The Future of Work with Zoom CEO Eric Yuan and Reid Hoffman

0:00Hey, everybody. This is Ben Kesnoka, co-founder and partner at Village Global Global, a network driven venture firm. And this is our podcast, where we go deep on all things business and technology with world-leading experts.

0:21Good morning, good afternoon, good evening, villagers. My name is Ben Kazinoka. Great to have everyone here. And we're delighted to have two of our Luminary LPs, Reid Hoffman, Eric Yuan, with us today to talk about all things AI. Reid, Eric, welcome. Thank you. Pleasure. Thank you. My pleasure. Thank you. And Eric, we must start with the obligatory thank you for creating, of course, the platform that's bringing us all together today. So awesome. You've said a few times that you think we'll be able to send digital versions of ourselves to most mediums so that we, the humans, can focus on what really matters.

0:54Could you share a bit more about that vision? I'm curious, Reid, if you buy that or how you think about a world in which we're all sending our bots to mediums, what does that look like and what's the timeline for that, Eric? That's a great question. So when you look at a number of meetings, is an external meeting or internal meeting? For external meetings, I never wanted my digital for myself, a digital to join, no matter how good or smart it is. But most of the meetings may be internal meetings. For all of us, right, for all those internal meetings, some meetings you must need to join. Some meetings you do not need to join, you will cancel.

1:28But a lot of meetings in between, You do not know if you need to join or not. You're not sure. Quite often, you join a meeting, and after 10 minutes, you realize, gee, why should I join this meeting? I should leave. For those meetings, what if you send a digital version of yourself to join those meetings and summarize everything and assuming you are in the meeting? Or maybe that meeting becomes so important, you send your note. Hey, Eric, please come to this meeting. You better join, talk with him, something like that. I think this is very important. And furthermore, as AI technology advances further, imagine even for external meetings.

2:06Say, Ben, you and I, maybe we are negotiating a contract. I do not want to send my digital twin to talk with you, but you can send your digital twin. I can send my digital twin. Those two have a conversation. Homemade license you want to build, you want to buy, and what's the current user and so on and so forth. And come up with a preliminary contract. Then we can just have a signature. So those cases, a digital warning of yourself can truly help. So read, does that resonate? Broadly, yes. Among other things, I already have a digital twin of myself that I send the things. Yeah, you have read.ai.

2:42If people haven't seen it, check it out. And I've sent it to give statements at conferences, classes, interaction in classes. I've had to do some even media interviews. And so it's actually already done a bunch of things broadly, externally, for a combination of when I actually can't go. And also when it's entertaining and interesting for that. And I think what Eric's gesture is fundamentally correct about is that I think there's going to be a set of different kind of external and internal places where there's more utility all around for the AI joining. Because we as human beings have a limited amount of time.

3:37For example, say there is a discussion in a class that wants to talk to my perspective on something, and I can't be there. I'm traveling and something else, but they go, look, it's worth it to us to have read AI. I think there's always going to be a preference for read over read AI, for an Eric over Eric AI, for Ben or Ben AI. But there's going to be a number of circumstances where that's valuable if it just doesn't work for us to be there. But for read AI, read AI is trained on a massive corpus of writing that you've done and so forth. Eric Yuan, when you think about everyone having a digital twin that's showing up to mediums, that's a twin that's trained on what data in order to properly represent that person's opinion.

4:22I think the trend of my digital twin is more like all the data I can access, corporate data, personal data, any data I can access. That's step one, but not good enough. Ideally, all the data here, all the knowledge here, I've seen stored in my brain. I do not think those data are available yet. That's the reason why my digital twin is also not available, meaning to match my own capability. It's not there yet. Okay, got it. By the way, there's an important set of different functions. The reason I just wanted to continue the thought, which is, look, if you said read AI to make decisions in place of read, we are very far away from that.

5:01However, being able to represent a certain set of things about why I think and how I operate, it could do a zone of that already. That zone will increase. But also part of the gesture to where I think it's important for people to pay attention to Eric Singh is to have read AI. say we're doing an internal meeting and to have read AI show up to record, give some probable proxies of early reactions can be very useful. And that's the reason why I think there's a bunch of differences. And I do think that one of the earliest interactions of a broad spread digital agent, it's not one that's necessarily trained on all your depth of corpus, but is actually, in fact, you're, it might be training a bunch, but who knows, is like a kind of a note taking, hey, I'm here to pay attention, because I'm going to be giving a summary to Eric a little bit later about this and other things, and may have Eric intervene, which could be suddenly useful across a much broader swath of, for example, people working in Zoom.

6:08Yeah, fascinating. Yeah, Ben, just quickly, really right on. I want to share with you a very simple, or basic use case. Yesterday we had our Q2 earning call. Used to be I needed to record audio or maybe read audio in the earning call. Since last quarter, I just generated AI, AI generated AI Zoom, Eric avatar. And we put it by avatar for earning call. It works so well. I never need to prepare anything anymore. So just to make sure I understand. So for the Zoom audience call yesterday, your remarks were, did you deliver them yourself or? No, it's AI generated avatar. So essentially I give a 30 second in my avatar to our marketing team.

6:52Marketing team use an earning script regenerate AI avatar. We play back that avatar. Okay, so the audio itself was AI generated? Yeah, a video too. A video too. Oh, wow. So it was a video of you doing the earnings apart completely AI generated. I love it. Wall Street's thrilled. The Eric one, we have a bunch of founders on this call right now, people who are building AI products and applications. I'm curious for both people, let's start with you, Eric. When you think about at Zoom building AI features internally versus relying upon external apps to provide functionality, how do you think about that?

7:29Like, when do you think about we need to own the note taker versus we can let people use external note takers? What other functionality do you think needs to be owned by platforms like Zoom and then Reid, similar to you, like in the Microsoft product suite, what is Microsoft going to own and operate versus where is there space for entrepreneurs to build valuable external applications? Yeah, great question. So in the case of Zoom, I think all those new AI stuff coming from three areas. One, I look at our core competency, right? For example, how we can leverage AI to improve the product experience or build something new or very innovative features.

8:04That's always number one. Number two is we spend a lot of time talking to customers to understand their pain point. And then when we come back, think about what we can do to level the AI, to deliver something faster than our competitors. That's the second area. Third area is more like, are there any new things, brand new things with AI capabilities we can monetize? It's a great brand new thing. That's the reason why we're looking forward to announce something new next month. I think ultimately, especially not for startup, but for the sort of mature company, you have to look at that. But startup company, opportunity-wise, is huge.

8:41You can level your AI, build whatever you want. There's so many opportunities over there for consumers, developers, enterprise. I think just to focus on one use case, to level AI can quickly get things done. And so just to clarify, so Eric, in terms of your roadmap, app, how many of the applications that people currently use as external applications that they plug into Zoom do you think you'll eventually build yourself internally? Oh, I think for, again, of a core company, a core product, for sure, we want to build it by ourselves. Like AI networking, support a cross-platform, for sure, we should own that, right?

9:14Other tools, our marketing team, sales team, and engineering team, let's see, take a cursor, for example. This is great tools. We all love that. We never want to build something like that, right? So we leveled more and more AI applications from external rather than we built internal. But again, we also want to have an agent to talk with all those applications to quickly get something done, fully automated everything. Yeah, Reid, what are your thoughts? I think it's still, the general principle is still like, where do startups and large companies have? And I think AI changes the landscape a little bit, but the principle is still the same, which is in the things that a large company is doing most centrally and where its position gives an advantage is a terrible idea for a startup to go after.

9:59So like you go, I've got a new desktop search idea. Like I'm just going to do desktop search. Now that's different than chat, GBT and so forth. But desktop search and I'm going to go after that's not a good idea because it's okay. Google and then even Bing and all the rest. Now being said, all these large companies have lots of efforts. And so you go, well, wait a minute. Google is doing this over here or Amazon's doing this over here or whatever. You go, well, actually, in fact, it's not in the core set and it doesn't leverage the key assets that the large company has and is broadly potentially available to startups because startups can take risks, do go to market, move faster, be much more focused.

10:44because there's a bunch of things that the large companies will simply not take risks about that's outside the core three, five things that they're doing. Maybe one is a top line. And so you can view all that stuff. And so it comes to a score of, do I actually have a really bold idea? Is there a really smart risk? Do I have a go-to-market? Do I have a distribution possibility? Usually distributions, one of the hardest things to get to. And so I think there's a broad range. And part of what to build on Eric's comment on, look, with AI, you can make a whole bunch of new things. You might see something that other people don't see.

11:19You might see, have a speed and a risk of I'll build it and I'll deploy it and I'll see what happens. Like just building and deploy it doesn't happen in the large companies. It can happen in startups and see what's going on. And so that's the kind of thing that I think is a very broad surface. And part of the thing that I love about being a tech investor is the things that I hadn't thought about that someone shows up and goes, oh my God, that's amazing. There's like the Airbnbs. It's hope. Whoa, that's a great idea. And to read your building startups, investing in startups, now that you spent the last few years on the Microsoft board, I guess, has that given you a new appreciation of the power of big platforms like Microsoft?

11:58Or are you newly energized by the opportunities for startups in a world where big tech is so dominant? Or how has that changed your mind at all in terms of the surface area for startups, now that you have such an intimate view of Microsoft? The short answer is both. And part of it is this question of Microsoft has a bunch of assets that make it a very challenging competitor in certain arenas. Large-scale enterprise, productivity software, a bunch of that kind of stuff. It has a bunch of things and makes it hard for a startup to do. On the other hand, like even take some of the things where you'd say Microsoft has a natural advantage, like coding with not just GitHub, but Visual Studio and a bunch of other things.

12:43And it's coding, the Microsoft basic and tools goes all the way back to the very beginnings of the Microsoft concept of being a software factory. And you look at it and you go, okay, cursor. Okay, Cloud Code. So the range of availability for doing compelling stuff with startups is actually, in fact, much broader than most people think. It doesn't mean that there aren't really central things like what's happening with ChatGPT and the OpenAI Microsoft work and what's happening with Copilot. And Microsoft's putting a lot of energy into how does this upgrade the core things that Microsoft's there with Microsoft Office and Active Directory and other kinds of participation in it.

13:27And so that's challenging for a startup. But on the other hand, there's a huge range, much broader than most people think for startups. That's the reason why the answer is both. Yeah, Eric, I'm going to switch topics, but anything you want to comment on there? I totally agree. So even for our company size, 7 ,500 employees is very slow. I think startup, you leverage the AI tools with 20 people. You can build a lot of things so fast. You can disrupt anyone you want now with the AI tools. Yeah, indeed. And that's why it's such an exciting time to be an entrepreneur. and we're thrilled to back so many of them with your guys' support.

14:02Reid, you've been thinking about AI for a long time. Both of you have, but Reid, let me start with you. You've seen different hype cycles, helped get OpenAI started before people really cared about OpenAI. What have you changed your mind about in the last six to 12 months, given that broad context and all the things you've seen and your moments of excitement, moments of doubt? It'll be a little bit of a two-faced thing. So look, one of the things that I still believe generally, so I did believe and do believe, and then I'll put the change of belief in the frame of that, which is the way that adoption will happen is in this kind of stuff, generally technology is slow then fast.

14:37And it's slow because people are slow to adopt new things, especially in work and all that. And then fast when they're operating on a competitive clock, they're like, oh shit, you're like, I'm going to be beaten out by company X or person Y or something else and do that. So that's how I've generally been thinking about AI adoption. What I think though, it's still interesting is, the AI adoption has been slower than I've anticipated in various ways. Like when you go and ask people about what do they do for AI and their own personal life, very few of them or work life are as advanced as just Eric is.

15:13Think about like just about every public company could do what Eric did with the earnings announcement. And I don't know if there's another one other than Eric. And so it takes at the moment very forward individuals and it's simply all there. but people aren't doing it. And so it's one of the reasons why in multiple ways, I'm always trying to drive my own AI use for myself and for my team to be more advantaged and like companies on the board of and else because the tools are not perfect at doing everything. Like you can't say, oh, go invest for me. It doesn't work that way. But like in everything I'm doing, I'm thinking, hey, how do I get really amplified by doing this.

15:55And I do like multiple AI things for work for my own work every day. That's, I think. So basically the slower, there's a bit of a slower adoption curve from the masses than maybe you anticipated, given how powerful these tools are. That's exactly right. I don't think it'll be slower forever. Yeah. Just thought it would have kicked off more strongly already. And Eric, have you seen that within either enterprise users of Zoom or consumer? what's been the adoption rate of your AI feature set? Has anything surprised you there? I think, first of all, I completely agree with Reed. But on the other hand, also say for any new technology, right, from internet or mobile or AI, I think the consumer adoption is always the first one, always faster than enterprise, right?

16:41I still remember early days, I carry my both iPhone and BlackBerry for two and a half years. And finally, I dropped my BlackBerry and only use iPhone. I think today most of consumers, they benefit a lot from using chatGBT. But also we have a lot of AI features, a lot of companies announce a lot of AI innovation features. But enterprise adoption rate always is slow. I think at least two or three years behind a consumer adoption. The reason why, because you take a Zoom AI feature, for example, when you're working together with our enterprise customers, legal team, security team, data retention policy, compliance, a lot of things you need to make sure they feel comfortable.

17:20They take some time. But however, I feel like between consumer and enterprise, there's a sweet spot, which is developers. They are part of enterprise, but they're not part of enterprise. And I think building something to target developers is always, I think, probably right away, if you feel like very slow. Otherwise, no matter how good your technology is, whenever there's a new technology paradigm shift, it's always slow for enterprise to embrace. Yeah, that's a great point on developers being early adopters. Eric, when you talk to CEOs of companies every day who are Enterprise Zoom customers, we have a lot of founders on this call who are selling AI products to enterprise.

18:04What have you learned or what advice would you give to a startup CEO when talking to a company and pitching some AI application? What have you learned these last few years selling your AI feature set into your enterprise customer base? I think if you want to target enterprise customers, make sure, right? And be patient. That's one. Two is, even if let's say you start ahead of any other competitors, you've got to think about, hey, what if other competitors catch up? So what you can do to make sure, even if the enterprise customers, they may not deploy your solution fully, but think about how to make sure the enterprise customers are part of your innovation cycle.

18:46Meaning, they can help you generate more data. The data also very unique for you, right? And also the data, they are not going to give to any other competitors. So down the road, when they fully embrace your solution, competitors, they cannot catch up because you ought to have a bunch of data that's very unique to you. That's extremely important. That's number one. Number two is focus on maybe some vertical market with some domain expertise and knowledge. And also you can leverage AI. Any generic big AI companies, they cannot build that. Other competitors want to catch up. They may not have a similar domain knowledge or expertise, right?

19:23I think those two, or maybe even for enterprise department, don't build a horizontal product. It's very hard. Other competitors or big ones, they can probably, they can build something and give it for free. You know, maybe engineer or sales any department. One, we recently did a masterclass with a CEO for the village community. And he made a really interesting point about selling into the enterprise, which is like enterprise customers don't just buy the best product. They buy from someone they trust who can help them navigate this very anxiety-ridden time where everyone feels behind, they care about security.

19:58And so don't just optimize on product feature set. Of course, this is a perennial truth, but especially for AI, also become the trusted partner to the customer and help them navigate this territory. And I imagine, Eric, with Zoom and your brand and track record, that's a role you can play is we're going to be your partner in this. Even if our particular AI product may not be the number one feature set on the market, we have trust with you and we can build and grow together. Does that resonate? That's absolutely right. However, trust is extremely important. Let's say that for your customer, they have quite a few solutions on the table.

20:34Which one to pick? For sure, if they trust you, they are going to pick up your solution. That's guaranteed. However, if something new, right? And they trusted some vendors, partners, but normally they're relatively big, very slow. If they do not have a solution, they still want to deploy the solution from startup. Again, back to the cursor. It's a great example. The brand new startup company, a lot of companies already embrace that. Yeah, fair point. In this AI moment in Silicon Valley today, the revenue ramp of some of these companies is truly extraordinary. Like their company is going from, you know, zero to a hundred million dollars of revenue in record times.

21:11And of course, there are also a lot of companies that are scaling revenue and then losing revenue or churning customers, right? So it's very perilous in that perspective, hard to know what's sticky and what isn't. Both of you have built institutions that have lasted, you know, in LinkedIn and Zoom. What do you think for some of these young hotshot founders that have created companies that have grown extraordinarily really quickly in the early days, what kind of comes up for you when you think about the necessary ingredients to build an enduring business over time? Because I think, and I know in the case of LinkedIn, it was a bit of a slower, the early years, Reid, right, were slower growth.

21:47Eric, I'm not sure the exact history of Zoom in terms of how quickly were you scaling, right, in the early years, but I imagine the slower start, Reid, does cause you to embrace the idea of patience and build for the long term and develop a set of principles that can last. How do you think Reid will start with you when you meet some of these founders that have gone from like zero to 20 million in two months, what advice do you give them in terms of figuring out if they're on to something real and enduring and how to build a company that can last? I think you always have to have a theory of the game of what 10 years out are.

22:19These companies are not valuable in a year. They're valuable in, you know, 10 years and ongoing. And your theory of the game could be, hey, people don't really understand what we got. We've got a very slow thing, but it's compounding and it's going to be really valuable when it gets there. All LinkedIn, it could be, look, it's going to go really hard and hot right now, right? But that is going to also put us in a place where we can go do stuff. And you've got to have a theory of that game and it's got to be forward. It can't just be the, hey, look, it's working right now. Like you could both make a mistake of being overly predictive.

22:56You go look at Uber and you say, it's only like the black cars, like the current limo fleet. And you're like, the most often one is current TAMs or what dedicates them. For example, when you look at Airbnb, you could either say it's tiny because you're comparing it to people using classified to rent rooms, or it's huge because it's the entire like stay in the travel industry. And you got to have a theory of the game of what that is and how you'll be doing it. And it combines to what some of your current things is, what competition looks like, what it occludes. So it isn't really a remember plastics kind of thing.

23:33It's a, hey, what's your theory of the game? What I hear you also saying is be honest with yourself. Just because you've had rapid initial success doesn't necessarily mean you have a long-term strategy or a long-term moat. And what explains the initial rapid success and how durable is that? Eric, do you want to react to this? Yeah, really right on. Actually, lesson I learned is super hyper growth is always good. But however, it's also very dangerous. The reason why the high growth might hide lots of issues, you do not know, actually. Because you look at revenue, look at the bookings, MRR, right?

24:10Wow, that's amazing. The company is extremely well. Behind the scenes, there's so many things hidden there. That's the reason why earlier this, I specifically told our team, we don't want to grow faster. We want to fix those problems. You do not want to grow that fast because so many issues suddenly become the most important issues you cannot fix. The perfect scenario, you can grow very fast. At the same time, you practically fix all those hidden issues. That would be the best scenario, but extremely hard. I love that advice. Reid and Eric, thank you very much for your support of Village and all villagers everywhere and everyone who's joined.

24:46Have a great rest of your day. See everyone soon. Thanks so much for listening to the Village Global Podcast. You can check us out online at villageglobal.vc. We'd love to hear from you, your feedback, your ideas, your inspirations. You can email us at hello at villageglobal.vc.