We’re talking to serial founder Julian Green whose most recent startup, Headroom is using AI to hack human conversation—starting with meetings. Julian’s goal is to make meetings and virtual communication feel closer to talking in the real world. Come to hear a very smart founder talk about his experience in startups and at Google, stay for Jordan pitching sound effects for every time Darrell mentions Canada.
We’re talking to serial founder Julian Green whose most recent startup, Headroom is using AI to hack human conversation—starting with meetings. Julian’s goal is to make meetings and virtual communication feel closer to talking in the real world. Come to hear a very smart founder talk about his experience, stay for Jordan pitching sound effects for every time Darrell mentions Canada.
Take our listener survey and let us know a bit about yourself and what you think of FOUND.
Links from the episode:
Connect with us:
Darrell Etherington 0:00
Hi, and welcome to found. I'm your host, Darrell Etherington. And I'm here with the thumbs up to my face palm.
Jordan Crook 0:06
Jordan crook. I feel like it should be the other way around,
Darrell Etherington 0:09
though. You think I facepalm? More than that's probably
Jordan Crook 0:12
true. No, I think I face more.
Darrell Etherington 0:15
Oh, yeah, you're right. But you thumbs up at the end of every meeting?
Jordan Crook 0:19
Really should be the thumbs up and the facepalm. And I don't know what you are. You're just,
Darrell Etherington 0:23
yeah, I don't make any sudden movements. Yeah. It's my management strategy. Also my life strategy. Display emotion that way they can't Yeah, I can't get you the evolution of man. So we're on found obviously, this is TechCrunch is premier podcast. I feel like I haven't yet ribbed equity into it. So let's go ahead and do that better than equity podcast. And here we tell you the stories behind the startups. And today, we have a terrific guest, Julian Green, who is co founder of headroom, which is video conferencing tool. So real hot right now. I mean, I feel like it's been hot, I guess for the past few years and will be for the foreseeable future. But headroom is different in that they're employing AI in a really unique way in the space to basically take care of everything that is tedious, and kind of the minutiae, I guess of meetings, but the stuff that really adds value. So taking notes, trying to get a sense of like, where people are at what their engagement level is,
Jordan Crook 1:30
like, who's talking? Who's talking? There. All right now, like you're using, it would be like, Darrell, you are monopolizing.
Darrell Etherington 1:37
I do, I used to run these through otter afterwards for transcripts, and I would have percentage talk and it would be like guests, by a large margin. Me. And then you and I did yeah, I was I described that to like, this is a technical problem, because Jordans at the time so talkative? No, your connection was kind of weird, or like he had some mic issues or whatever. Right? So
Jordan Crook 2:00
but it's not like doesn't add up to what you know, me to be the quietest person on the call. Right. But
Darrell Etherington 2:05
now I want when I'll have to look at it. Because it could be I have to take a look gender inequity or something. I mean, I'm willing to admit this.
Jordan Crook 2:12
Yeah, file a claim. Please, please. Thank you really face.
Darrell Etherington 2:20
I think that actually is like a poor thing. We actually didn't talk to Julia about as much as I'd like to have done. But like, it's one thing that the platform can do is highlight that in a way that is maybe difficult to do in a virtual meeting environment, maybe easier to do in person, right. So we don't have
Jordan Crook 2:37
the data either, right? Like, that's actually happening, right? If there's a five person team with one woman, and she often feels like she's being talked over you, they talk over me, but like, point to it. Well, I don't have every meeting recorded. You know, like, What do you mean, I'm telling you what happens. So you actually have like data now? Yeah, that's one use case, but could be an impactful one.
Darrell Etherington 2:56
Yeah. And we talked about a lot of the use cases. And also you know why Julian wanted to do this in the first place. And a lot of it came from his experience that big tech companies couldn't Google. But yeah, no one can explain it better than Julian. So let's go ahead and get right into the episode.
Hi, Julian, thanks for joining us. Hey, Darrell. How's it going? Great. It's going great. We're doing this meeting on Zoom. But maybe you should tell me whether we should be doing it on headroom, which is the company that you co founder?
Julian Green 3:33
Well, I'm glad you asked Darrell, it is the future. I'd love to show you to the light.
Darrell Etherington 3:38
Great, we're eager to be convinced I would love for you to give our listeners an explanation of what headroom is just a very high level overview. And then maybe a bit of background about yourself as well.
Julian Green 3:48
Sure. Headroom is really about fewer better meetings. And I think we've all sat in meetings wishing we were somewhere else. Yes. We want to chart and
Darrell Etherington 3:59
kind of test. She'd seen me glaze over. Yeah.
Julian Green 4:03
And it's strange, because meetings should be fun. Right? We're meeting with interesting people. Sometimes some meetings. So not,
Darrell Etherington 4:12
ideally, I think so. I agree. I agree. Continue to
Jordan Crook 4:17
be plot goal, I guess. But okay.
Julian Green 4:20
All right. Let me let me let me say some meetings should be fun. But there's lots of different type of meetings. But you know, some of the best moments you have meeting with cool people about cool stuff and learning and having, you know, maybe even creative conversations. I think meetings can be great, but most often they are not. And that is a problem. Because we all spend all of our lives in meetings, particularly now. We're stuck in our homes and the way to reach out to people is to have a meeting with and so you know, you have these days filled with meetings that don't allow you to do any other kind of work. And yet you don't have a lot to show for it right at the end of the day. I'm not the best note taker in the world. And so you know, occasionally I have some chicken scratch to show you know that I did do some thing with my day at work. But it's this funny conundrum. So what Henry was trying to do is two things. We're trying to make meetings themselves better in the meeting, and try and allow people to communicate in a more human way and take all the work out of the meetings. And then after the meeting, we're trying to make that meeting information useful. So there's, that's really our approach. And I can talk about how we're throwing some technology at that problem.
Darrell Etherington 5:27
Well, yeah, I mean, those are all admirable goals. And I think ones that basically anybody who has to do meetings would recognize as useful, right, like, based on their own personal experience, I'm the same as you like, I'm not really a note taker. And I also hate the thing of like, it feels bad to designate a stand out or something, but like, somebody should, because I'm not.
Julian Green 5:47
So one of the ugly truths is that there are very few people who actually take good notes. And unfortunately, you know, the people who are best sort of get tossed with it. And that's all they can do, they can't really participate. They can't be in the meetings, because when you're, you know, when you're doing this, people talk at 120 words a minute, and the number of people who can type 120 words a minute is vanishingly you know, maybe 60, maybe 80 Unless you're a professional, you know, stone agar fat.
Darrell Etherington 6:13
And then it's rare to have one of those on your meeting. And there's not one in each
Julian Green 6:17
exactly the number of professional stenographers is not broadly distributed. So, you know, the question is, do you want to take verbatim transcripts, and that can be useful. Often you want to take the most interesting bits of the meeting and sort of note that down, you know, what actions you need to take and what decisions have been made. And so your observations, but humans are dreadful at multitasking. So if I'm trying to communicate with you, if I'm also trying to communicate with myself through through transcript on notes, I'm essentially not communicating with you.
Darrell Etherington 6:47
It reminds me though, I mean, I think it's like every single meeting that we're in Jordan, like, Slack is the obvious example, right? Like people are in Slack. And you're like, Oh, this is great. I can do both these things at once. But then it's like, what were your thoughts on that Jordan? And it's like, hi, yeah, uh huh. Uh huh. Yes, I agree. You clearly have no idea about it to everybody thinks they're greater. That's the weird thing about it.
Jordan Crook 7:13
I always default to yes. And I, like, you know, I'm not like a decision maker at TechCrunch. But I do make decisions. And when I'm caught fading, yes, is the answer I land on. And hopefully, it wasn't a terrible idea. It's like, Oh, that's great. Let's do it. You know,
Julian Green 7:30
I mean, we shouldn't feel bad about multitasking, because most of the time, we're not super engaged in meetings, you know, people sort of feel bad about it. I think the solution is to make the meetings better, and make it you really wanting to be there rather than feeling bad about doing something else when you're in a meeting.
Darrell Etherington 7:45
So I did, I wanted to ask about how did you personally get interested in this problem? You've obviously participated in your fair share of meetings over your career? I think Google is I mean, every tech company is known for having lots and lots of meetings. But was it a pain point that you just saw, and you were like, well, my particular skill set applies to this, or what was your kind of realization of like, this is a problem and needs fixing, and I'm the one to do it,
Julian Green 8:11
it was a convergence of two things that brought me to want to do headroom. One was, as you said, sitting in, you know, 25 years of meetings in Silicon Valley, and realizing that the smartest people with the greatest resources on the funnest problems still have dreadful and unmatched also with sort of a broader, more intellectual, you know, observation that the politics in which you know, a generation seems to have served up a stunning lack of leadership and people can't have constructive conversations. And unless humanity can start with for people knowing what they spoke about, and what they decided, what hope is there for what hope is there for humanity.
Darrell Etherington 8:52
Yeah, that's a macro point that I did not consider. But it makes a lot of sense. Like, because if you're broken at that fundamental level, then how can you how can things work at a mass conversation or communication level? Right? Like it doesn't stand to reason.
Julian Green 9:05
So we're saving humanity when meeting at a time?
Darrell Etherington 9:07
Wow. I mean, it's like is that literally what you use when you're going out? And recruiting? Because it's a competitive environment, right? Like is that part of the way that you get engineers and people excited about joining a company is like, no look like this is a bigger problem than just like, what's your day look like on a on a business basis?
Julian Green 9:24
That's part of it. I think meetings are a little unsexy. And so fixing meetings doesn't this isn't necessarily the rallying cry. And I think thinking broader than that, of how it were really understanding human conversation. And so we're using technology to truly be able to summarize and give insights into human conversation. And I can talk more about that. At the same time. A lot of what motivates the best engineers to work on headroom is that we're using real time AI and some pretty cool ways on real time communication. So that's a lot of it.
Jordan Crook 9:56
So I want to talk about all this because like, I've been pitched a lot and essentially since the pandemic on, like everyone who's quote, fixing meetings, and so much of it is kind of similar to what you said early on, actually, oh, let's make meetings fun. So you have like, right, and there are a few others that are like, what if we could put you in some virtual environment, you know, you put a picture, and then you have a GIF, and you, you know, have a little sticky note, and blah, blah, blah. And it's like, it's almost like, you know, if artists who rarely have meetings got together and decided this is what meeting should look like, it was like, very random, you know. So it's interesting to go the like, I'm wearing a tie route. I'm sure that's not You're not wearing a tie right now. But like, comparatively to what I've been pitched, you feel very much like you're wearing a tie, like this is actually a real problem. And I'm going to go solve it with AI. How does real time natural language conversation processing happen across every industry? Or like, how did you decide to roll it out? Right, because what's important to us are how we make an assignment just industry based, right? It's like team based, like how Darrell and I, and a group of five other tech ventures come to a decision is different from the next media organization and the next media organization and what that sounds like, and the terms that we use, and the lingo we have, and the jargon like, seems so hard to scale that out and learn that and understand what is the highlight of this 40 minute meeting,
Darrell Etherington 11:19
right? The domain seems so broad is what I think that's
Jordan Crook 11:22
what I'm trying to say. Yeah. Many words.
Julian Green 11:24
So yeah, I mean, every meeting is different, right? That the interesting thing is the experts on what's being discussed in that meeting, or in the meeting. And that's the crazy thing, you know, sort of the domain experts in there. And the insight is that they actually label the data in real time, by their reactions in the meeting. And so if you use what we were using, which is use multimodal AI, use the different signals. So one, you know, one fun example is, you know, if you see two people crying, you know, they are at a wedding or their funeral, you need more context. And if someone says, yes, I'd really like to work the weekend, are they for real, or so you need to look at the transcribe language. And you need to understand that and there's been a lot of innovation with GPT, and transformers and understanding general language, you need to be able to look at people in meetings and see if they're excited, if they're engaged, if they're less, you know, engaged. This is a, this is a visual joke for an audio gesturing to each other, if someone's looking out the window, or you know, has left the frame, or you know, it's playing on their phone, and so you know, the language and the vision, and then the audio and their intonation that picture their voice, and on the video, you know, their eye gaze and their body language. So you pull these together, and you start to have an understanding of human conversations. And what we've done is we take all the best technology to do that. And then we've labeled a bunch of meetings for what are the most interesting moments. So if you take notes, the other thing is, you know, in the meeting in headroom, you're taking notes the same time. And you can do it in a couple of ways, you can click on some transcribed text, and just make that a note, like highlight sort of Daryl Jordan, you know, delivered a beautiful paragraph or pearl of wisdom, I can just click on that and make that a note. Or I can, you know, make those myself. And so every time you do that, you're essentially voting that that moment in the meeting is interesting. And so all these different signals go together and allow you to in the most easy way to understand take a one hour meeting, and choose the five minutes of the most interesting moments of that meeting. You know, it's like a sports center highlight
Jordan Crook 13:25
reels, NFL redzone, for meetings. Yeah,
Darrell Etherington 13:29
that's exactly what I want. Because like, immediately afterwards, you're like, I don't even remember what just happened, let alone highlights, like, I don't remember the thing at all. So there's no chance that I'm going to pull the highlights out of it
Jordan Crook 13:38
encouraging. Someone has in so many meetings
Darrell Etherington 13:42
Jharna, don't tell my boss, you
Julian Green 13:45
the forgetting curve on meetings is insane. So we've done some, some research on this. And you know, even just two days after meeting people basically remember nothing. So if you go back and say, you know, what was the company that Darrell mentioned, or what was who was the person who I should talk to about that. And so what we built is sort of a Google for meeting information is where you can go back and you know, you have total recall, if it was said in the meeting, you can find it, you can go to that moment in the video. And there's a lot of solutions out there. And you know, you talked about everyone saying that fixing meetings, we feel we're obviously biased, we feel that we've taken a very complete approach to this. And so there are some people who you know, you said is they're doing cool things with video filters, so you can sort of make it more fun to look at things on video conferencing. There are some people who didn't transcription right there are some people who are doing you know, gesture recognition and meetings to try and add some nonverbal communication in there are people who are taking the video of your meeting and transcribing it and making it you know, editable and shareable. We're doing all of that.
Darrell Etherington 14:44
That is to me the thing because we use otter previously because like it's a reporter's best friend in a lot of ways right? And then they introduced their integration with Zoom and I think we tried that a bit for meetings, but it is isolation. You're right like all these elements are like either too complex for people to just all bring together and remember, turn on this thing or whatever. It's a lot, especially cross organization or between organizations to keep track of. And then it's also just like insufficient, like, it's not giving you like, Okay, well, it's giving me the whole thing. But really, that's just, it makes it marginally more searchable than say, like video, but like, it's still not telling me what was the most important thing? Or what did I actually want to remember?
Jordan Crook 15:22
So we're getting to the point in the podcast, I think we're gonna need a sound effect for it, where I with 10 minutes worth of information, start brainstorming on something that oh, this
Darrell Etherington 15:33
is what you do your products. suggestion? Yeah.
Julian Green 15:37
Let's do it. Do it.
Jordan Crook 15:38
So like, proactive information delivery, right? Like how many meetings have action items at the end? Like they're all by Tuesday? Can you get this and I'll do this by Thursday, and we'll meet back again on Monday. That's something that you guys could do, right? And be like, hey, Darryl, Tuesday's tomorrow, and you said you would do this, and it's here and a transcript and based on evidence,
Julian Green 15:58
exactly right. That is not yet in the product. That is absolutely something that we're hoping to add. And if you think about some of the previous attempts, it's been more like Alexa or Siri, you've been invoking a an assistant, remind me to? Yeah, well, that was either right and voice voicea. So it was like, hey, either enterprise voice assistant, I have an action item, then the assistant could help you take it. But the language technology is now so good that if you say we should do this, I'd like to meet you next week. I have an action item, you don't need to invoke an assistant. And I think probably most useful is to suggest an action item rather than saying, you know, we're going to get it right every time. And we absolutely know, you know, we want to say hey, you know, was this an action item? And then you just click it and say yes. And so that's the plan. Yeah.
Darrell Etherington 16:43
And then yeah, your tool will get better over time. Because you're like what action items look like. This reminds me so much of like, this is an anecdote from when I just started out in my career, I was a desktop publisher, actually, for like a very small consulting firm. This is a long time ago, but the managing director of that once called me in and he was like, do your smart guy or smart guy, what I need you to do is program an Excel sheet so that it can take recordings of our meetings, and then to a transcription of that and then output the point of the meeting or the point of the person's comment in like another cell. And I was like,
Jordan Crook 17:21
how do you expect me to do this?
Darrell Etherington 17:23
I went to a sheet. He was the type of guy he was the type of manager where like, nobody said no to him directly, like people just said, okay, so I was like, I can't do that. I can't do that I was new. So I'm not savvy enough to be like diplomatic, I'm like, that's impossible. What are you talking about? And then his assistant was like, Oh, we'll get it done. And then like, took me aside was like, Look, you're just going to email me a summary with a description. And then I'm going to put it into an excel sheet. And we're going to tell him that the Excel is made up to my point is like, at the time, it was totally absurd to imagine that anything could do that. I was like, I think that there are people working on their PhDs and stuff like this now, but it's like not a thing that anybody can do. Especially not in like consumer grade technology.
Julian Green 18:14
So I think it's everything. And I think what when I was using machine learning at Google for the computer vision and the AR we're working on, and then at Google X to solve all these crazy deep tech science problems, we were trying to get machine learning to help us solve, what I realized is that the computer vision and the natural language technology is now right at the right moment to try and solve human conversation. And meetings are a great place to start because they're valuable interactions. There's enough value there for people to pay you to solve the problem. They're relatively structured, hopefully, you're not just shooting the breeze
Jordan Crook 18:50
and literally are like, I'm telling you the agenda for the works. Yeah, I mean,
Julian Green 18:59
but I mean, that can be fun. And that's, that's also a goal of getting together. But yeah, there are enough meetings where really stuff needs to happen and structured and you know, you know, who's in them, and there's hopefully an agenda. And so really, it's a solvable problem. And that's why we're excited about the fact that there's all this data for the first time meetings have audio and video being captured. There's this technology which is now at the right time to be able to help and if the new crazy reality of remote and hybrid working is going to stick around we need something better to make it with Yeah,
Darrell Etherington 19:33
and I mentioned it is because like some people argue like Will there be a reaction that is like aggressive the other way or like people are just okay, I don't want to have anything to do with remote meetings anymore at all because it was so long, but like, Jordan, I need a sound effect for the time in the podcast where I reference when I worked at Shopify, but
Unknown Speaker 19:54
yeah
Darrell Etherington 19:58
I did that like we were It was the pandemic was not a twinkle in anyone's eye, right, it didn't exist. But we did the hybrid all the time, all the time, every meeting was hybrid, you just you went into a conference room, there was a huge video screen and a bunch of people from Montreal or Ottawa or whatever joined via the video screen, right. So I have to imagine that it is now just entrenched and the norm and de facto not going anywhere. I think
Julian Green 20:23
the beautiful thing about the whole future work debate is nobody has any idea. But you guessed that there's going to be more options, and people are going to be doing more different things. Not everyone is going to be nine to five with a tie in an office every day, it's going to be much different. And we've seen as a remote startup, the power of being able to hire people, not just in Silicon Valley, but around the world is crazy helpful,
Jordan Crook 20:50
I want to go back to the product. So there are certain categories of enterprise or SAS tech that are very clear on the bottom line of a company, right? When they sign on to use the software, like, okay, I get it, even if it's not directly related to like my product cycle, right, or my customers or whatever, I can see how this saves me money, or how this makes me money, one of the two, whereas this feels like it's in a separate category where there's a lot more that's qualitative. It's like, Hey, I know Meetings Suck. Like I know, I don't quite remember what happened. And I know that things get lost in translation, and blah, blah, but like, what does a good meeting actually look like? And what does like success with this new tool actually look like? And how can I measure it and make sure that the money I'm spending on this is worth it? So how do you do that part? Right? When you're talking to customers? And like actually selling the value of the product? Do you have numbers you can attach? And will it work for us? Like the first thing I think about with TechCrunch is like we're so bad at meetings, I think of Darryl, in almost every meeting be like this isn't what we were talking about five minutes ago, like where how did we get here, like he was frustrated.
Darrell Etherington 21:57
My character, my meetings character is the guy who towards the end says Remember, the purpose of the meeting was excellent.
Jordan Crook 22:04
We had like five minutes already upset, like grumpy old man.
Julian Green 22:08
So there are a ton of very tangible hot benefits. And the folks that are most motivated, who are in meetings, where if you make the meeting better cash comes out the other side, you know, those folks are already using transcripts, they're already looking at recordings, and they range from salespeople who, if they remember to sell in the right way, and to listen more than they speak and all the other best practices end up selling way more. And there are good statistics around that what we've done there is you know, something like gong or chorus allows you to look after the meeting, how much did I speak that I mentioned that a key selling propositions, we do that in real time. So you can actually look and see, oh, I'm speaking too much I should shut up. And then other types of meetings, like interviews, where instead of taking three weeks for a bunch of people to ask the same questions of the person, you can use asynchronous meetings to share the first interview and then be much more targeted around what you ask people UX interviews, you know, you guys in media, where the meeting itself is sort of the product, having good recordings of that and knowing what information was in there health. So there's tons of meetings where there are super hard benefits to better meetings. And then as you say, there's also the fact that most people are gathering their eyes out in meetings most of the time. And that seems like a bad way to spend your time.
Darrell Etherington 23:22
Yeah, the felt pain seems like it would go a long way. Right? Like it's even if it's not something you can easily quantify in like, you know, bottom line numbers, every single person has experienced it. And it's one of those things that it's hard your brain is just like, I know, this is a problem. And the numbers
Julian Green 23:34
that come out of our pilot users is that they say one or two ads a day, because they don't go to a bunch of meetings, like big meetings, where you're not presenting where you're just listening, like webinars and the types of meetings where you where there's already a recording, and you can listen data, we basically turn everything into a webinar. So if you want to listen later, you can and you can listen to the summary and do it in five minutes, rather than an hour. And you can search it and go, Oh, the five minutes I need to look at is the five minutes at the end.
Darrell Etherington 24:03
One of my meeting Gospels is like if a meeting over a certain size is not a meeting intranet exists, right? Like just doesn't make sense and more of a speech. Yeah, it is a speech and async video stuff is really great for that. Right. And to your point, it's like make it into a webinar and then have a q&a feature after at some point that's participatory, but the rest does not need to be st.
Julian Green 24:22
So another fun stat. I don't know if you ever saw the information about judges and when they give verdicts in their day, you know, so after lunch, they're way more likely to give a guilty verdict. Oh, wow. If you've seen that research, fine, which we get don't go to court.
Jordan Crook 24:40
I say from a holding cell, it's close lunch.
Julian Green 24:43
So what because we're looking at the engagement of each person in a meeting and instead of using that to call people out and go Joe's not paying attention, or Julian's goofing off, what we do is we give that back to people as an average, you know, sort of temperature of the room. So you can say oh, boring people they were all super interested in minute ago I've gone on to. And then after the meeting that helps us do the summary. But we look at those statistics. And what we've discovered is for headroom, ourselves, meetings after lunch are basically a waste of time. And the engagement is super low. And then after about half an hour, in any meeting with more than about five people, the engagement drops off significantly at like the 3035 minute mark. And so there are these insights where you hopefully can work out how to meet better.
Darrell Etherington 25:27
Yeah, I was gonna ask like, perhaps even more generated, right? So you actually, you're gathering a lot of data, and you're gathering it in an area that a lot of places probably haven't previously had any data or done any kind of like serious scrutiny, right? So when we're talking about companies moving to online models, or remote models, or hybrid models, it would seem like there's tons of opportunities to provide them a big piece of the puzzle, analytic side that like, this is how your business is working now. And is that something you've explored? Or is that something that you've talked about with customers, a lot of bigger
Julian Green 25:59
companies have wondered, you know, how can I instrument my business and run it top down more efficiently, we have a slightly different philosophical approach, which is the we don't want this to be big brother. And we think that if you can enable individuals and small teams to have better meetings, then they'll be able to work out at the most efficient themselves. And obviously, all this information is the most private, confidential information. And so it only goes to the people who were in the meeting, they choose whether to share it more broadly. So we don't really, you know, have stats that we give on about like
Jordan Crook 26:31
broad stroke stats, though, right? Because if you have 100 clients, it's not even just like, This is how my business works. But maybe it's true what you said of headroom across the board, right, like maybe 99% of meetings after the 30 minute mark, if there are more than five people engagement significantly drops off, have you thought about kind of educational tools?
Julian Green 26:51
I think in the aggregate that sort of general best practices and meetings, data will be hugely useful. And it's probably going to sound a lot like common sense, right? Don't have meetings that are too long. Don't have meetings that are too big.
Darrell Etherington 27:03
Yeah, all the things that we share, and nobody listens, but at least now you'll have stuff to back it up.
Julian Green 27:08
Yeah, be good to have the data and know, you know, is it 35 minutes? Or is it 40 minutes,
Jordan Crook 27:12
there's conflicting best practices, right? Like, you're not supposed to have a meeting that's more than 30 minutes. But you're also not supposed to, according to like, like anything on EQ tells you like, don't just jump straight to work on a meeting either. Be like, Oh, what are you doing? Your
dog was cute, like, the weather? Right? Like I care about.
Darrell Etherington 27:30
You said these things to me and they sound sincere when you say them to tell
Julian Green 27:35
that to you, Sean, I thought you can
Jordan Crook 27:39
care about people I work with. But I think there are a lot of people who don't give a shit about the people they work with. But know that the best way to get the best out of them is to be like I really care about you. I think
Julian Green 27:47
the data will show that treating people like humans and icebreakers and connecting with them before you jump into the you know, what have you done for me lately? I think that will show through. As I want to hear you saying that.
Darrell Etherington 28:01
Yes, I like to hear you say that. But I also like hearing you say that your philosophy is like ground up kind of like because I think that's true of we've talked to other founders about remote and remote first approaches and transitioning to remote and whatever. And like TechCrunch has always been remote organization. That's one of our strengths, right. And like, it's a thing that I think that legacy organizations don't really understand is if you start from that place, it's like trust your people trust them to know what to do with their time and how to allocate it, and how to get the work done. And they will get the work done. Whereas if you're coming from a legacy, and you're trying to transition, I think the tendency, and that's why I mentioned that is to say like I need to understand what people are doing, because I no longer have direct line of sight over that.
Julian Green 28:44
Right. So transparency is huge. And when you can't see people in the office, you wonder if they're doing anything. And that's just a normal human reaction. So when you can share meetings in a way that just doesn't say, Hey, is too much information for you to look at to show that I'm doing stuff when you can share it in a way which is, you know, look, here's a five minute summary of a really cool meeting I had on this topic and people can watch that I've actually increased the you know, the respect I have for what my colleagues are doing, because I can now see into it a little
Jordan Crook 29:13
bit. I think also though, there's like another piece with the legacy players, which is that not only is their instinct to know what the hell's going on, so they're like clinging to that. But I think there's also if you're an employee of a place like that, and historically been in a cubicle with someone knocking on the door every 10 minutes being like, what are you doing what's going on? What's going on, you're likely to when the world changes and you go back home probably be less good at allocating your own time and maybe even like feel the need to abuse it like freedom, you know, like the first time as opposed to if you start from a place of my employer, trust me and as long as I get the job done, I'm good to go. You know, it's like a overreaction on both sides potentially.
Julian Green 29:54
I agree. You know, when you spoon feed people, they get used to it. What's been amazing with remote in our company. Everyone's like, oh, it's sort of up to me, no one's checking up on me. And so I need to, I need to get it done. And then being able to share some excitement, because it's quite lonely, you know, being remote. And so you need all those human connections. And when someone can share a bit of what they've been doing through a meeting and share that, it's, it's fun, it makes it more human. Yeah,
Darrell Etherington 30:24
for sure. Yeah, that is the biggest challenge. I mean, you know, we try to do virtual get togethers once in a while that are like just, you know, nothing like, Let's just hang out that
Jordan Crook 30:33
it's like not even close to a parallel for what happens when we're all in the same room in person.
Julian Green 30:40
Let me talk just for a second about how to make meetings a bit more like being in person. And this is a challenge for remote. And so one of the big challenges we talked about is that at least some of the people in the meeting will be typing furiously, and taking furiously and not able to really participate. And so we try and take away those distractions, with technology with the transcription and the notes and the fact that you know that there's a meeting being saved for you to be able to look at in a convenient way. So you don't have to worry about capturing everything. But some of the other stuff is, you know, just things like gesture recognition, right? So one of the worst things about Videoconferencing is that there's a Stanford study on Zoom fatigue, is that it's very hard to know when to talk, you can only have one person on the stage in the limelight at any one time. And if you want to get on that stage, you have to elbow them out of the way, right. So you get a lot of this. After you know, you go and then basically what happens is because people have these social collisions, and don't know when to speak, they basically shut up and they withdraw, and then only one or two people in a big group can actually contribute. And then you're sort of like, why are we here as a big group anyway, so one of one of the very simple little things is if you can say, Hey, Dad, I love what you're saying, and I do a thumbs up. And you can only look at one person at once. So if you're looking at me and Jordan do a thumbs up, you wouldn't know. So we just sort of augment that and we send the thumbs up out in real time, and a few folks are trying this. It's actually identical problem. And if you do false positives, like there are some tools where they try and do the hand up and hands up a guy all the time. I didn't put my hand out. Yeah,
Darrell Etherington 32:09
cuz a gesture, you're like,
Julian Green 32:11
I'm Italian. I use body language. So you know, everyone's putting their hands out apparently all the time. And they said did Julian did you put your hands I know. So getting the false positives dialed down to the point where you can actually recognize the gesture and only recognizes it when you're actually doing it when you're just waving your hands around. It's actually a hot technical problem. But the benefit is hilarious. So when people use Hedra, and they go back to the tools, they go back to zoom, and they're doing gestures and zooming like, why is this not? Why is this not working?
Darrell Etherington 32:41
It's, that's a really good problem to have if something becomes the behavior that you assumed to be default. That's terrific. That's
Julian Green 32:49
right. You know, whenever you have to change what you do, it's hard to have to remember to say it, okay, Google,
Darrell Etherington 32:55
hello, Google are
Jordan Crook 32:58
up at the end of almost every meeting just like this. And it's my way of saying goodbye. I don't wonder how to be like Jordan really liked that the meeting ended.
Darrell Etherington 33:10
You need to do one for vaping. And then it's like Jordan, Jordan.
Julian Green 33:16
You said she would speak right now. But she's taking a hit. The fun Easter egg is the is the facepalm. So you know, when things are going badly, and headroom, you can constantly.
Darrell Etherington 33:31
But that'll be great, because then afterwards, like, eventually, you could generate, or maybe you're gonna do like a per user sentiment analysis. And it's like, you know, Jordan overall, is
Jordan Crook 33:42
disappointed, and only happy when the meeting
Julian Green 33:44
ends. So I mean, seriously, we have that data, and we're working out how to make it most useful. But if you for example, looked at your week of meetings, and realized every time you met with Darryl, you had a great time and every time you met with Julian you facepalming 90% of the time, you might think about who you
Darrell Etherington 34:01
want to meet with or like how to improve the underlying relationship
Jordan Crook 34:11
surprised me the other day that my dad brought this up, who is obviously much older and works for like a big, huge legacy business. And one of the things that I think is so lost that I don't know that technology can fix is like the side conversations like the buzz, right? My dad was talking about, like his weekly team meetings in person and how there were always people like buzzing all the time. There's always like, smaller. We could do the you know, like that, like little side conversation. Sometimes it's for fun. Sometimes it's Darryl and I rolling our eyes at the same silly comment. Right? And it's just us like building on our friendship zone. I was like, actually a step of the meeting, right to like, get a little cohort. Okay, like, let's talk about this part now. Right. I know the
Darrell Etherington 34:55
indicator that an idea is very excited, right? Yeah, it's already starting to Spin,
Jordan Crook 35:00
can we I guess we like Slack, right? Like in the middle of a meeting. It's like, oh, what do you think about that? What he just said, like, I think it's kind of silly, or maybe I think it's great or whatever, but it doesn't feel the same, you know, like, it just doesn't. Again, it's not a one to one for what we do in the real world.
Julian Green 35:16
Yeah, the real world has some great advantages. This sort of in meetings, there's sort of this truism that the best meeting is who you happen to go up in the elevator with and talk to on the way to the meeting. And then when the meeting stops, and two people stay behind to say, did you really mean that or are you know, I wasn't sure what you meant at that, you know, that those moments can be the most valuable. I think that's sort of a real world versus virtual, you know, problem to solve. I do think hopefully, if you enable people to sort of be on any device and sort of meet, you know, from anywhere, then then maybe you can capture some of that with breakout rooms and side chats, as you mentioned, and sort of enable that. But it is it is a problem, because some of it is being right there with someone we believe some of that is the nonverbal communication, right? So the buzz, you talked about the bus, right? In a room, you can tell when people are leaning in, and are interested and engaged, you know, there's sort of moving in the city, you know, when you nod your head, they nod your head, you know, there's some mirroring going on, there's a bunch of stuff that happens when you get people who are able to perceive each other in a conversation when they're not staring down, you know, the tubes of the internet each other and they can only see this two dimensional representation of each other. And so with gestures and other capturing nonverbal communication, augmenting it with trying to give people the feeling of people leaning in, or people, you know, lying down and not interested. So there's some of that you can do, but the having the virtual elevator on the way to the virtual meeting, you know, we'll have to work on that.
Darrell Etherington 36:45
Well, that I mean, that kind of is the next question I have was because, you know, there's a lot of, especially Facebook, obviously, or meta now, but like, they're very interested in the virtual meeting rooms and virtual meetings. And I think they would argue that a lot of this stuff is solved by doing that. And by encountering people in those spaces, right, but like,
Jordan Crook 37:06
just getting sick in your kitchen as you meet what the team and they are.
Julian Green 37:11
Is this the part of the podcast where you play the metaverse
Darrell Etherington 37:14
sound? Yeah, yes. The Metaverse theme song.
Julian Green 37:17
So here's the deal. We think we are closer to the metaverse and anyone. Let me let me let me take that back. I think meetings are basically a 2d representation of the metadata, right? So you, you look into my world with the webcam, and then you can see my reactions, and I can see your reaction. So it's this two way screen and sensor that's two dimensional, and it doesn't have an amazing, you know, three dimensionality and some of the, you know, the haptics and some of the other things. But the question is whether it's most important that you have all the fancy headgear and you know all the stuff, or whether you've found a way to effectively communicate with each other. And so the augmentations that were doing have meetings with all the metadata of, you know, transcripts and notes and nonverbal gesture, augmentation, and so on, are sort of baby steps into being able to understand each other's realities.
Darrell Etherington 38:15
I mean, like, I agree with you for me, I asked that question just because I anticipate that question from others. I think the metaverse is awful. I hate it and shouldn't say that, like on the dock. That's for everyone. But I mean that that version of it, I guess the version of it of like little bobblehead, whatever guys floating around in a virtual boardroom is not something that appeals
Julian Green 38:34
it's a classic, you know, we'll throw technology at it. And it will solve everything sort of fallacy, right? Is you read it and Neil Stevenson, you're like, Oh, my God, that's cool. And then you see what meta did we have, you know, robots, Zuckerberg bubblehead. And you're like, Okay, these two the same thing?
Darrell Etherington 38:52
Yeah, like they're pretty far apart on the spectrum.
Julian Green 38:56
And so the question is, if you just keep throwing, you know, more smart engineers at it, does it get better? Or do you actually have to think about how do people want to communicate?
Jordan Crook 39:05
I do feel like I would be more engaged. If we could have our meetings in a setting of like the bridge on a spaceship. That'd be cool, though. Like, see, I'm falling for it again, Julian, do you see the artists who are fixing meetings there in my head,
Julian Green 39:18
we have a great have a spaceship bridge background and headroom. And that goes some way towards it. I think that's really exciting innovation going on with sort of virtual offices and meeting spaces and serendipitous, you know, bumping into each other and breaking down the very traditional scheduled meeting approach some of that inner were doing with the asynchronous you don't have to be in the meeting, you can consume it later and share it and move the information around so you don't have to be stuck in meetings all day. But I think something will come out of the new approaches to how and when you decide to meet where and so.
Darrell Etherington 39:54
Yeah, but I think your way in is probably the one that will lead to the most like It's like start there start where people already communicating and have the means that have proven effective, and then figure out how to expand that into what becomes eventually a virtual shared office.
Julian Green 40:09
And we think the disruption because it's obviously hard to go up against these big video conferencing companies, whether zoom or Microsoft, or Google, you know, they got lots of users, they have big teams working on their products. And so what we're doing is doing something different. It's not just video conferencing, it's communication and collaboration, right. So there's all this meeting information afterwards. And we sort of had to build everything from scratch. So
Jordan Crook 40:35
stocks should do that. He did that suck. I
Julian Green 40:40
mean, on the communication side, web RTC has made massive strides towards incorporating all the amazing innovations, zoom, and others have made on how to get pixels to fly around the world in the right way. So you can see and hear each other, right. So those, those are largely more solved problems, but the ability to do real time AI on pixels as they fly around the world. That bit is where we've made big advances. And it's actually quite hard for traditional sort of voice over IP picture phone, you know, sort of signal processing technologies to do that, for some architectural reasons. You know, if it's a peer to peer conversation, which is zooms to person meetings, there's no server in the middle, you know, there's no way to add any value to the pixels. If you do you know, end to end encryption, there's no way to add value to the pixels and possible. So you need this sort of media server in the middle, this SFU and low latency real time AI technology to be able to do it, and then build a proper security model around it so that it works for people's privacy and confidentiality. So you know, we can now basically run any neural net in real time on real time communication. And so that's exciting for all the different features that people are gonna want. Alright, but
Jordan Crook 41:51
to be clear, you didn't build from scratch video conferencing software,
Julian Green 41:54
that was the place where there was web RTC, we could leverage that, and lean on.
Jordan Crook 41:59
Okay, exactly. Yeah, that would be a sucky place to start. Right? Like,
Julian Green 42:03
that's a lot to do. Yeah, we don't, we're not massive. So yeah, all the API we've had to build and the ability to run it over the top of the web RTC data channels, but the Web RTC was something we could take a great open source implementation of, and it's been that community is amazing. So that's been good.
Darrell Etherington 42:21
Yeah. And the latter sounds like a great note, like the thing that you've actually been
Jordan Crook 42:24
on the product side, though, do you know when a meetings going off topic to
Julian Green 42:30
define off topic, Jordan?
Jordan Crook 42:32
Curious, because I'm sorry, but I'm stuck in the Space Bridge? And like, what if there was like an incoming missile and someone went off topic? That's a cool blend of the AI being like, are you gonna stay on track? And like,
Julian Green 42:46
I mean, we can definitely tell when people are engaged and excited about an incoming missile and less excited about whatever it was they were talking about that got interrupted by the mist. And that's, that's how the highlights work. Right? There are moments when nothing's being said, but people are going crazy. And that tends to end up in the highlights.
Darrell Etherington 43:05
I wish we were using header because yeah, you could just take the notes and the highlights, and they could be your product roadmap, and like all journeys,
Julian Green 43:12
I like it. I like the Excel spreadsheet with the
Darrell Etherington 43:17
I wanted to ask, we don't have too much time left. But I do want to get into a bit of kind of like your personal journey, because you were a founder, right? Like you found a jetpack. And that's how you got to Google. And then a good stint at Google and then back to being a founder again. So what was it? Why did you want to go back to the entrepreneurial life.
Julian Green 43:35
So this is my fifth startup, and I love startups, it is a ridiculous opportunity to try and do something very hard. Get the group of people all of whom you have to work with you and try and assemble a crack team and do a ridiculously hard problem. And I don't know too many opportunities to do that in the world. It's it's crazy fun. I was at Google after they bought my first startup jetpack for longer than I thought. Google is an amazing place. And they let me do some ridiculously fun stuff. I got to work Google X on a variety of moonshots trying to solve crazy problems. You know, the imposter syndrome was off the charts. I think my first day at Google X, I was asked to go and have lunch with a Nobel Prize winning physicist and see if his technology was any good. I said, How am I gonna? And so you know, lots of fun problems. But the reason I decided to dive back into startups was both this particular opportunity, they're coming together, oh, my goodness, meetings still terrible. The technology is there, the data is there. We need to do this. And then also just the attraction of going back to startups where you have a small team, when you need to make a decision. You basically ask yourself, do we want to do this? And the answer's yes. And that's the decision making process. I like that.
Darrell Etherington 44:51
Yeah. I mean, yes, that has a lot to recommend it, I think but do you miss the Nobel laureate sniff tests, or do you get to do any of those? I
Julian Green 44:57
mean, it was it was like goodbye. University I was amazing. I was learning so much stuff. And being part of projects that are attempting to save the world was incredibly inspirational
Darrell Etherington 45:08
as the change any kind of your like fundamental approach to found like, do you feel like you're you're very different in your approach as a founder than you were at jetpack or a house
Julian Green 45:17
or I was taught by Google X and the moonshots we were trying there to think bigger. And so it's as much work to try something big and hard as it is to do something small and less hard. And so you might as well reach for the stuff.
Darrell Etherington 45:31
Yeah, well, you definitely reach it. And I hope you get there. Because honestly, I want everything that is on offer from ad group. And more and more, we demand more of you get back to work. Well,
Julian Green 45:42
it's funny, because it's a problem that everyone is on video conferences all day every day, right. So doing something where everybody has an opinion, is both a great opportunity and a big challenge. And then doing something where if it doesn't work, you know it because you're on your product all day every day. Super chasing right? So yeah, right. So we feel like we have a great privilege and a great challenge.
Darrell Etherington 46:06
Thanks very much for joining us. That brings us the time, but it was great chatting with you. And I think, yeah, we figured out some some stuff we have to work on at TechCrunch. And maybe we should give headroom a try to solve that.
Jordan Crook 46:18
How many? How many doll hairs is it?
Julian Green 46:21
It's it's currently zero dollhouse?
Darrell Etherington 46:23
Oh, all right. That's what we like to hear. That's it awesome.
Julian Green 46:30
This is completely free. What we think we'll do eventually is charge for you know, storing meetings, but we want to try and launch we want to try and keep the the fully featured video conferencing bit free, because I think communication wants to be free.
Darrell Etherington 46:52
All right, Jordan. So that was our chat with Julian, someone who knows meetings better than most folks. But I mean, everybody knows meetings, we could all feel that pain when we started talking about it.
Jordan Crook 47:02
Yeah, I mean, I definitely feel like if we switch to headroom, I'm gonna get called out a little bit. All this gesture recognition. I feel like my teammates and stuff can see it. But I don't like the idea of being recorded forever somewhere
Darrell Etherington 47:14
about how like you're constantly giving everyone the finger like if it just says like,
Jordan Crook 47:19
that's exactly what I do. Thank you. Yeah, listeners, I'm giving them the finger. But, you know, I do think that like, the overarching data is interesting to like, it's cool that this could transform the way teams do meetings, because I think we all like you said, we all agree that there's pain there. So they're not as productive as they should be, etc, etc. But I think like on a holistic level, some of the stuff he was talking about where like meetings after lunch are basically useless. And meetings over 30 minutes, with more than five people engagement significantly drops off. And some of that stuff is just like useful tools to have. And if you can, as we think about like the future of work, when we think about startups, who are building companies and trying to, you know, I feel like you and I do spend a lot of time thinking about process, where it should be where it shouldn't be and what it should look like. And startups do that. You know, this seems like a useful framework to start with, right? Yeah, meetings are like the skeleton of a business a lot of ways. So if you can nail that you're starting with one foot forward, right? Yeah, for
Darrell Etherington 48:17
sure. And I think like one of the things we did also didn't talk too much about, but that people, depending on their industry will probably like be, you know, selling silently, their computers, but like sales teams use a lot of software that does a lot of these things already. When you're doing sales meetings, a lot of what you want to do is track what happened, take detailed notes, be aware of when things seem to pique someone's interest or convert them or whatever, right. So like, there's a lot of impetus to like want to do that there and a lot of motivation to have those kinds of tools there. But I don't think it's been something that a lot of other folks have focused on or been interested in, right. But now because we're in remote people definitely are. And yeah, they're looking for anything, anything that can help them kind of like form more useful practices around meetings, which like Jordan said, like, they're the backbone of the company. So obviously, you want to make changes there and improvements there. I think it's interesting that like, Julian was pretty adamant about, like, you know, everyone's information is private. And this is not something where we're sharing detailed stats at the organization level, or they're not interested in doing that right for now, for now, right. And that's the thing where it'll be the one to watch. Because as a company grows, and as a company, especially a startup, you know, like maybe their runway gets smaller, or they run out of options that side or like they're not seeing the growth they want in particular metric. And then those options obviously become more palatable. And then pressure from investor can put them on that path, even if they didn't intend to do it in the first place. So like, that's all this is a watch. I don't necessarily like I'm not making a moral judgment of that. I think it could be an interesting tool for an enterprise using the right way. And not necessarily just something that's like a nanny state type feature. But
Jordan Crook 49:56
I mean, on a positive note, the idea that he created It's something pretty general, like you were talking about this being used for sales staffs Right? Or like how software is probably been developed, particularly around selling a certain thing, right selling x you we can give you really good notes and details on this is like pretty general AI that is measuring things that could be used in a million different organizations across a million different teams. So that's a pretty, you know, phenomenal feat. And it'll be interesting to watch this company as it grows. Yeah, what they do,
Darrell Etherington 50:25
yeah, that because a lot of what we talked about was how the reinforcement learning works and like how the software is getting better as it's like building its database of, you know, what is a behavior that indicates a positive reaction on a virtual meeting or whatever, right. And I think that that is a net new kind of like database of information that to your point will have like, tremendous value. I think there's stuff that it will end up being valuable for that we don't just understand or have knowledge of yet because we're still in the very early stages of like, what is a remote company and like, what's the remote workforce working? Well look like?
Jordan Crook 50:59
Definitely. I agree. Well, we'll see AR You know, anything could happen.
Darrell Etherington 51:06
Found is hosted by myself, TechCrunch news editor Dale Etherington and TechCrunch Managing Editor Jordan crook. We are produced by unshackled Kearney and edited produced by Maggie Stamets TechCrunch his audio products are managed by Henry pick of it. You can find us on Apple podcasts, Spotify, or wherever get your podcast and on Twitter at twitter.com/found. You can also email us at found@techcrunch.com and you can call us at 510-936-1618 and leave us a voicemail. Also, we'd love if you could spare a few minutes to fill out our listener survey at bit. Ly slash found listener survey. Thanks for listening. We'll be back next week.
Transcribed by https://otter.ai