Dan Siroker of Rewind.ai and Optimizely on AI and Memory

Michael Koenig: Hello and welcome to between two COOs, where phenomenal Chief operating officers come to share their knowledge, advice, and at the very end of Crazy Story, I'm your host, Michael Koenig, and occasionally I'd like to bring on a CEO to bring their perspective and also learn a bit about their industry that in mind. Our guest today is Dan Roker, co-founder and CEO of Rewind. A personalized tool that uses AI to basically augment human memory. Dan calls it a co pilot for your mind. And if Dan's name rings a bell, it's because he found it optimizely and grew it to 120 million in ARR. He was also a key member of President Obama's campaign back in 2008, where he helped pioneer the use of data to optimize fundraising, media, the entire thing, which was a major factor in the election win. It's also when we first got to know each [00:01:00] other, albeit briefly. Dan, welcome. Thanks for being here.

Dan Siroker: Thanks for having me. So

Michael Koenig: here's how I'm thinking about this. There's so much to talk about. Let's start with Rewind dive into some AGI and then get into operations because after all, this is an Ops podcast. Sound like a plan? Sounds great. All right, wonderful. Well, I mentioned a little bit about Rewind. Give us the [00:01:22] nickel

Dan Siroker: tour. Yeah, so Rewind is a truly personalized AI. You know, there's a lot of AI solutions out there today that lets you ask any general question and you'll get the same or similar response to anyone else. the inspiration for our product came out of this realization that we can use technology to augment our human capabilities and give us superpowers. Uh, for me, that happened personally when, uh, about, uh, 20 years ago when I started to go deaf. Um, and in my twenties, I, I realized only when getting a hearing aid, how much I had lost and that ability to gain it all back in an instant really felt like getting a superpower. You had to lose a sense and gain it back again [00:02:00] is quite magical. And that put me on a hunt for ways technology can augment human capabilities. And so, uh, so lucky for, for us is that, that intersected with this now very exciting wave of technology around, uh, uh, AI. So we, we've kind of pioneered using all the data about you, everything you've seen, said, or heard on your Mac. And now, um, as of yesterday on your iPhone. And, um, and making it searchable and making it, uh, AI able, uh, if that's now a verb, uh, where you can now use AI to ask any question about anything you've seen, said, or heard. So simple things as writing you an email or reminding you like, Hey, how do I know this Michael guy? And going back and actually seeing all the times we've interacted, seeing the touch points of our past synthesized for you is, uh, really does feel like getting a superpower.

Michael Koenig: And I love it because I have a horrible memory and it's getting worse, right? Because that's what happens since we get older. And so when I saw what you were working on, I was like, Oh, this is pretty interesting. I call him like my partner. She we like to joke that she's my [00:03:00] external hard drive.

Dan Siroker: Mine, mine is mine as well. Maybe this will. Yeah. Yeah. It was interesting. Actually memory. Well, I w I was, so I, I started this journey where, you know, I had a hearing loss and I started going on this research expedition on what are other human capabilities like hearing that people don't realize is quite as bad as it is. It gets worse as you get older. And the one that really stood out was memory. You know, studies show that we forget 90 percent of our experiences after just one week. It follows what's called the forgetting curve. Um, and not only that, but typically your memory gets worse every year after you turn 20. So you turn 20, your memory peaks, and then every year thereafter it gets worse. Um, and on top of that, most people don't realize it. So you've got this like already base level of memory. That's pretty bad. You don't, you don't know what you don't know. So you don't realize how much you're forgetting often. Um, and then it gets worse every year and that's kind of a boiling frog dynamic where you, yeah, you kind of thought you had good memory and then all of a sudden you start realizing, boy, I don't know how I know this person or what was my action item for that last meeting? Um, and that's where we saw this huge opportunity to use AI to augment. our natural biological limited memory [00:04:00] with a, you know, super intelligent memory. This is

Michael Koenig: interesting because we're talking about memory augmentation. We're also talking about what happens as you get older. Every now and then I like to take a question from the audience and this one comes from my colleague Justin Riley, who's the CEO of WaveLow, which is one of the companies that makes up the Two Cows family. Justin asks, and he takes a very similar approach, by the way, he thinks about AI not as replacement, but as augmentation. And so he asks, how does rewind thread that needle of being a useful recall tool, but not changing the fundamental makeup of how a human experiences memory. And I think that's the end of the quote, but I think an extension of that is if we rely so much on something else to remember things for us, does our memory just, go. Out the door because we're no longer trying to recall.

Dan Siroker: It's a, it's a really fascinating question. Uh, one that I've thought about. And, and if you look at the history of this [00:05:00] question, uh, the analogies can actually maybe be pretty illuminating. Um, you know, there was a time in my childhood where I had to remember the phone number of my friends. If I wanted to actually call them, I had to remember it, and I had to dial it in. I didn't actually have to put an area code in, which is probably crazy for kids these days. Why? Why? You don't have to put an area code in. You just put in the number, and it would, the phone would ring, and you'd, you know, and I knew my Best five friends. I knew their phone numbers today. I don't know anyone's phone numbers. You know, I know my wife's number. I know my own number because people ask me for that. But then everything else is in my phone. And like, am I any worse off because I don't remember my friend's phone numbers? No, like it's just a piece of technology that can augment something that otherwise I would have to memorize. And now I don't. You know, another example is, is getting from point A to point B. You know, when I grew up, you had to know cross streets. You had to know, like, how to get from one place to another because of, you know, that's the way, you know, just how we navigated the world. And then, you know, this website called MapQuest came along and then now today's Google Maps. And you don't have to know any of that. You don't have to remember cross streets. You just put in your destination. Uh, it will dynamically decide how best to get you there through the traffic that's out in the road. And your better, [00:06:00] life is better for it. So I, I don't think we lose anything by outsourcing parts of our, Mind or parts of our memory to technology. Um, you know, it's not everything. It's the parts that technology is well served to store things like numbers, things like directions and locations. And there's so many other parts of your mind that are like that, that you don't need to remember. It feels maybe nostalgic to think, Oh, well, would I be my own self if I don't remember all of these facts about something? Um, and I would argue that it actually creates space. It creates creativity in your mind. So you're not obsessed with what was that thing I supposed to do? You're not racking your brain. Every moment you're racking your brain trying to remember something is a moment. You're not thinking about the future or connecting the dots or being creative. So I see it very much as a pedestal in which, you know, our minds can evolve to become greater and better, not, um, you know, worse and limited because we're now in some dystopian, uh, you know, Wally universe. We're just sitting around in a scooter, uh, you know, disconnected from reality. I think it's quite the opposite.

Michael Koenig: And it's interesting, another area, because I was also thinking about how [00:07:00] does tech augment my memory in general? And I have Google Home displays throughout my house. I also have two little kids. And so many of the memories that I have forgotten come back to me as soon as I see a photo of my kid eating ice cream upside down on, something crazy like that. And I'm like, Oh, wow. And the more I see these pictures, the more I remember it. And I assume the same thing is happening for my kids. Maybe they remember things. In greater detail or even in greater breadth at age, from age two than they would have otherwise. So I think that'll also be an interesting experiment.

Dan Siroker: Yeah, in many ways our memories are encoded in a way that are very hard to retrieve linearly, but can be retrieved through recognition. You know, photos do that, music does that for some people, smells do that for others. So people even with advanced dementia can recall and like retrieve parts of their mind that many had thought [00:08:00] were gone forever. Through these kind of sensorial experiences of smells and music and dance. And so that's, I think, another way, again, we're augmenting, making us better through technology, not making us worse, because the worst thing is having in your mind, you can't retrieve it. It's like a, you know, a vault with no key. Uh, but with, with technology, you can, you can get to those memories and, and, and cherish them.

Michael Koenig: Yeah, absolutely. So we talked about rewind. And a big part of it, the way it works is it's a desktop application that you install and it essentially sees everything on your desktop. It listens to your conversations, both the input and the output. And then it does this magical thing of somehow compressing it to not take up a lot of space and then integrates chat gpt as you described so that you can query everything, it's all indexed, etc. There's obviously some privacy. Things that we need to talk about. How have you approached

Dan Siroker: that? The approach starts with the realization that there is nothing more important than the trust and [00:09:00] privacy of our users. And we started with that. I mean, it's exciting, you know, we call it privacy or private by design or privacy first. We started as the, as the foundation of thinking how we built this product. Um, we could have built what we built entirely as a cloud based web app. That's my background. Optimizely was a SAS web product. I know JavaScript quite well. I could have helped build that. Instead, we did the crazy thing, is we pivoted. from what was a product we had before called Scribe, an online meeting bot, and we pivoted from a cloud based technology that I knew intimately well to an entirely new domain of technology I knew nothing about, native macOS, low level system, you know, optimized software, because of the fact that now we could offer a product entirely local, or at least all the recordings would be stored locally. That was a decision we made, recognizing that the convenience and value of what we're offering was quite high, but at the same time, If you could give that and give people the assurance that no one like us like if we if we're compelled as a company to look at somebody's recordings, we [00:10:00] can't like we were subpoenaed, we can't do it. There's nothing it's it's on your desk. We don't have access to it. Your employer doesn't have access to it. Only you have access to your data. And so we thought if we could do that and give you the convenience, that's the best of both worlds. So that's where we Um, started with was just this premise that, okay, we need to focus on privacy. First approach. What we do is unique. It's new. It's different for people to get comfortable with it. Initially, it needs to show that we care and not just that we talk the talk, but we walk the walk. What we found is interesting. So we started that approach and surprisingly, a lot of people were like, Oh, this is amazing. I'll use it right away. And then as soon as they get to that aha moment, using our product. It's usually, you know, an example would be something that they couldn't have found any other way. Or maybe they're writing an email and they hit discard instead of send and, and, or they, a tab crash or some information that they really needed disappeared and, and rewind was able to. to go back to it. Uh, as soon as that first moment happens, people's minds shift. Now it's all about convenience. And then people are starting to ask, Hey, can you synchronize this data across my phone and my Mac? And [00:11:00] can I ask questions on my Mac, my phone and all of the things that we were so, so careful to avoid in the beginning, uh, when people become. Uh, you know, activated and have seen that first magical moment, then their minds have shift shifts. And so now that's why we're launching. We just launched iPhone. You know, soon we'll be launching the ability to, uh, in an end to end encrypted way, synchronized data across your devices. So we're, we're sort of paving the path as our users are pushing us toward more and more, uh, of what they want. But it all, again, starts with a privacy first approach. Everything is stored locally on your machine. We don't have access, access to it. Um, and I think that's sort of foundational to giving the right, uh, return on investment for users as they think about adopting a new different kind of technology. You have

Michael Koenig: integrated chat GPT in one of the videos, or maybe it was on your website, you mentioned that the information, because it is stored locally. Isn't used to train any sort of models or, oh, I'm there. How does that work with chat GPT? Because everything you pop into it [00:12:00] is used to train

Dan Siroker: the model. Yeah, actually that's incorrect. Um, if you use the API for GPT four. Open AI has been very clear that they do not, uh, train any of their models using the GPT 4 API. If you use the consumer chat GPT interface on the website, they, I think, likely train. But they're very clear. It's like black and white on their website. Anything used through their API is not used to train. And not only that, but it's only retained for 30 days. So after 30 days, anything sent to the GPT 4 API is deleted. Um, and again, even in those 30 days, nothing, nothing is used for training.

Michael Koenig: I mentioned privacy and security. You were recently on a, and there's a little bit of a lead in here. You were recently on this week at startups with Jason Calacanis and he came in, by the way, pretty [00:13:00] fiery, but you calmed him down at the end. Nicely done. It's no easy feat. Thank you. But he looked at it, and he described his reaction to this as terrified and intrigued, and that he was uncomfortable using it now. He also pointed out, potentially some of the legal liabilities. You already talked about how this stuff can't be subpoenaed. One of your use cases on your site though is for execs, like our COO listeners. The most sensitive matters of a company Come across our desktops. How do we get

Dan Siroker: comfortable with? Yeah, I think if you zoom out and look at the broader, um, world and imagine where we're going to be in 10 years, the companies that succeed or fail, are we the ones that embrace AI as a way to get. 10 percent 20 percent in some cases, two or three or 10 times better. And in those situations, the people who are actually the most successful in the roles as COOs are the ones that don't just [00:14:00] say, my job in 10 years will be the same, but slightly different. It's going to be, it's going to be fundamentally different. Like every decision I'm going to make is going to be informed through not just data, but AI synthesizing the data for me. Uh, I'm gonna be able to be in five meetings at once, not just one. Uh, it's, it's thinking that every minute of my day is precious. And if I can use technology to augment my memory, my capabilities. to give me superpowers. That is better for me. That's better for my company. That's better for the world. So I think it's, it's in that broader landscape that you have to ask yourself, okay, if in the world, unless you think AI is just hype and it's going to go away, it wouldn't, it may be some, maybe that's true. I, I think more likely are fundamentally the way we work every day will be different. And the companies that can actually, Credibly, like us, claim that privacy is at the core of what we do. We store data in a way that is almost impossible. It's by sort of the way it's foundationally designed, um, is the most privacy centric model. Um, we, I think, will be your partner in, you know, Surviving in that world, you know, so you either have to say it's [00:15:00] not going to be a big deal. The way I think is not going to happen. Um, or you have to say, you know, I'm just going to pretend like it is a big deal. I'm going to be one of the companies that decides, you know, it's not important for us. And then your company is not going to exist. So like in that sense, like I didn't create this whole AI wave. I'm just trying to help companies write it. And if I can do that in a way that respects the privacy of their company, their data, my guess is maybe today your CIO is not going to be trusted. That keen, but in three or four years when they see the alternatives, which is maybe entirely cloud based solutions or ones in which data is used to train models. I think then it'll be pretty clear that this is the path. I

Michael Koenig: hadn't expected to have privacy so early on. in AI. I had expected this to be something much further down the road. The, we were to follow the trend of social media or search or things like this, privacy really doesn't get added in until people are up in arms about it. And so I love that you took the privacy and security approach first. [00:16:00] And John, my, my CISO are we cool with this now? Can I use it? I really want to. No, I'm sure he still says no, but I hope that we can bring him along. You talked about 10 years out, and you started talking about how the companies that don't embrace AI now are just set to fail. Similarly, I think something can be said around remote work and productivity and learning to be able to work and excel in that field. If we think about operations and the applications of AI, And I think you are very qualified to answer this question, having scaled optimizely from nothing to 120 mil. What do we think the application beyond Rewind of AI In ops

Dan Siroker: could be, yeah, the one word I would use is leverage. And the analogy I'll draw is like, imagine thinking about ops in a world before computers and a world after computers. [00:17:00] There's a lot of people in your company doing operations who had to do wrote manual tasks, uh, with pen and paper. And, uh, you needed an army of people to do things that with the advent of computers now could be automated away. I think similarly, if not more so. AI is going to do the same thing. It's going to be a giant lever in which any one human being at your company can do 10 or 100 times more. Um, and to do it joyfully, not because they're working harder, because they're working smarter with technology. And so the future looks like smaller teams doing more. Companies that don't embrace AI are going to be bigger teams that are going to be greater, uh, you know, competitive threat, uh, at greater competitive threat markets will look at their company and say, Hey, why, why do you have, you know, every, everything you think today of what is a good, uh, PNL or balance sheet for your company, uh, in the future, when you compare your company and your industry with one that is a competitor, that is braced AI, it's going to look fundamentally different. So expectations will rise from the public markets, from investors. Um, and, and it'll be, it'll be silly. Like I look at our company today, today, we're 18 people at [00:18:00] rewind. Many of us use AI every day to do our jobs better. I, you know, no knock on Optimizely, but I feel like as a team of 18, we're able to do more quickly than a team of 450 I had when I was running Optimizely. And a big part of that is AI, it's technology, it's leverage. And so I think the future, if you're, if you're a COO and thinking about your company in the future, you should ask yourself not, you know, what is my, you know, how do I get more people in my org chart? It's like, how do I make every one person in my company better, more augmented, more capable? Sure, there'll be some parts that maybe no longer exist, some teams that no longer need to exist. But for the most part, the leverage you're going to get isn't from replacing people with AI, it's going to be augmenting your existing people and making them 10 or 100 times more productive.

Michael Koenig: I completely agree with you, of course. When GPT 4 came out, Basically, we said to everyone in the company, Hey, look, we don't know what this is going to be like, but you can definitely do more faster. And, I have the app installed on my phone, [00:19:00] and I just ask myself at the end of the day, How did GPT 4 make my life faster and how did I do more today with it? I think the real big challenge for the company is twofold. The first is around privacy. We saw what happened with Samsung, how they had uploaded, all of this proprietary IP. So first, it's making sure that everyone throughout the company, and when you're a 1, 300 person company, that's complicated, but making sure everyone has an appreciation. Okay, don't Upload this or don't, use your work email. The other part to it is that it's such a broad tool and it asks people to essentially be creative in figuring out how it can augment it. You guys have gone in and you've taken a much more specified approach to it. And I just, I love that. I don't know [00:20:00] that there's a question in there, but those are just more like observations and feelings.

Dan Siroker: Yeah. I'll share, I'll share one insight we had over the last few weeks, which is, um, related, which is, you know, you're right. That AI today, chat GPT is so broad. You don't even know what to ask. We have found that the most powerful and valuable features we offer are ones where we sort of bridge the gap between that ambiguity and the task you're doing. A perfect example is today with rewind. If you're in a meeting, a zoom meeting with a colleague or a team meeting, um, the first thing we'll prompt up is say, join and record. We know you're about to join a meeting. You join, that's pretty helpful. At the end of the meeting, you get a little prompt that says, uh, summarize and draft email. And what it will do is take everything you said, everything your colleague said, summarize it. And turn it into like a five bulleted list in a draft in your email inbox to the attendees of the meeting, knowing you because we know who's in the meeting. We know their names. And that's a perfect example of using technology to bridge the gap between the ambiguity of what's possible with the actual like that is. You know, five things less, you know, you know, if you're especially if you're back to back [00:21:00] meetings, there's no way you're going to write a summary. This makes you look good and for your colleagues, you look like you've got a superpower. Uh, and that's the thing that I think AI is best served to do is like bridge that gap between the things you're already going to do. But now instead of spending 15 minutes after every meeting, summarizing, sending notes, action items, have the AI do it for you. That's

Michael Koenig: phenomenal because oftentimes we'll have meetings, there will be great ideas that come up, great action items. And so often, unfortunately, things just get dropped. And. Talk was great, but then action didn't happen. This kind of ensures that at least people remember what the meeting was about. And then, go to the next step and integrate with Asana, turn

Dan Siroker: them into action. Yeah, we found, meetings are a black hole for information. It's like you go there, everyone talks, like I said earlier, the forgetting curve tells us 90 percent of what's going to happen in that meeting is forgotten the next weekly team meeting. Uh, and so just taking what's already said in, in, in a very. Um, expensive, uh, uh, setting everyone's salary for that one hours is, it's not, it's not, it's not insignificant. [00:22:00] And then taking that and turning the value insights of that into actual action, follow up, um, that's super valuable to companies. You

Michael Koenig: mentioned 18 people. You have built so much, right? A small team doing very big things here, 450 people at Optimizely. Obviously the tech has come a long way in enabling that. You have a phenomenally low burn rate and. A lot of cash in the bank because of that. A lot of runway, a lot of ability to get out there, figure things out. How has AI, and I was going to say, oh, how have you done that? AI, we use AI, right? Outside of Rewind and the feature you just talked about though, how do you all think about Applying a I to do more with

Dan Siroker: less. Yeah, I mean, it's not just a I, by the way, it's a phenomenal team. Like I think hiring people well above, you know, if I look back the 1st 18 people at at at optimizing the kind of people I work with, they wouldn't have worked with me. I didn't know what I was doing. I was just like they could have worked anywhere. They would have been making three or four times as much money working as [00:23:00] you know, fancy, fancy, um, individual contributor at You know, Facebook or Google at the time. So I just think I, I, I feel fortunate that I've been able to hire truly exceptional people. And, you know, the, the myth of the 10 X engineers is certainly true, and with ai it's the a hundred x engineers, the one that. You know, can do more with less. So that's I think a key part is is really senior folks. It honestly, if I'm candid, makes me really worried what it's like for kids graduating today, you know, in a world where, you know, people who are have experience, who can use AI to make them even better. You know, it's gonna be a tough, you know, I personally, we don't hire junior engineers, we only hire, you know, senior staff level, principal level engineers, designers, designers. Um, and, um, you know, it's, anyway, so that's a, that's a, uh, that's a topic for what that means for society. But, um, yeah, I mean, to do more with less comes with great people. I think another thing that I have learned the hard way is focus. I, you know, I always at Optimize used to think I was focused, but then in hindsight now, boy, was I, uh, was I not, you know, and, and, um, you know, Steve Jobs has this, [00:24:00] uh, famous quote, uh, that, uh, Johnny Ivey recently retold, which was that, you know, focus to him. meant not doing something that with every bone in your body you want to do, but you decide not to do that thing that you really, really want to do because you're focusing on something else. Uh, and that is a test, you know, and I, I remind myself and my team, they're probably sick of me hearing it, but, uh, I, I say this saying that the main thing is that the main thing should say the main thing. Uh, I say this over and over again because that is the, the curse of, of death for, for any. Organization, especially startups is doing too much. It's indigestion, not starvation. It's saying, you know, let's do Mac and iPhone and Android and Windows all on day one. You know, we, we took a very different approach. We started with Mac very, very purposefully. You know, we were getting requests. We had a huge waiting list of people dying to have our product on iPhone, but we waited. We spent, you know, really. Uh, a year perfecting it, learning, and then when it came to iPhone, we could do it in three months because we learned all of these things around compression, around [00:25:00] integration with AI, and like all the things that made it possible to do it right the first time. So I, I think that's probably the broader lesson is great, great team, really experienced team, great AI, and, and deep focus. I love that.

Michael Koenig: Okay. The main thing is the main thing is the main thing. Three times? Is that it?

Dan Siroker: The main thing is that the main thing should stay the main thing. Ah, should stay the main thing. And I think I heard, I first heard this, I think HP had this as a cultural saying. I think somebody else famously quoted it, but I'd heard it from somebody who formerly worked at HP and she said this and I thought, ever since, and this was maybe 10 years ago, I heard it. And I only in the last three years, uh, as I've been working on, Uh, Rewind, have I really embraced that mantra?

Michael Koenig: That's fantastic. Okay. Might have to take that on. Normally my thing is, if someone's hey, this seems broken, I, but what do we do? How do we fix it? I just, I default to four words, which is just who does what, when. It's actually all quite simple. Figure that out, and it'll be smooth. The main thing is that the main thing should stay the main thing. You started [00:26:00] getting into what jobs in the future look like. And I swear I didn't share my notes with you, but I actually have a question on this. Let's shift gears to be, or shift back. Let's talk about kids and education. We're both dads. Right now, education, there's a huge focus on STEM. And in light of how quickly we're seeing the tech advances, especially for solving these structured problems, is STEM going to be as important in 20 years? And I'm asking you to look in the crystal ball here, but

Dan Siroker: Yeah, yeah. I think, uh, yes. Uh, and maybe even more important, you know, maybe, uh, maybe we will be standing on the shoulders of giants so much so in 10 or 20 years that like, it's just too much, too inconceivable to be so low in the implementation details of, of science, um, implementation digital science. I mean like the physics, the underlying, cause there's, you know, I [00:27:00] think somebody once looked at all of the things you would need to know to make a common, you know, pencil. And it was like, Huge amounts of science, like metallurgy and, you know, chemistry and all this. Like, so like maybe it's, you know, and I look at my education, there's so many things I learned that I probably don't use every day today. So like learning how to learn will always be important and learning how to build will always be important. I think stem really helps you learn how to build, how to be creative, how to use the tools that are at your disposal at any, at your disposal at any time to build something, you know, today that's maybe much more rudimentary than it will be in 10 or 20 years. But the idea of always taking what you have around you and building something with it, um, I think that's, uh, going to be. It's hard to imagine that that's not going to be the most important thing people have to be good at in the future. Uh, I was lucky in, in high school, I was on the robotics team, and that was like one of the things I really learned. It was like how to build physical things out of the things around you. We had this competition, and it was all about, like, how do you use the constraint of the materials you have to build something great? And, um, and I, I [00:28:00] similarly, you know, and optimizely, I, I viewed that very much so. It was like, you know, what, what can JavaScript let you do? That wasn't possible before, uh, with browsers getting as fast as they are. And that's what sort of inspired the first version of Optimizely, client side A B testing. That was easy for marketers to use. And similarly today with Mac, it's very similar. Like, we, we would not exist today if not for Apple Silicon and M1 chips and M2 chips. And the hardware, you know, you know, we started the product basically realizing, wow, actually, This stuff is really good and fast. How can we use it to solve this problem that before this technology existed, we couldn't solve. So I think that meta skill of sort of looking around and thinking, how do I use technology to solve a real problem in the world? Um, that I think STEM teaches really well and I think will be important, uh, for our

Michael Koenig: kids and their kids. Yeah, we still have a really long way to go before we have intergalactic travel. It's really not going anywhere, is it? Yeah. Yeah. You mentioned M1, M2 silicon. I was going to I was very interested. There was a podcast that I saw. I didn't have time to listen to it. But this is very deep [00:29:00] into tech for you all to go, Oh my gosh, look at the M1 and M2. Like, how can we leverage that? That didn't happen overnight. Take us into your brain.

Dan Siroker: Yeah, and by the way, you said something that I just want to remember, remind you of. You said, there's a podcast I didn't have time to listen to. That is the persona of the person for whom our product is perfect. Instead of listening to the podcast, ask the question, what was that thing Dan said about M1, M2? Like, those are the kinds of things that, why would you spend an hour of your life listening to a podcast when you can synthesize it, get the action that you're looking for? Anyway, I just thought that was like a perfect example of why our product and products like ours is going to make your life much better in the future. Yeah, I mean it,

Michael Koenig: so sorry. Would I be able to play it on 10 x speed and it would get it and we'd be no . Yeah. I

Dan Siroker: mean, or you just ask a question. But like, if you have podcasts, you just send it to Rewind and ask rewind, Hey, like, what did, what are the takeaways? Summarize it, or what, when Dan talks about M1, what did he say? You know, those are the kinds of things that you, you know, maybe you do now manually. Like, I do this too. Like, I'll be like, oh, I swear I saw this quote in a book. And I'm like, I, I, I'll, I'll listen to it on Audible. Instead I buy it on Kindle. Then I'm, [00:30:00] you know, command ing in the Kindle to try to find it's. Totally broken. I should just be able to send the thing I heard to Rewind, have Rewind do the synthesis for me, tell me, hey, this is the thing you were looking for. Um, and oh, by the way, here are the three other references to in the book. And like, you know, anyway, a bit of an aside there, but yeah. So your question is, you know, how do we get to this conclusion around M1, M2 and sort of the technology, the name of the technology? Well, it started with, you know, we, I kind of alluded to, we pivoted from another idea. So when we started the company, We had built, uh, what today is called like a meeting bot. It joins Zoom meetings, it's in the cloud, captures everything that's said in the meeting, transcribes it, does sort of like some, you know, gives you kind of like a pie chart of how long each person spoke. And that was, uh, that was kind of our first foray into our approach in giving humans superpowers. We thought meetings was a good place to start. But very quickly we realized there's so much more context you have outside the meeting that would make, Um, that, that you need to really make that successful. And we also realized that, you know, we are in this point where, you know, everyone felt weird about bots. Like, do I have a meeting, but the bot is kind of creepy, it's out in the [00:31:00] cloud. That's really a data privacy concern. Like that bot is somewhere, some, you know, some hard drive somewhere, you know, Google Cloud has the data of that. And so we thought with, with M1 and M2 coming out, we looked at it and, and the vision that I had for the company from the beginning finally became possible. Everything that we're doing in the cloud, we could do locally and if we could do it because the technology was good enough to do the compression, we could do transcription all locally. That's what we transcribe what you're saying all locally. So the privacy goal now is finally realizable. Like it wasn't before the technology like your computer would crawl to a stop if you try to do what we do today. Before Apple silicon. So that was really the reason we invested in it. It was a big, you know, we had a big debate and we had a whole strategy doc. Five different strategies were considering of which one of them is the one we ended up pursuing this kind of better memory strategy. Um, in fact, three of the people on the team voted that the last. Choice of the five. So it was not an obvious decision to focus on, uh, you know, learning, you know, Apple or Apple native development and learning Swift and how to build a native app, [00:32:00] but it was the one that was the most likely if it were to work to be a transformative technology. And it turned out it did. So I'm glad we did. You know, that was another example of something I learned from from optimizing is being decisive and focused. And it wasn't obviously the right choice, but in hindsight, working pretty well.

Michael Koenig: I would love to see those other four. Okay. So let me ask and only because I I've been, I have scar tissue. Talk to me about platform risk. This is something that runs on a machine, right? We talked about this at, ad nauseum. Do we, do you worry, right? You just did the iOS app. Apple is famous for being very private when it comes to iOS. Do you worry that some of the moves that Apple might make further down the road might disrupt Rewind's product?

Dan Siroker: You know, it's, it's definitely something I think about. There's four advantages we have today. Uh, and I'm [00:33:00] more worried, frankly, about Apple becoming a competitor than really, Pulling the rug out from underneath us because, you know, the good news is, uh, you know, when we started this company, there's one of two paths, either what we're doing is a good idea or a bad idea. If it's a bad idea, it would fail. It's a good idea. And so, uh, with it being a good idea, it will attract the, the, the attention of big tech, including Apple. Um, but we today benefit from a couple of things. One, what we do is weird. And when something's weird and maybe even borderline creepy, lots of people at a big company will tell you not to do it. You know, I know for a fact there are people at Apple who think that like the future of Apple services revenue will be something like what Rewind does. But for every one of those, you know, people fighting the fight, there's a hundred people at Apple to tell you no, and here's why not, and brand, and privacy, and concerns. And so it sort of comes down to basically the innovator's dilemma. Like we are, uh, and that's going to be a huge advantage for us. Um, I, I hope for a few years, I think eventually Apple will do what we do. Uh, but if you look at, uh, you know, something as comparable, if you look at like the Dropbox versus iCloud, you know, Apple tried to buy [00:34:00] Dropbox, they didn't want to sell, they sort of competed and built iCloud. That was first of all, many years after they probably should have. And in that market, that is a very low switching cost. You just drag the files from one to the other. We are a very different product. Everything you capture with our product is encrypted locally on your database. It is, you know, it's by design, it's not something that snooping eyes or, or, um, you know, other products can consume. So it's the switching costs are incredibly high. So that's our huge advantage and speed of execution. You know, we release 11 times a day. Uh, I listen, I give my cell phone number to every pro subscriber. Uh, when Tim Cook starts doing that, then I know they'll take us seriously. So, um, so yeah, so those are advantages today. So, um, that's more broadly around Apple as a competitor. I think from a platform perspective, you know, we don't break any rules. We play all by the same rules that they have in the APIs. You know, everything is a published API. Like if anything, they love what we do. You know, one of the first things we got when we announced our product was, this is so crazy, I did not expect this. We let, we first announced November 1st. Within days, I was getting people emailing me with receipts of [00:35:00] purchasing Apple, new Apple laptops, because we said we're only gonna work on M one chips on his people saying, I love what you're guys doing so much. I just went out and bought a 1900 laptop to use your product. So in that sense, we're apples. That's the reason Apple has built this great platform for developers like us to build software that drives hardware sales. So, um, so I'm not too worried about them pulling the rug up for us, given that they get quite a bit of revenue from, uh, From users who want our product.

Michael Koenig: Yeah. And that's also like the question that some will ask is what happens if Google does this? Then either they buy us or we're out of business. I don't know what to tell you, but. Nonetheless, this is how things develop. Well, I, I mean, I

Dan Siroker: think, I actually think we have a credible claim to build a successful independent company. If you look at the big tech companies, Google, this would be make, you know, and I, there's rumors even later in Sergey, when they built Google Desktop, their vision for that long ago was something like what we do today. But like I said, technology wasn't there yet, couldn't do it. But even Google, like they have a fundamental business model conflict. Their innovator's dilemma is like as soon as, if Google offered exactly what it did today. We would crush them because all we would have to [00:36:00] say is, do you want ads targeted better because of the data that you collect? And like, so they fund, I mean, we are a subscription service. You pay us money. We don't have access to the data. Google's approach would be very different. Uh, so I think they're the least of my concerns. I think Apple is the only of the big tech companies who could do this. But again, what we do is weird and creepy. Hopefully no one important at Apple is listening to this. Uh, and, and, uh, and you know, if as long as they think we're weird and creepy and we can fly out of the radar, I think we can build years of a head start. And you know what? Honestly, one day I, I love Apple. If we were to join forces with them, uh, I'm very happy to acquire them. I have no problem. Uh, you know, I have no problem with that. I think culturally it'd be a good fit. Um, it's just, you know, it's going to be a while before I think that's going to work out.

Michael Koenig: Yeah. No, it's, you gotta have goals, right? You gotta have goals. Let's get back into some operations a little. Now you have a chief of staff. Is

Dan Siroker: that correct? That's right. Yeah.

Michael Koenig: And you had that at Optimizely and now have it at Rewind as well. Tell me a little bit about that decision. [00:37:00] How do you, like, why a Chief of Staff? And then, how do you work

Dan Siroker: together? Yeah, um, yeah, we have a Chief of Staff here. I also had a COO at Optimizely, so I've definitely had that experience. But, yeah, we work together, uh, very collaboratively. I think of her as an extension of me. So the things that I would otherwise do but, um, but don't have time to do or She would be just as well suited, if not better, to do. She, she owns. Um, so there's a set of things in that camp. A lot of the administrative stuff, um, thinking through, we do quarterly retreats. We all go together. And so thinking through how to, to, to set those up for success. Um, you know, just recently, you know, a topic that was very important that I wanted to spend time as a group working through was psychological safety, talking about what that means and having a candid conversation with that. So she did a great job of owning that, thinking through how to have that conversation. Um, and so, you know, in that sense, I see it as just, um, an extension of me as a CEO. And, um, and she's able to bring her unique skills and perspective. You know, I'm much more excited about what's [00:38:00] possible and open minded and optimistic. She's much more, you know, execution oriented and sort of traditional COO kind of mindset. Uh, which is also a really good compliment. Um, and similarly at Optimize, I had a COO who also, you know, he actually ran a huge part of the company, but was fantastic as being a, a compliment to myself. And it was one of those things where, you know, one plus one equals three, where, you know, I could bring my unique perspective, that orientation can bring theirs, and together we get to a much better rational view of the world and, and a path. Interesting.

Michael Koenig: Tell me about. Did you have the COO and the chief of staff

Dan Siroker: simultaneously? I did, yeah. At Optimize? I did. Yeah. Yeah, yeah.

Michael Koenig: What was that like? How, what was that interaction? How did they work together?

Dan Siroker: Yeah. Um, I didn't see often how the two of them worked, you know, when I wasn't there. But together, we as a team worked really well. I think, um, the COO at optimize the, uh, fantastic, um, uh, one of the best executives ever hired, uh, he actually was promoted from within, he was our CFO actually, for a long period of time. And then, uh, you took on customer success and [00:39:00] manage that part of the company, uh, became our COO, uh, was a fantastic thought partner and, uh, was very, um, very much wanted in when I think coming to the company wanted a partner in the CEO to work together to. Build something great, uh, much more strategic. Uh, my chief of staff, I had some fantastic chief of staffs at, uh, at optimizely, many of them have grown up, gone, gone off to be, uh, uh, chiefs of staff to the stars like Elon Musk and, you know, the console brother. So I feel really good that I'm like a stepping stone on their ever progressing career of success. But, you know, I know what good looks like because I can, I can see, you know, where they ended up. So. So in those roles, I think, uh, you know, one of the things that I really valued was somebody who could take a very ambiguous, ill formed thought of mine, like I could, I could utter a sentence without much structure and they could actually translate it into the organization, kind of what you said earlier, like who, what, when, or whatever, like, you know, they could be a good at making sure that these often dumb ideas, but [00:40:00] sometimes there are some kernels of good ideas turned into new ideas. The next steps. They didn't just die on the vine, but they were sort of translated in the day to day. They're good at holding, sort of keeping processes in place to hold people accountable, making sure that the things we were talking about in one setting are translated to another. And, uh, yeah, it's, it's a fantastic role. Highly recommend. I can't imagine being a CEO without chief of staff or eventually probably

Michael Koenig: a COO. That's fantastic. Now you just. said, you know what good looks like, clearly. Everything that you just discussed though is, things that you learn and experience once they're in the role. How do you screen for that?

Dan Siroker: Pretty poor. I, I do not know because I've never really, I'm more lucky than good, you know, I've made some mishires. Um, and I honestly, I'm just more lucky than good. I do not know how to screen for really good chiefs of staffs or COOs. Um, like I said, our CEO is actually promoted from within. Uh, it's a really hard, it's a really hard job because a lot of it is also just that unique partnership with the CEO. And like, how do you, that, that intermeshing, you know, each CEO is different, what they need [00:41:00] in this role is different. Uh, you need somebody who's different enough from you, but compliments you in a way that actually together you're able to be, you know, if I were a perfect CEO, I could do all the things a CEO, CEO could do, but I'm not. I'm, I've got my flaws, I got my strengths, my weaknesses. And I need somebody who can compliment those and you know and mutually we respect like they're great at one part of the world and how to lead and I'm part great another and together we can do more. And I think that's actually if I look back at the best CEOs, it's them also recognizing that they're not the end all be all like they couldn't really do the job of CEO. They know that actually you know what I'm best suited for on this planet is to be a great COO. partnering with a CEO, and it's not so much like, Oh, I could just do their job. And I'm not there yet. It's like, it's easy. That truly is true partnership. Not a, you know, well, I'm just in some, you know, some other role. And one day I hope to be a CEO role.

Michael Koenig: When you were at Optimizely and steering the Helm you you were at, I think, 450 people or so. And Optimizely was ranked the number one, best place to [00:42:00] work by the San Francisco business times. You talk now about psychological safety, that certainly goes a long way to people feeling comfortable. How did you approach achieving that? How did you approach building such a fantastic place to be?

Dan Siroker: It's, it starts with values, um, and not just talking the talk around values, but, but walking the walk. So we early on at Optimizely made a commitment to a set of values, uh, it's spelled an acronym Optify. Ownership, passion, trust, integrity, fearlessness, and transparency with a Y at the end. So Optify. Um, and for each of those values, we had a very specific set of behaviors that lived up to those values. And those behaviors were very, they weren't just obvious, you know, integrity is kind of obvious, but a lot of companies like Enron on the day of their collapse, and we've walked in the lobby of Enron, it was actually emblazoned into stone, the word integrity. So, you know, there, you know, it's, it's, and that's what I mean by walking the walk. So I think we did that well, you know, I think one of the things I learned on how important it is to be. Uh, to at least put it pen to paper. People [00:43:00] talked, at Google, one of the things that killed me was, you know, I interviewed like a thousand people at Google, even though I was like a very low person on the totem pole. Like, we were hiring like crazy. And every person I interviewed, I had to assess them on how googly they were. And there was no definition for googlyness. And that was like the thing that killed me. Like, I decided if I want to start a company, at least I want to define what our culture is, so that it doesn't become a recipe for unconscious bias. It's not like, oh, they weren't a culture fit. Well, like, there's no definition of what our culture is. So, you know, and so that was my first goal. I was like, define what the culture is. And then the second was, how do we live those cultural values? And the way I did that was, anytime I had to make an important decision, I would always open a culture doc. We had a document that described it. I always went back to that as kind of the constitution. You know, I, I thought of it as very much like I'm a Supreme Court justice. I've got to look at the law to ask myself, what's going to best live up to these values. And some of the hardest decisions I made, I use as our, as our culture, as, or the values in our culture, as the. As the, the, the sort of deciding factor. And then when I made the decision, I would explain why. I said, the reason why I'm deciding to do this is because as our culture talked about, [00:44:00] fearlessness is important and much rather, you know, many decisions are two way doors. Blah, blah, blah, blah, blah. So that was the thing that I think helped it optimizely. I've evolved that substantially here at Rewind. Uh, I think we mistook cute and simple, uh, as a, as more valuable than like comprehensive and. Um, and, and, and so in, in Rewind, we have actually 14 cultural values, no fancy acronym. We have two high level themes of impact and teamwork, but, um, and we put our culture doc up there. If you go to rewind. ai slash careers, you'll see it. There's a culture doc link. It's the same notion doc we use internally as we share externally. You can see all the comments and that defines really, and like, and, and every quarter we do three sixties and every three 60, the key criteria, everything you evaluate each other on is. How well does this person live up to our cultural values? And that's another factor that like shows us we walk the walk. Like I'm held accountable to our culture by my team. Similarly, I hold them accountable. So we just create cultural values as the DNA of the company. And I think that's Uh, when everyone has that shared understanding, I think that creates psychological safety. Everyone has a clarity on what's expected of them. [00:45:00] Um, so that's how we've done it, I think. Fourteen,

Michael Koenig: do you foresee your values evolving similarly to GitHub, right? GitHub, or no, was it GitLab, excuse

Dan Siroker: me? Yeah, they absolutely, it's a living, every new hire, I tell them, our culture is a living document. And you can go to again, you literally go to rewind. I assess careers. You can see a culture memo there. We explain what our values are. You can see the revision history. We tweak them all the time. They're pretty well baked today, but we asked ourselves at a regular cadence. We do retrospectives every two months. We asked, we, we want to make sure that we're. We've, we've defined these values in a way that serve us. They're not meant to be lip service. They're not meant to be recruiting tactics. You know, we hope people come to us because of our values. Um, but only because it's a match for them, not because there's some kind of marketing gimmick, uh, because this is also how we hold each other accountable. Like we, we know every quarter, you'll know, how am I living up to transparency? How am I living up to the decisiveness? How am I living up to resourcefulness? You know, these are values. That we think will make us successful. And that's why we use them as the basis for

Michael Koenig: performance management. And then talk to me about [00:46:00] psychological safety, which is something that you mentioned previously. Tell us about that. Why was it something that. You focused on why now, and how?

Dan Siroker: Yeah, I mean, we, we focused on it in part because we felt there was an opportunity to have a shared language and vernacular. Uh, in many ways, just talking about it creates it. Uh, because people know, oh, Dan seems to care about psychological safety. Maybe that's something that's important to him. Uh, so I think, if you're thinking about talking about it, do talk about it, because I think the more you talk about it, the more people feel it. It's also something that, um, if I look back at some of my regrets about Optimizely there, there's a two by two quadrant. I can try to pull it up in psychological safety. And I think the two by two quadrant quadrant, um, the, the axes of the two by two quadrant one is, um, comfortable. Yes, no. And the other one is like accountable. Yes, no. Something like, or maybe it's something like. Um, uh, And, uh, And each of those quadrants, you know, they describe it as a certain thing. The ideal of psychological safety is like comfortable and candid and comfortable and accountability. Like you have the balance of, [00:47:00] you feel like you can say what you think, you don't have to worry about the repercussions of what you say, and you're being held accountable to a high standard. And that high standard doesn't create a lack of psychological safety. And optimizely, we started that way. And then over time, as we got successful, we had like, you know, amazing series a and B. Like we, we got this collective. disease of, um, of getting, you know, comfortable, but without accountability. We were, and we were kind of like the way I think it was like, we all had locked arms kind of kumbaya off a cliff. You know, I'd much rather somebody yell, there's a cliff, uh, you know, and be willing to admit that then everyone feel this false sense of, of security. And that's the thing I really wanted to avoid at Rewind is like, even though, and especially I think things are going well, I feel like it's my important, my job as a leader to remind people, like, it's not always going to be up into the right. Like there will be moments in time where. It's going to be hard. Like we're going to have times when who knows Apple launches exact clone of what we do and offers it for free. Like in those moments, we need to have that candor and that muscle around kind of a wartime mentality and high accountability for one another [00:48:00] that you lose when things are, things are good.

Michael Koenig: And it's operating as a unit. When you're talking about that, you can't operate as a unit. If people aren't feeling safe to be able to contribute.

Dan Siroker: Yeah. And it's, it's, it's also related to, I mean, just knowing that. Everyone, I expect everyone on the team, if they see something, say something like their job isn't just to be a cog in machine. I hired incredibly experienced, highly capable people. I pay them top of market compensation, not just because of the title that they were hired before, but because of their smart general per perspective point of view. And so if an engineer sees something wrong in our marketing, they should say something. If, you know, if our designer thinks that I'm making a bad decision, they should say something. You know, like that's part of the, the contract I've, I'm trying to build is that. I, I know I'm not perfect. I know I have a lot of dumb ideas. Uh, I want to create an environment where I can throw those dumb ideas out there and I get, the bad ones all get ripped to shreds and shot down. The good ones get, you know, watered a little bit and turn into something great. Uh, and not just for me, but everyone on the team as well. Yeah.

Michael Koenig: How do you Maintain that you [00:49:00] grew optimizely, right? You've talked a little bit about the lessons you've learned, what you've wanted to do differently. And this is your 18 people now. And, rewind is going to be around for, long, long time. Hopefully it will. I don't mean to throw in the, hopefully that was grim there. No, but as you go on it's just so

Dan Siroker: important. I think, well, first, yeah, I mean, I hope to do this as my life's work. I want to be doing Rewind for decades to come. And, um, you know, our vision is to give humans superpowers. And I think that's, it's going to be quite a while before we achieve. You know, the full scope of what that could mean. So I do think it starts with then, I guess, with a long term orientation. And from that, ironically, comes a conclusion, which is to hire slowly. Uh, you know, I, I've been very purposeful in, like, not making the mistake I made at optimizing. In optimizing, you raise a 28 million Series A, and then a 57 million Series B, and then, uh, That alone would have been fine. We raised a lot of money. Great. Got to cash in the bank. But then we coupled that with a secondary mistake of then spending it. Uh, you know, like it was hard for me as a first time CEO. I hired [00:50:00] this amazing head of marketing. They're like, Oh, we should, we need a comms person. We need this person. Then I was like, okay, that's what you say for marketing. Great. And not having that constraint, that willingness to say no and focus and say, let's try to do more with 18 and see how that goes then. You know, to do maybe less or even the same with 30 or 40, which you could have easily done by now. I think that's the, that's going to help a lot. That's how you maintain kind of the culture. You don't let it go from what I saw in optimizing, which is like a high psychological safety, high accountability to a, you know, um, safe, but, you know, but, but kind of, you know, like I would say low accountability culture, that's. Part of it is it's growing slowly. It's, it's being very purposeful in the recruiting process. I've, I've very, very much believe in references. Now, I didn't do a good job of that in my first company leaning into references. Now, like that is almost the most important part of the recruiting process is, is doing really good references, backchannel references, um, and, and being willing, and I've done this, being willing to say no to a phenomenal candidate who checks. All the boxes on. Can they do the job? Are they a great engineer? Are they [00:51:00] great? You know, but they, they fail on the cultural aspects and they fail in the references. Um, you know, we're willing to say no to great people because we think that it's more important to build the right DNA and the culture up front than it is to have a amazing tactician for the next six months or a year. But ultimately will be kind of a cancer to the culture.

Michael Koenig: Do you have a favorite reference question?

Dan Siroker: I do. My favorite reference question is usually I do 15 minute references and then most of it is just build a report to get them to be honest. So that when I ask this question, they actually give me an answer. So the question is near the end, I'll ask, Hey Michael, look, we're going to hire, we're probably going to hire Bob. Um, but you know, uh, if we hire Bob and in six months you and I bump into each other and I tell you it did not work out with Bob, what is your first instinct as to why? And usually with that lens, with that accountability of, look, I'm going to bump into you and you're going to tell me. And people like to predict the future. That's where I'll get the truth. Then I'll get something like, well, you know, what would probably happen? Bob probably got disengaged. You know, Bob probably got excited for the first few weeks. Uh, and then maybe Bob decided he wants to go start a [00:52:00] startup or everything. Okay. This person's maybe not committed. Maybe they're not focused. You know, you hear the things, uh, that otherwise don't come out when you ask for strengths and weaknesses. And anyway, that's been so far, my most successful question.

Michael Koenig: Such a great question. Where'd that come from? Is this

Dan Siroker: I don't know. I'm sure I stole it from somebody. Uh, I mean, for a long time when I did references, uh, you wouldn't be surprised given my experience optimizing, I would AB test my questions and I literally would just, every time I would do a reference, of course you did. And this question just took, I started to ask it. I probably heard it somewhere and it just started to get things that had I not asked that question, I would have not had information. Uh, it's really hard at the back channel reference to somebody you don't know to get them to be honest with you in 15 minutes is really, really hard. Like the stakes are pretty high for them. The return for them is not very low. So this question for whatever reason strikes the right balance of like, Hey, you're going to bump into me. Okay. I don't want to look like an idiot. And I'll tell you the one thing you really should know about this person.

Michael Koenig: Yeah. Important distinction there. That's a back channel reference.

Dan Siroker: Big believer in the back channel. I ask the same question of regular references as well. And usually the other trick I [00:53:00] do there is I never ask the candidate to introduce me to a reference. Instead, I'll ask the candidate, give me a list of five references, title, their email, a little blurb, how you know them. And then I'll reach out. Even that subtle difference alone, you get a lot more honest. If, if there's a direct one on one introduction, oftentimes the reference won't be as candid. And then usually I'll ask at the end of those references, um, by the way, this is something my chief of staff now has taken on and she's on fantastic. And she's now does these references, uh, for me and can do them just as well as I can, if not better. But usually at the end of those. Five references we asked for. We asked them who's somebody else who might give us a unique perspective on this candidate. Uh, and then you get to that, that first order. And then usually the thing is even if that first reference who knew that they were given doesn't want to say anything bad, what they might do is give you the person who's going to give you the real dirt, you know, because they won't feel as bad about that. They're like, Oh, you know who you should talk to. I think Dylan might have a really unique perspective on Bob, like go talk to Dylan. Uh, and then we reached out to Dylan. Dylan's like, Oh, do not hire Bob here. The fight, like, you know, that's, that's when you also get some truth because people want to be honest. They don't want you to be saddled with. You know, a bad hire. So, uh, it's just [00:54:00] sometimes it's not in their best interest. You see

Michael Koenig: me smiling because that's my last question that I love. Not necessarily the last, but it's in there. I actually had someone ping me and he's Hey. You're talking to 25 people. What's going on here? You're reaching out, like just make sure that, someone doesn't find out that I'm looking. I'm like, I know. But yeah, there's so much value in doing those. You, and I'm mindful of our time here. I am. This is just a great conversation. You mentioned getting people to go at ease right in the beginning so that you can ask that question and get a true answer. So building that rapport what's your approach?

Dan Siroker: Um, usually I'll start with, um, letting, well, I, I also, the approach generally is I want them to leave this experience very positive because often maybe we might be recruiting them down the line. So part of it is me selling them on our idea or vision. Um, I don't miss the opportunity to tell our story, explain why we're doing how it was inspired of my, my losing my hearing and the dire to create superpowers. Um, so I don't make it as [00:55:00] transactional. I make it more relationship, even though it's a short meeting. I also don't want to make it too long because I don't, you know, they're doing them a favor. So, um, so trying to strike the right balance. And, um, yeah, generally it's around respect, treating them in a way that I would want to be treated if I were asked as a reference, uh, letting them know how much we appreciate it, how meaningful this is to the, to the decision, uh, how much we value them as an expert on this person, like, you know, making them feel as important as they are, because they are, I truly believe, like, I have been asked to do references for people and I can tell the person asking me and interviewing me on the candidate, Has already made up their mind, or they don't care what I have to say. They're just checking a box. And when they do that, I'm so much less likely to want to give them the actual honest answer. Cause I'm like, look, why would I take a risk and tell you something that might piss off this person? Um, because it's true and probably not something they want other people to know, but if you're already going to make up your mind, you're going to hire them. Then like, what's the point? So that's another part. It's like, I try to treat the reference or the person I'm talking to you the same way I would want to be treated if I'm being asked to do

Michael Koenig: a reference. We've gone way over [00:56:00] time. I had so many questions that I wanted to talk about that we didn't get to. It's unfortunate. Maybe we have time for my final and favorite, which is, as execs we have had those times where something comes up and it's just never thought I'd see that. That is absolutely crazy. Do you have one that you can share with us? The

Dan Siroker: best one, the first instinct as to the question, uh, is related to something I said earlier, which is, it's going to sound so dumb. This AI thing is pretty freaking good. This whole AI, like, the first time I saw a summarization, like, just this morning, we had an executive team meeting, 90 minute meeting, went a little long. Uh, I was running rewind on it and we summarize it into five bullet points and it was so freaking good. It reminded me of my first job at Google as an associate product manager. My job is like to take notes straight out of college, taking notes for the senior executives, just, you know, typing away. And I couldn't have done as good as the AI did in summarizing the content, the substance, the, the objectiveness. [00:57:00] So, uh, I am, if you have not had these moments in the last. Six months since Jackie PD came out where you're just not blown away by what AI can do that. Probably a human couldn't even do Then you're not looking hard enough because these things are just so magical So i'll just leave it there to say like if you have you not had that moment You're in you get your head out of the sand and and try some of these products because the things they can do today are just truly mind blowing. They're changing the world. I

Michael Koenig: have that moment, but another moment that I think of that is like the moment, was when I first held an iPhone in 2007. And I was outside in Colorado in an open air market. Before that I had this LG flip phone which I actually have right here. Crazy enough, my daughter was playing with it. The screen is broken. But I held it and I went holy crap, I have Like humanity's knowledge in the palm of my hand. So it was remarkable. I had that same thing. Dan is awesome. Thank you so much. I've so enjoyed this conversation. We might have to have another and rewind for that. [00:58:00] See, I was going to do that. I wanted to use the word rewind to recollect something multiple times. And I held it off until now. Anyways Dan, where can people go to keep up with you and

Dan Siroker: learn more about rewind

Michael Koenig: rewind. AI. Perfect. There you have it. Well, thank you very much for listening to Between Two COOs. I'm your host, Michael Koenig, and a very special thank you to Dan Soroka for joining us today. Tune in next time for our next COO chat on Between Two COOs, and be sure to subscribe on Apple Podcasts, Spotify, blah, blah, blah, blah, blah, blah, blah. You have the idea. Tune in to the next episode, and until then, so long.