meeting at least once. And then besides that, it's a matter of actual kind of conversation exchanges on Slack, internally on calls as needed on specific topics, right? In terms of, I think your question was how would you characterize his relationship with your principal, right? Or how would you describe your work? I would say it's more that I think he takes the mentality of, look, you're here to make sure that the organization is moving. You're here to make sure that, you know, hey, the vision that we collectively come up with and I come up with for the organization gets executed properly. Here's a problem that you need to focus on right now, which this year and last year is engineering and how do we scale the company as we go. But generally the expectation is whatever it takes to make sure that we actually hit our targets right. And then I think this is both can be very confusing and also very liberating because there is nobody that tells you what you need to do. I think what is helpful for me here is having had the long track record and history and context of the organization, I can kind of see smokes or fires or, you know, potential areas for leverage and opportunity from a distance. And then I can at least go because we start pulling at the threat and see, hey, it's just something that we need to do to continue on your answer, I think one of the questions was what about the case where maybe the reality does not immediately look like what your principal is thinking, right? I think your your point was how do you align those right? Speaker 2 Where you may have to push back or contradict or. Speaker 3 Yeah, that's actually a very good, very recent example, actually. So coincidentally, it also touches upon the topic of AI. So what we do is, you know, as part of steering the company, we run an OK R system. So objects and key results, that's basically how we set targets for every team. And then that's done once on an annual basis, but also then on every quarter we do reviews, right? So most recently this was April when we reviewed the transition between Q1 and Q2, right? AI is a big topic. Everybody's kind of looking into how do we make our work more efficiently, right? But one of the realizations that I had at the time was, hey, actually as we were attending all of these and reviewing the company performance in between these conversations, we had a kind of offhand conversation with my principal, my CEO. And he's like, which goes along lines of, hey, how much do you actually believe? You know, what is the future? What is your vision on what an AI enabled future looks like? And he had his view on it and he shared it. And my point to him on that call was, and well, from everything that we've seen, it doesn't seem like, you know, this is actually aligned, right? This is what the team is executing perhaps on the ground, ground level, right, versus perhaps what you're thinking ahead, you know, X years into the future. There's a bit of a disconnect there, right? And then from there onwards, there were, let's say 2-3 follow up conversations in our weeklies to then come to the conclusion together and see the point that, hey, I think it might be good to make this gap maybe more explicit, pull in certain folks from the organization. And now basically that's one of the work streams that I'm pushing forward. Speaker 2 That's incredible, and I want to get into that work stream in just a second. I want to pick out two points that you just mentioned. You used this term the first time we chatted, you called yourself a loose radical on the management team, which was a phrase I really like and stuck with me, and I've used it since to the point that you were just making. That only exists when your principal truly has your back to be that loose radical moving across the organization, helping see kind of projects and initiatives through without any clear kind of direct oversight from other managers, right? That means that you truly do need to step into a leadership position to make sure that you're focused on the right things and to say no to the things that are, you know, not going to be a priority for you that day. And that's not easy for I think a more junior chief of staff to do. But it is clearly one of the stepping stones that more senior leaders in this role have found themselves driving. I think what you shared was a great example of that, the kind of final example that you did share in terms of seeing how your CEO was talking about the future of AI and kind of protons place in that future. And then acknowledging that you having worked from the ground up day-to-day, right? And seeing kind of what the teams are focused on and where that energy is being placed to say, hey, that's actually not what the team is doing. Like you might be saying these things to yourself or out loud, but our OK, ours, our priorities are not aligned to that vision today. And so how do we adjust course? I think that's one of the clearest examples of where a chief of staff could truly be effective. Because too often, and not that I'm calling your CEO living in the clouds, but it's very easy for an executive team to be too far away from the actual work to acknowledge that like, hey, actually maybe we're not doing the things that we said we're. Speaker 1 Actually supposed to be doing. Speaker 3 Yeah, I think you're absolutely right. If I ever can just go back on two of those items, right? One is the shift from having this kind of very loose and loosely defined mission, what you figuring out what you need to work on? I think one thing that was good and at least in my trajectory was I think deciding on what is important to work on is basically at the heart of the product manager function, right? It's really like, hey, figuring out what is it that we actually need to build for the customer, right? And which problem do we need to tackle in order to, you know, have the highest impact. I think that was part of the product function that was transferable for this particular role. So that was helpful for sure. And then the other thing also just to to relativize a bit right in my description before, of course, I want to give teams credit where it's due, right? It's like I'm, I'm exaggerating for a fact that, you know, hey, there's a complete difference. It's not that, you know, our teams aren't completely doing AI at all. I think it's more the question of the intensity level of how much are we engaging with this, right? And that you would want. And is it on a trajectory that you would want in terms of speed and of execution of actually hitting the vision that we want? Some teams are further than others. Some teams are, you know, tackling different problems, though by all means we have some very good teams in here. Speaker 2 To your earlier point, this is one of the challenges of having an organization whose kind of culture and structure is based around business units. And it's not that that was the wrong decision. Lots of companies are built this way. But it comes with its own set of trade-offs that can, you know, be very difficult to kind of walk back. And so in the sense that it's created lots of different silos and teams are moving at different speeds, they may be focused on different things that are not totally correlated or aligned with the broader vision. How do you, as a chief of staff, counterbalance some of that? What does business unit coherence actually look like when you're working against it? Speaker 3 I wouldn't say the best outcome is 100% cohesion given for the nature of where we are. So we're building a workspace, right, with lots of different products and we're organized by big business units. We're still launching new products, right? There's a good pipeline of, hey, here are the new products that are coming. So naturally, some of the business units will be in a different stage of the development than others, right? So you will have business units that have been up and running ProtonMail, right? For example, it's our products in like 14 years. It's very established large customer base. They have a very different risk profile, very different kinds of problems to tackle. Then say a new product that we're that might be coming out, you know, a couple of months from now, right? They're still building V1 of this and they're thinking about, hey, how do I go to market? How do I make a splash? How do I get my first customers right? So the things now that at each of this stage that you need to worry about is very different. So again, I think one thing that helps is having had history of working through various different types of products, I understand roughly kind of like which are the main challenges each business units face at their life cycle. So that's helpful. So I can tell that for example, going back to our current very specific tricks for engineering, right, a very established product on the engineering side probably needs a tighter leash on, you know, hey, how reliable is our service if we're about to break into the business segment, business for business customers, reliability is a much, much higher bar than for example, a, you know, soon to be launched product. There are kinds of challenges are, you know, standardization might not be a problem yet for them, right? So they have very different natures of challenges. So to make the general point, I think really it's absorbing context, understanding how the business works, what are the main activities of the business does right. And I think Proton is basically split into figuring out which problems do we solve and how do we solve it with building a piece of software. And then on the marketing side, how do we attract them and how do we retain them by these are kind of like the very, very broad strokes, the three main targets. And then we can think about, hey, if there is misalignments between the different business units in these three areas, does it matter right now at this stage, yes or no? Speaker 2 I'm curious when you say absorb context, what are the tools or frameworks that you're using to try to understand or ascertain like what is happening across the business? Who may have potential blockers? Are these team stand ups? Are they one on ones harvesting Slack for, you know, channel data between team members? This is scraping all of your tools. Like how do you kind of think about context, especially in this state of? Speaker 3 Yeah, it's very varied. So if I had to try and structure this, my activities, I would probably structure it around 2-3 areas. So one is more, I would say targets and execution plans like, hey, understanding where are we trying to go? What are our goals? What are specific targets we're going to hit for each of the major ties we're going to hit, which are the major big pillar investments, right, that we need to see to completion? And what are the plans for the people responsible, right, understanding, Are those things on track, right? So this is more classic, well, hey, which goals are we chasing? What's our strategy to actually make this happen? And how do we do execution on there? Is execution on track, right? And I think the second part is people, I think maintaining very strong exchanges with key partners to the things that I need to drive forward. That is, you know, be it our CTO, together we shape the engineering organization, be it with our people team, because a big part of growing the organization is of course, levelling up our people, getting the right people into the door and creating the right environment for our people. And then the last one is perhaps I would say tools and automation. This can be things from basics like hey, I create my own dashboard and our internal analytics tool to pick out the metrics that I care about. We have an internal tool that we use to display our internal metrics, both business as well as engineering or other product metrics. For example, I might care about. Hey, I will look at are we generally on target for the whole company, right? Are we actually going the right direction? Which business unit is you know, progressing more, which is less? Which business unit is, you know, having problems with user acquisition versus churn versus something else? And then all the way down to, hey, maybe for, you know, scaling our engineering organization, some of the metrics aren't even there, right? So part of my work day-to-day will be to work with the teams to make sure that, you know, the right metrics are put in place so we can track the evolution of the organization as a whole. So I think goals, product tracking, right? This is classical strategy, product management, people making relationships and exchanges with them, hearing from them. Also what is happening in the area. And that might be relevant for mine because at the end of the day, running an organization is still very much a people related business as much as tooling and AI comes into play. Proton's Approach to AI and User Privacy Let's talk about the AI since we're on the topic and it's been mentioned a few times already. You're working in an incredible organization again and has a very strong mission since day one, privacy kind of being at the heart of that. And yet the company, like many companies, is trying to move very quickly on AI, both inside in terms of the tools that you and your team are adopting, but also within the product itself. And you know, the entire identities, right, really built on not compromising user trust. Talk to me a little bit about the tension that may exist within the company in terms of business units may have different versions of this executive team members may have different versions of kind of what that world looks like. And don't get into details that are going to get us in trouble. Talk to us a little bit about kind of what that looks like and what your role as a chief of staff is in helping shape that common vision. Speaker 3 So First off, credit words due, right? I don't think we have a lot of smart people. We have a lot of dedicated people who are very mission driven and they care deeply about what we're building. So all the differences that we were talking about, right? I think one thing that is clear is that people understand the mission by privacy by default is really that, hey, whatever solution we cook up, whatever solution we build for our end customers, it needs to have the interests of our end users who are also our end customers at heart. As an example, nobody is going to build, you know, an AI tool at Proton that is, oh, by the way, it's going to sell all your data to, you know, a third party in order to monetize this better so we can actually run this right. So luckily, I don't need to do any work there. I think where the tension shows up, you kind of mentioned it a bit, right? There's like 2 areas. 1 is what does AI have on the products that we built for end customers? What do end customer expect from us, right? And then second part is like, hey, how do we apply this internally? How do we reap whatever benefit there might be for efficiency, for execution without compromising our values, right? On the external one, I think it's pretty straightforward. So we have one part that we call Lumo. Lumo is our Mai assistant team. It is built from the ground up with the same principles that we also have for our other products, right? So we run in a controlled on our own infrastructure. We bake in the promise that hey, your data is yours, we cannot access it, we reduce it, right? We have 0 knowledge encryption. We cannot read the conversations that you are having with Lumo so that you can have proper, sensitive, very deep conversations about topics that actually matter, be it, for example, personal finances, your personal health issues, your personal whatever, right? Things that are very, very sensitive. And our customers may care deeply about that. You know, only a restricted group of people or nobody should have access to, right? In this case, we also don't. So this principle is built still at the core of our products, even if it's AI, and that needs to stay where it becomes more interesting. And also a part where I'm actually working on actively is, you know, how do we deploy AI internally across all different organizations in a way that, you know, even though we want to build fast, even though we want to go, you know, make sure that we are staying competitive and bringing the right products that our customers expect. We don't want to compromise user data or privacy as part of that process, right? So for us, what does that mean? It's hard to know exactly because you know, people will always say about the either super fast or super slow and adopting AII think we are taking a very measured approach, right? One thing is we spend a lot of time clearly defining for example, on the engineering side, the policies, the rules that govern our AI usage for engineering. That is one part we can dive into that right. The second part is luckily also thanks to the very mission driven nature of our company, the people that we have attracted, they really do deeply care, right. So if we for example, were to set a policy that were to go against the values that we stand for, the company would call it out. People will immediately tell us and and we get a lot of feedback even with this good policy. But like, hey, have you thought about this and what happens with customer data there, right. If we really were to even try and force AI at all costs, we don't care. I think half the company will probably walk. They are really, really keeping us on our toes in terms of how we deploy also AI internally and then also the process in like how we deploy it, who has access, access controls and so on and so forth. Without diving too deep into the weeds policy wise, I think 1 hot buzzword is like AI slop, right? It's like AI creates too much stuff for a very concrete thing, right? I think the headlines that you read nowadays is, you know, AI is making junior software developers, you know, irrelevant because it's, it's of course faster than any any human ever could, right? Fair. I think that's true. And then you'll immediately hit the next problem, which I think a lot of people still trying to solve is fine. Now you're spitting out code 10 times faster. But who's actually reviewing this? I think the principles that we have, which are also codified in our policies, AI right now is a tool. You are the software engineer and you are still the person responsible and accountable for whatever comes out of this, right? So you need to have read every single line or understand and be able to defend a merge request in front of an audience and say like, hey, this is the reason why it's being done. And then do not ask AI to create a piece of code that you don't even understand. I'm going to create a piece of code in the language I have no idea about. I'm going to try to pass this off as good quality and try to deploy the production. No, that's unacceptable, at least for us internally right now, right? And then the human is still in the loop. So I would say we're not in the fully agentic world yet where, you know, 5 agents can autonomously decide, hey, this is a new feature we built. I'm going to deploy automatic to production. No, that's not where we're at right now. So Proton really right now has an iteration where we say, look, nothing gets pushed to production to our customers without human oversight. And the human is still accountable and needs to be able to say that, you know, this piece is good and has the same quality bar that our customers have come to expect from us. Speaker 2 Proton being kind of an engineering LED or product LED, let's say organization, do those same principles and values, do you tend to find kind of equal counterparts in other parts of the company, marketing, finance where the data can be just a sensitive, but there may be less of this kind of engineering principle approach to how we use the tools and what the output is and how we gauge that output being good, better balance or not good at all? Speaker 3 Yeah. The short answer would be we might have that, but I think we have the right controls in place to not have that. I talked about engineering, right, because roughly speaking, Proton being, you know, 600 plus people, half of that organization is hearing, right? So engineering is a big part of it, right? So we need to solve it for a big part of it first. But then of course the other half is also just as important. We take a similar approach, right? So we will have the same policy, the same kind of people and the process that we have in place. For example, one general thing that applies everywhere is right, no customer PII can go into third party tools externally and this is throughout the line. For example, in addition to using perhaps third party AI tools, we have an in house to ourselves that we use, right? Lumo is also used internally for our own work flows because we host that, because we control that, we know where that is going. This is the default answer actually. That is the default preferred tooling that we have. So we know that sensitive information stays in house. So we have a very fine grained, I think, policy and and rule book of what can go where and people need to adhere to it. Speaker 2 This is why I think Proton is such an interesting case study, an example for other organizations going through AI transformation, but with much looser values, much looser guidelines, if at all. You're not starting from scratch because Proton was literally built on these same sets of principles. And so it's really just applying this to kind of a new landscape, new sets of technologies and tools. And yes, there will be some adaptation in terms of how things translate, you know, not apples to apples, but for many companies that don't have that. We hear this first hand from lots of chiefs of staff who are finding themselves caught in the middle of their CE OS vision for AI and a radically different interpretation of that across the organization in terms of how it can actually be applied and trying to make sense of that. It doesn't sound like that's the case for you. Why Culture is Key for AI Adoption But since you aren't working on these things today, what what would your approach be? You know, as a chief of staff working in a company that like may be having a much harder time on a, a transformation. I'm curious if you have any words of advice. Speaker 3 If I had to think of, you know, what are the factors that are making this relatively easy? I mean, it's not easy. We're still figuring it out, right, like everybody else. But I think that the factors that are making our AI adoption easier and easier to guide is I think that there's this term that people are making sure have heard more culture eats strategy for breakfast. And I think what we have underpinning all of our actions that we're taking, the policies that we're writing, the people that we have there is really the culture that we've actually built here. Proton, right? It's really mission driven. People really do care about this, right? And if you find yourself perhaps in an environment where that is less pronounced, look, there's nothing wrong, right? Because if you're inherent business is not privacy, maybe this is not what your business needs to care about, right? So I think maybe for other chief of Staffs, you should have that conversation with your management team, with your principal, right? What are actually the guiding principles that we care about? What are our values? What is the culture that we want to enforce, right? Because that will permeate into every decision you take, whether it's a new tool, whether it's a new technology, whether it's a new competitor engineering field, right? That's the underpinning, very soft but very sticky driver for all the decisions that you make. Speaker 2 AI transformation is not about which tools you're adopting or how you're using those tools. There are much greater foundational elements in an organization and even makes that a possibility. And it's more important, arguably, for a chief of staff and the executive team as a whole to be focused on those things first before they try to figure out, you know, what are we going to be doing with whatever tool we're paying for now. Speaker 3 Yeah, because I could imagine if that's not clear, right, as you said, you maybe you find yourself in discussions about, hey, why aren't we doing XY and Z connecting, you know, all possible data sources to this tool that we're buying or why are we doing this? And then somebody else pushes back. What about, you know, retaining access to a certain data X, right? Like this seems to be a fundamental disconnect, right. And if you probably see some of those conversations or questions being asked, I think maybe it's good to take a step back and ask ourselves like, hey, what do we care about as part of this whole roll out, this whole transformation? Looking Ahead: The CoS Role and Key Episode Insights Yeah, so long. We've talked about your past, we've talked about your present. I'm curious for you, what are you most excited about in the future, whether it's for you professionally or or things going on at Proton? Speaker 3 Proton's still very exciting, as in it keeps me up. As in it keeps you on your toes, right? It's a good way, I hope in a good way. In the past, people have asked me like, hey, what's it like working at Proton, right? It's kind of like a choose your own adventure book. The good news is it's a company that's on the right trajectory. People have internally used the word, you know, hey, it's like flying a rocket, but building the engine or building the the airplane while you're flying it, right. In that sense, there's still a lot to be accomplished, lots of interesting challenges as we scale Proton. So definitely looking forward to that. The challenge is not going to get any easier as we go and as we grow. So that's part on a personal level, look, I think the chief of staff role is actually super fulfilling, right? It is ambiguous. It is the two sides of the coin, right? It's like because I'm a loose radical on the management team, right, My role and doesn't stay the same over time. So I think you were asking also previously, hey, doesn't this set you up for irrelevance, right? Eventually. But actually, I think depending on how you look at it, this is the opportunity, right? Because you get to reinvent yourself basically every two years. You get the freedom and the risk that comes with it of, you know, picking the challenges that are the most valuable and the hardest to solve along the way. And I think that open field and where this is going, I think that's something to look forward to for me personally. Speaker 2 Long thank you so much for the time. Appreciate you being here. Speaker 3 Scott, thank you so much for having me. Speaker 1 All right, that's a wrap on today's episode with the long blue from Proton. A couple things I'm still thinking about from this conversation. That phrase he used, loose radical on the management team. I think that's one of the cleanest descriptions I've heard of what a senior chief of staff actually is. And it only works when your principal genuinely has your back. That's the part that doesn't get talked about enough. Then there was the AI piece. I thought his framing was really useful. Before you figure out which tools you're rolling out or how fast you're moving, you need to know what actually the company stands for as an organization, because that's what's going to guide every decision you make. And if that's not clear, you're going to keep having the same circular conversations anyway. Really grateful for Delong taking the time today. Links to connect with him are in the show notes. If you've got something out of this episode, please subscribe and leave a review or share it with another operator who might find it valuable. It genuinely helps us grow. We'll see you next time.