The Exchanges

Q And what are those, out of curiosity? When does, uh, synchronous communication make sense? Obviously you talked about off-sites, but when else, especially in a digital environment?

A Yeah, I can give, I think maybe I'll give some specific examples, and then maybe we can try to extract a principle from that, but, uh, one is, say, a project kickoff. It's something that I personally like to do. Not every project has one of these, but it's like, we are officially starting this group of people on this project on this timeline, and we're all here. We can all answer questions, and we, there's like an emotional resonance of that Moments of like, we're starting this project for the next eight weeks. We're going to be working on this together. Uh, another is daily standups. We default to doing them async, but if you're working on a really high velocity project, just having 30 minutes of sync time in the morning really makes a big impact. And we found for a lot of our projects, people might do two weeks of daily standups that are synchronous live Answer questions, get more context, higher velocity, and then they eventually transition into async standups, where at the end of the day, they just say, this is what I worked on. If you have any questions, let me know. And so, uh, I think other aspects, if we do a lot of our code review live, we default to async. You might record a loom walking through the code, but if it's a really big pull request with, I don't know, a thousand lines of code changed and It's on a part of the code base that people are unfamiliar with, and y…

AI assessment note: “one is, say, a project kickoff. It's something that I personally like to do.”

Q You talked about some of the, the biggest surprises that you had when you went back and looked at two years worth of data. Can you talk about a few of those surprises, things that you weren't expecting to see when you looked at the information?

A Yeah, I think, The biggest surprise, if I had to pick a singular one, is how there were things that I assumed took up a lot more of my time than they actually did. Uh, specifically recruiting was one where I felt like I had been spending a huge chunk of my time on recruiting. And it was actually a tiny, tiny fraction of my time. And I write a lot of memos. I write a lot of strategy. In fact, right after the end of this podcast, I'm going to be writing a strategy document. And I assume this was, you know, 20, 30% of my time, some very large number, because I, the output is so high, but it was really, again, five percent, some small number. And so, uh, it, it made me realize that it's, having these feedback loops, measuring things is super important. Uh, if you, and I'll give a specific example recently, I was talking to a CEO friend of mine who was saying he's probably spends 20% of his time on just board management. And that might be true. What is definitely true is that he feels like he's spending 20% of his time on board management, which means it is sucking the life energy out of him, and that, that alone is a problem, but he might be surprised to discover maybe it's one percent, maybe it's 40%, but he really doesn't know how much of his time is being pulled into that, because he doesn't measure it. So, uh, I think having an empirical record when you have a, a stated set of …

AI assessment note: “specifically recruiting was one where I felt like I had been spending a huge chunk”

Q I want to talk about, kind of, your three processes for async, but just on meetings for a second, So we've talked about, kind of, like, what are the situations in which having synchronous meetings makes sense? To you, what is a good meeting? How do you actually have a good meeting? Because knowing when to have one is important, but also most people are really bad at running meetings.

A Yeah. I think we'll specify meetings that are internal to the organization, because that tends to be the most soul-crushing of the types of meetings that people have. In general, there should be an action item at the end. Often several. So I would say that is more of a, an artifact of a good meeting is we had some, we had a, we had an hour and a half meeting today with our eng team discussing a number of things around how we do documentation, a number of processes, and we ended with three or four action items. So we're going to fix this thing. We're going to change this to that. We all agree that we're going to start doing this thing that way. Um, that's, I think a good form of, uh, of, of, Recognition of when a meeting was successful is if you have a series of action items. I think every meeting should have an agenda and you, I, I'm pretty militant about this. If I have one-on-ones scheduled for all of my direct reports, but I've made it very clear and everyone's on the same page that a one-on-one calendar hold is not an entitlement. It is, it is a hold so that I don't double book myself. And if there's nothing on the agenda in the notion doc that we share, Then the meeting gets canceled. No question. Like, the night before, I go through all of them, and they all get canceled. It's actually, that process is handled by my EAs. They actually go through themselves, and they cance…

AI assessment note: “In general, there should be an action item at the end. Often several.”

Q that comes to mind is High cultural importance, and I don't know how else to say it other than like increased motivation or energy around a, um, around a project or a goal where there's a level of emphasis and importance feels like it can, almost like because you have an async culture, when you have a synchronous meeting, it naturally signals importance. What, what are your thoughts on that?

A I think that makes sense. I would add a third category to it as well, which is, uh, There's a degree to which alignment really matters for these things. Uh, in the early days of a project, uh, there's a rule in aviation. It's like the one in 60 rule. I forget what it's called. Where if you're, if you're one degree off for a certain, you end up in a completely different country. Yep. It's, it's surprising how a very, very tiny difference in direction can compound as you, as you continue to diverge. And so for early projects, when it's really important that people are aligned and you correct and you get to that point where it's like, okay, I'm now confident that if you spent the next five days in a cave and came out with a deliverable, it would be the deliverable that we need. As opposed to you come back and it's like, what on earth is this? How, how did you misunderstand so horribly what it is that I thought I was saying that we ended up in this place where we have to undo all of this and we have to start over. So I think in the early days when alignment is a really critical piece, feedback, constant communication, I think that's, if, if, if you ever expect there to be more than one or two back and forth with a person, this is, this is where tools like Slack are really, really bad. I could nerd out on, on internal comms all day, but Slack is a, it's a synchronous tool that masqu…

AI assessment note: “I think that makes sense. I would add a third category to it as well”

Q never seen empirically actually how, um, Time spent for CEOs evolves over the course of building their business, and I think it's wildly helpful for early stage entrepreneurs who are trying to understand how should they be spending their time. So I'm just going to give you the floor with one simple question. How do you spend your time today, and how has it evolved over time since starting levels?

A Yeah, I think it depends. It's going to depend a lot on the role. As the CEO of the company and as one of the co-founders, I largely fill gaps wherever it's needed within the company. And so for the first couple of years, when I was working the most hours, things were the most busy was during our, our seed round fundraise. Today, it's a lot of software development. I took almost a two year hiatus from the code base, which in retrospect was a bit of a mistake, but I'm now fully back in it, getting ramped up. Um, I think when you have a software product, I think having leadership who knows how to build software is really important, and the further away you get from the actual building of what it is that your product is, uh, the worse your product is ultimately going to be. So the last two, three, four months, I've been really back hands-on in the code base, and so as As growth is more of a priority, I focus more on growth. As we had a big recruiting cycle, I focus more of my time on recruiting. And so I think the, the nature of being a co-founder CEO is you have to be 80% good at basically everything. It's, uh, it's kind of a strange role. You have to be able to write a passable growth strategy and a passable clinical strategy and a clinical, you have to be able to write a passable regulatory strategy. You have to be able to write Read and write legal docs and contracts. You don'…

AI assessment note: “Today, it's a lot of software development. I took almost a two year hiatus”

Q Yeah, but I think what's interesting, at least around that bucket, is I feel like a lot of it is related to unblocking, and one, one of the things that you talk about as you look at your time is you say that unblocking is a top priority. Can you talk about what that actually means and why it's so important for you to be the one to unblock?

A Yeah. The, the CEO ultimately is the final decision maker on anything that needs to be done at the company. And so having the flattest structure possible while still being able to manage effectively, uh, I was asked recently if I could have dinner with anyone in the world and learn something from them, I would pick the CEO NVIDIA because he has A very interesting management structure. He has something like 40 direct reports, which is crazy, but he does it in such a way that it's not super burdensome, and the intent behind it, from what I understand, is that the closer you get to the ultimate decision maker, the faster decisions get made. The fewer committees that you have to form, the faster your company will go, and so figuring out how he managed to build that type of structure I find very interesting. So whenever there's a tie, whenever things need to be collaboratively escalated, when there's a misunderstanding or disagreement, the faster that gets to a decision point, the better, and the CEO is always the tiebreaker in any discussion. If the CEO says, we're going left, then we go left, and there's no further discussion. So that's, that's really the, the, the underlying intent there.

AI assessment note: “the closer you get to the ultimate decision maker, the faster decisions get made”

Q say, 80% good at all these things if you just hire the right leaders? If you just hire the right leaders, then why would you need to spend your time on any of these things other than, you know, Hiring the right people, go forward strategy, and maintaining the right culture? Because clearly you've done more than those things over the last number of years. What would be your answer?

A Yeah, I think it's, there's a degree to which that's true, but your ability to assess and know whether they are the right people, if you're willing to put that to luck, then sure, you can just hope that you hired the right person, and you have no way of assessing the quality of their work output. If you can write a strategy that's okay, If you have advisors, if you have people you can lean on and learn from, you can really only assess the quality of those people by knowing how to do it yourself. I, I've, I've seen a lot of friends of mine who run software companies. I'm not going to throw any of them under the bus, but a lot of them are non-technical and they criticize their engineering team. They're upset about product velocity, but they don't know how to write software. And so these criticisms are come, they come from a place out of frustration, probably frustration from their lack of knowledge on how this actually works. And so I think Knowing how to do it is really important.

AI assessment note: “you can really only assess the quality of those people by knowing how to do it yourself.”

Q like masquerading as As one of those talks when it really should have been a more structured, uh, talk that actually had an outcome and an agenda. And one last question on meetings is, uh, do you have a cutoff on number of people to have in a meeting? Do you think, like, once you get to a certain number, just ever the marginal benefit, uh, continues to go down?

A Yeah, we, I would say there's not, like, a strict rule. We try to limit it to three is really the number, but there were absolutely meetings where it's useful to have more. Like, we have a weekly, uh, The whole engineering team meets once a week to go over anything relevant to the whole org, and right now our engineering team is only 10 people, so it's not a huge meeting, but these things can become really onerous. I, I have a friend who runs a larger company than I do, and the problem is once somebody gets invited to like the exec meeting You, you, it's really hard to then uninvite them. They feel like they've been demoted by not being invited. And this is weird to me because like, if somebody told me I don't have to go to a meeting, I'm thrilled. I, especially cause I know I can always watch the recording later, but there is a poll to like being in the meeting where these things are happening. I think it becomes less important as people get used to our culture. Cause all of our meetings, including our one-on-ones are recorded and shared by default to the whole company. So our board meetings, generally speaking, are shared with the whole company. And so people can see what we're talking about. It's not a secret. So people don't feel like they need to be there necessarily because they can still see what, what's being discussed. So, uh, I would also add that we, we do have water…

AI assessment note: “We try to limit it to three is really the number”

Q Um, I should have asked this in the beginning, but I'll ask it now. Why did you decide to start tracking things so methodically by the 15 minute end? For, for people who are listening that maybe want to, uh, embark on this maniacal journey like you, how does one actually go about tracking by the 15 minute over an extended period of time?

A You know, it's, it started out of necessity. I was doing a lot of consulting work, software development, contracting work, and I just had to keep track of what I was spending my time on in order to build the right people, and so it came out of necessity. Just, I spent 30 minutes on this project, I spent 90 minutes on this project, and being able to retroactively go back and build those people the correct number of hours was ultimately what started this, and then I started to realize I started adding more and more of what I was doing to my list, and so knowing how I'm spending my time, some of these things, like, I keep track of how much time I spend with my wife, and some people find it weird that I have that in my calendar, but it is a real priority for me to know that I'm spending an appropriate amount of time, and I mean that in, like, the positive way. You can get a whole week and see, holy cow, I only spent three hours with her this week. That's really bad. I need to change my priorities. So knowing how you're spending your time and having a record of that allows you to just live the values that you, you have.

AI assessment note: “it started out of necessity. I was doing a lot of consulting work”

Q into action, because I love what you do in concept, but, um, maybe this is a story that I tell myself, but it feels like a lot of work. It feels like it'll take a lot of time to track your time. One, is that the case? And two, you If I wanted to start doing this, like, what is the system for tracking in the way that you do?

A I think one is starting by being proactive, which is the, the biggest, the biggest thing to do is to get rid of your to-do list and put all of your action items on your calendar. Block off the amount of time that you think it's going to take. You have a to-do list. Everyone has a list of things that they want to do, and just translate all of that into time on your calendar. And then as the day progresses, check your calendar, see what you're supposed to be doing right now. A lot, there are a lot of failure modes with this. One from somebody I was talking to recently is he found that when the time came to do task X, he just didn't feel like doing it, and then he would bump it to the next day, and then he wouldn't feel like it, and he'd bump it to the next day. I think that's, that's not a problem of the process. That's a problem of you trying to force yourself to do something that you don't want to do. This is a, you're, you're trying to solve An upstream problem with tactics. Really, if you don't like doing a certain thing, like writing investor updates, it means that you should probably have somebody else do it because clearly you're not a reliable person on this specific type of task. I, I tell my team this all the time when they ask me, okay, Sam, we need you to do X, Y, or Z. I say, look, I can do it once, but you cannot rely on me to do this because I'm going to find this …

AI assessment note: “get rid of your to-do list and put all of your action items on your calendar”

Q didn't have a chance to really dive in, but at least from 2019 to midway through 2021, When I look at the distribution of time, obviously things vacillate, but by and large, the two biggest buckets look to be network and then information processing. Can you talk about what those actually mean in practice? Like, what you're actually doing in those buckets and why they're worthy of that much time?

A Yeah, network is a tricky one because it's, it's in many ways a catch-all for interpersonal engagements that don't have a specific agenda. So, like, I met with a founder this morning, and we were just talking about industry stuff. I might be able to connect him with a couple other people who are looking for their next roles, but there's no specific agenda there. Um, a lot of those, if I was to retroactively update this, which would just be way too much effort, but if I was to retroactively do this, it might, it might turn out 20% of those are actually recruited in retrospect, because they ended up joining, or You know, 15% ended up being investor relations, because in retrospect, I now know that that person became a levels investor. And so it's sort of a, an ambiguous bucket of no real intent.

AI assessment note: “network is a tricky one because it's, it's in many ways a catch-all”