Mike Ghaffary (Canvas Ventures) - Could AI Be The Next Consumer Platform After The iPhone?

0:00This episode is brought to you by Vobin from Carta. Vobin from Carta is the easiest way to launch and run your venture investing. They offer SPVs and fund vehicles for GPs at all stages of the journey, from your first syndicate to operating a multi-million dollar venture fund. If you're interested in investing in startups, stick around after the episode where I chat with Gabriel Shin from the Vobin from Carta team, who shares his perspective and tips about how to start investing and how Vobin from Carta can get you set up. The link to Vobin from Carta's website is in the show notes.

0:44Hey, I'm your host, Mike Gelb, and this is a Consumer VC podcast where we discuss the intersection of venture capital and consumer innovation. If you're enjoying the show, also subscribe to my newsletter at theconsumervc.com, where you will receive new episodes straight to your inbox and a weekly recap of all the consumer deals that are happening. All content episodes are for informational and entertainment purposes only and is not investment advice. This episode is brought to you by Vobin from Carta. Vobin from Carta is the easiest way to launch and run your venture investment. They offer SPVs and fun vehicles for GPs at all stages of the journey, from your first syndicate to offering a multi-million dollar venture fund.

1:26If you're interested in investing in startups, stick around after the episode where I chat with Gabriel Shin from the Vobin from Carta team, who shares his perspective and tips about how to start investing and how Vobin from Carta can get you set up. The link to Vobin from Carta's website is in the show notes. Our guest today is Mike Ghaffari, General Partner at Canvas Ventures. Mike is the first third time guest on the show. The first episode, we spoke about how to invest in marketplaces. The last episode I had with Mike, we spoke about what's the next consumer platform after the iPhone. But funnily enough, we didn't mention AI.

2:04And so we thought considering everything that's happening in AI, we need to revisit this question and discuss the opportunity in AI from a consumer technology lens. Without further ado, here's Mike.

2:21Mike, round three, your first third time guest. Thank you so much for coming back on the show. Oh, thank you for having me. And I'm honored. What an honor to be the first third time guest and an early guest. And I've just, it's been amazing to watch your podcast grow and rise and now a premier tech podcast. It's pretty, pretty amazing and great to see. Oh, Mike, that was, you were one of the big reasons and big catalysts for it to, to kind of continue and to enroll with, with you being one of the OGs, man. I, well, I think you're going to do it anyway with or without me, but I'm, I'm lucky to be along for the ride.

2:57Yeah, really excited to kind of diving in on generative AI. I guess before we get started, I think we both want to do like a huge shout out to James Currier at NFX. So he was kind of what sparked this episode as well. He was kind of a generative AI. I brought from that conversation, our previous conversation about what's kind of the next thing after the iPhone. And so, yeah, big shout out to James. They're kind of spurring this without the intention to. So, yeah. Yeah, that was fun. I was listening to that episode because I just thought, hey, James is on. I want to hear what he's saying. He's always great.

3:36And it was about AI. And I thought, this is super interesting. I want to listen to it. And then you guys gave the shout out to me like total coincidence. I was like, wow, that's really nice. And then that led to you and I catching up on all of this. So this is great. Totally. Totally love it. In our last episode together, we broke down what are the different scenarios of what's the next big thing consumer platform after the iPhone. We talked about like Web3, crypto, the metaverse, VR. And then, of course, is like remote work? Could that be like a platform? We didn't get into at all generative AI.

4:13I feel like generative AI was not nearly on the horizon back then. When did you start thinking about generative AI? Well, it was kind of two parts to that question, right? Because I was a computer science major, so I've been thinking about it for over 20 years since college. But that doesn't mean I had the foresight to say this is going to be a big tech platform that's going to hit, right? So it's kind of two separate questions. I remember even probably seven or eight years ago now, I was really interested in this book called The Most Human Human about like a Turing test in AI. and I was fortunate to be, I was co-organizing this dinner with the author of that book when he was in town.

4:53And one of my friends who was helping me organize said, okay, we've got Elon Musk to come to dinner. He's super interested in like AI and generative AI and language, you know, and this was what this stuff was all about. And so once he was on the guest list, we're like, let's just keep it small because he didn't want, it was like seven of us, six of us having dinner times Arthur. So like these conversations have been going on for so long, But it just seems so speculative then, right? Like, literally, I think flying and landing rockets for Elon, that was more near term. And I was this kind of longer term at some point out in the future that people were thinking about more in the science project, but just not ready to be commercialized.

5:31And people have been talking about it for 20 years. You know, my firm, Campus Ventures, and one of my partners there had actually invested in Siri, right? that was bought by Apple and turned into like Syria was a startup before that came out of Stanford Research Lab. So it again, I think thinking about it has been a long time for all of us. But I think embedded in your question, the second is really like, when did we start thinking about it as this is going to be a consumer platform this year now, like investable that started. And I think there was a watershed moment around GPT-3. GPT-2 was a sign of it.

6:06Dolly and Dolly 2 were a sign of it. Like it started building. And then GPT-3 came out and that was kind of this explosion too. And it really kind of opened everyone's eyes like, wow, the moment is here. And so there's been several moments people are calling the iPhone moment, because we all remember the launch of the iPhone. And then a year later, the iPhone app store, and everyone's getting some of those vibes now. And it's got everybody excited. The jury's still out. It's not clear that this is going to have the impact that the iPhone had on that kind of timescale. In the long term, it probably will.

6:39But in the next few years, the history will still be written. So talk to me a little bit about what was so different about chat GPT-2, since I know you come from a software engineering background, from the second to the third. Yeah, I don't know if this was GPT-2 versus GPT-3 specifically. There's also, first of all, I think OpenAI does get some credit because you had Google, you know, sitting there with their and now they're doing a bunch of stuff with BARD. But they were sitting with a bunch of their work behind the scenes. You had Facebook has an amazing like LLM and kind of engine there. Right.

7:19Everybody had these different efforts, but they were all kind of keeping it under wraps. Right. Apple has Siri integrated and it's out there publicly. there was no major improvement in Siri. That's what we thought was the state of AI. It's like, I use it. And it's like, hey, set a timer for me. Hey, call this person. No, no, no, not this person, not other person. And it's kind of not always understanding you. You know, Amazon Alexa and chatbots got people really excited. But it wasn't quite the state of the art the way OpenAI was able to show us like, wow, you can generate like real stories, real essays like high school or college level essays you can start to like answer exams i think there were a few of these use cases where people would read it and say oh man and it's it all comes back to the turing test that i referenced earlier earlier that this most human human book is about um and famous computer scientist alan turing proposes like can the the computer actually write something um or say something where you can't tell you know is this human level intelligence at first blush when you're reading it, at least.

8:21And it takes a lot of work to figure out if a human had written it. When you get to that level, that's pretty impressive. No, totally, totally. And I mean, I think also, as you say, with like OpenAI, it also just became a lot more accessible as well to people that they can actually, you know, punch up your own query. That's right. So instead of keeping their locks, I think Google might've had this, right? I mean, they did, but it wasn't available, right? We had an EIR at our firm who was doing AI and had access to a lot of this stuff behind the scenes. and we were making some investments, but this wasn't public to be a platform.

8:52It needs to be publicly available. It needs to be that developers can build on it. And then OpenAI, like I went on and just signed up right for free. And then boom, I was generating, you know, my friends who are completely outside of the tech ecosystem. My mother-in-law was sitting there using GPT to generate fun stories and do interesting things and create ideas. And so it just, once you had that entrepreneurial energy, I think then the light bulb went off. Like, wow, this maybe really is open to everybody. There's a certain openness, hence the name OpenAI. Now, it's not OpenAI's kind of game.

9:24Now there's a ton of competition. So we'll see how it all shakes out. But I do think they get a lot of credit for pushing to the forefront. Well, as you say that now, it's like, oh, wow, I now have this platforming and can build off this platforming and create my own companies based off of generative AI. As an investor, where do you feel like the opportunities are when it comes to the different the different layers when it comes to generative AI? Yeah. So I think a couple areas. First of all, there's generative AI, which we'll probably talk about most. There are these other areas, right? There's AI infrastructure, and there's been a lot of investment there.

10:02I mean, I think literally to show you how this is, just today, there were two fundraising announcements on open source vector databases, just today alone, right? WeV8 raised$50 million from index battery NEA. And then I thought I had deja vu. I read about QDrent, another open source vector database startup the same day funding, you know. So it's like there's a lot of AI infrastructure tooling, a lot of people trying to be like layer zero, trying to be the next open AI or compete with them. People are raising on that. So there's a lot of effort going there. But that's not necessarily consumer, although it might be someone trying to create a consumer platform.

10:36And open AI ultimately is consumer right now with, by some measures, unprecedented growth of getting to like 100 million people trying the product out. So then you have, you know, you mentioned James Currier and his episode with you where he talked a lot about analytical AI being another bucket outside of generative. And, you know, I spoke with a lot of founders about that. Hey, what would you call the different categories? There's a company, OfferFit.ai, where I'm on the board. And the founder, George, there, he calls it applied AI maybe, but I think there's some similarities as analytical AI.

11:07And sometimes that could be using an LLM, you know, language model like ChatGPT. but sometimes it might be using, you know, another kind of area of AI. And so, for example, in digital health, right, I have, I was thinking about this, I have all this data of all these tests I've done, all these cholesterol tests and blood tests and glucose tests, and it's all unstructured. It's all kind of a mess. I'd love for, you know, AI to go and kind of analyze that for me, do something with it. That's kind of a consumer use case, could be generative AI, but it could also just be applying kind of the analytical aspect of AI.

11:42There's also in kind of contract management, there's use cases there, a lot of B2B use cases. But I think we can spend a lot of today because I think what's got people most excited on the consumer side is really this generative AI use case. And there are just tons of categories there. By the way, one more category we should talk about too that kind of hot off the presses of capturing people's imaginations people think could be bigger than generative AI agents, right? AI to do work for you, AI to go, you can imagine like, finally, are you going to get that personal assistant who can go out and just complete tasks?

12:16If you think about the movie Her, where you just kind of have in your ear, the AI is always there. That was a combination of a language model, but also an AI agent who's able to do task completion for you, which is important. And OpenAI is doing a big, you know, kind of trying to create a platform to do task completion as well. So you You can go like book travel or buy your tickets or do whatever else through plugins also. But OK, so those are all the preambles. And now it's a long preamble, but diving into like, OK, specifically consumer founders trying to look for generative AI startup ideas.

12:56I think there are a ton. The first place everybody starts is let's just port over all the existing experiences onto the new platform. They did this with the web. They did this mobile. like how will we search with AI, right? Displacing Google and everyone's saying, okay, like what are the new search? There's perplexity AI, right? Open AI, you know, people are kind of using also for it. How will we shop with AI, right? Displacing Amazon and others, right? Like what's gonna, just think of these big categories of where online and mobile has captured lots of eyeballs and dollars and that's all gonna move.

13:26How will we connect socially with AI, right? Is Facebook at risk here? Are we gonna now use AI to connect to each other socially in new ways, Facebook, LinkedIn? how will we do photos with AI? So Instagram and others who are doing photos, how will we do news? Thinking about my former company, Yelp, where I worked for many years, how will we do local? How will we find restaurants and places to shop and bars to go and electricians and roofers and doctors, dentists and lawyers, like all those services we need, how will AI impact that? Sometimes AI could actually do the service. Other times it might help us find the place or service.

13:59AI is not going to really become your restaurant, but it might help you find a great restaurant. So all of these are interesting. And then on top of that, there's stuff that only AI unlocks. The way only mobile could really, like the iPhone app, could really unlock Uber's opportunity. Uber was like the first time you could have a phone with GPS. It was always with you and it was a mini computer. And then all the drivers could have it and you could displace taxis that way, right? You could change a whole taxi industry? What's AI going to do there where it's not just porting over some web or mobile experience and saying this is the AI version, but reinventing another legacy industry?

14:36And the possibilities there are almost endless. There's tons. So there's lots and lots of opportunity here. It's actually very exciting. And I'd love to hear if there's any more you want to ask about or you were thinking of as well. No, totally. I think that those are great starting points in terms of use case. I love when you said the AI agents, it reminds me a little bit about James as well was saying on his episode about how you look at The Simpsons and instead of it being one, like a team of writers writing for The Simpsons, it might be one person that's actually then talking to like an AI bot and then kind of like screwing out different ideas and what have you.

15:12Riffing, exactly. So you have different ideas, actually then creating episodes. So using it kind of like as an assistant in that capacity that can actually learn from and actually develop anything. How, one of the things too that James talked about as well is that we discussed is where is the, will the incumbents take up most of the value or will it be the emerging companies? Like you mentioned, for example, Yelp. And in terms of how, do you think Yelp and some of like the larger organizations into all these different kinds of verticals that you're talking about, do you think that they're just gonna, not just because it is a huge undertaking, don't get me wrong, but are they going to come out with incorporating generative AI or AI into their platforms?

16:00Or do you see actually more so on the innovation side for new startups taking that value? So the answer here is yes, the incumbents are a big deal. And it's kind of like incumbents, some incumbents will win, but there's always big startup opportunities. And I think the best way to think about this is maybe just look at the analogies of history. History doesn't repeat itself, but it does rhyme, right? And so we go through, okay, look at the web to mobile shift. Actually, web 2.0 companies in particular were very good at incumbents actually winning. They were still kind of startups and growing. So Yelp's a good example.

16:36Yelp started as a web company, right? It was on your laptop and the desktop web, But mobile was perfect for it. And Jeremy Stoppeman, the founder, to his credit, realized as soon as iPhone came out, he said, oh, my God, this is like the perfect platform for Yelp. Local and mobile are inherently linked. We need to go all in on this. Right. And bought it the day it came out, built the iPhone app when it came. I remember, you know, the weekend when the final work was happening on the first Yelp iPhone app, the launch of the iPhone app store. and Jeremy was like stressed out about it, working late on it.

17:07And like, so it, you know, that's where incumbents and Google was search was able to, now Google had to go to great lengths. Google had to acquire Android, do all these negotiations with Apple. Like Google was able to retain a search kind of lead on mobile, but that wasn't obvious, right? Like Apple could have put an upstart in there and displace them. Or if it wasn't for Android, Google might not have default search on all those other firms, on all those other phones. So and then there's examples where incumbents get displaced all the time. Right. So like Instagram, right, is major social. What's like all these new social apps came out.

17:44Facebook bought a bunch of them. So that's another like through M &A, they can kind of do stuff. So it's hard to say. I think you're going to see all there are all kinds of brand new companies that and there are many mobile native companies that got built. There are going to be AI native companies that get built and you can't rule out the incumbents. incumbents will capture a lot of value they're going to try their best i will say the incumbents it's interesting like when you talk about a company like google or microsoft is actually doing a great job right now of capturing a lot of that ai value integrating it and getting it out there but his uh like right now these incumbents are getting more long in the tooth there are these massive organizations they're a lot older they feel a lot more like ibm was you know in those prior eras or some of these older companies where they have less kind of founders and entrepreneurs maybe on the front lines.

18:36And that will make it harder for some of them. You've talked about a number of different use cases for AI pertaining to consumer. I'm sure you get pitched now a lot of AI companies, not to call out anybody, but what are maybe some of the use cases that you actually don't really think are much use cases at all? Well, high level, even before use cases, I think one kind of yellow to red flag is just huge amount of money raised, huge or raising huge valuation pre-launch, you know, pre like we're just we have the idea, we have this and we're such an amazing team, which they are. We're going to raise it this like massive valuation.

19:14And there's a Goldilocks amount. There are some teams that are really good that have made that work, but way more often than not, that's been if you look through the web boom and mobile, like you've seen this time and time again that sometimes you raise too much money and it's actually um kind of uh creates a failure case for the company because you start burning so much out of control and you don't have product uh market fit necessity is the mother of invention and like you need some creative constraints actually sometimes um it's helpful to have some cushion but if you're raising 100 million dollars or 50 million dollars and you really don't have much that can sometimes kill the company.

19:51So that's always a little skeptical and that's tough. So it's good to be disciplined, I think, as a founder and sometimes turn away big piles of cash, even if you can get it, until you think you can really put it to work or it makes sense for the company. Some of the use cases that are tough, I would say when I think people, what I'm seeing a lot of is people taking existing pitches that wouldn't have been called AI before and they just slap on like, hey, we're using AI. You saw a lot of that with crypto too, like Web3, we're doing an ICO. And so like you said, I won't pick on any one company or whatever in particular, but it's a tough time too as a founder, because I think everybody's got to think about your AI strategy, whether you're a startup, a public company, or something in between.

20:36It's kind of like when the web came out. And we'll see if it's as successful as that. But you can't just say, we haven't thought about it, we don't have a strategy. So you have to think about your strategy. But a lot of companies are also losing credibility by just slapping AI into the pitch, thinking that's going to drum up VC interests. And I think if AI is really not at the core of what you do, it's good to be honest about that and just say, hey, we're this kind of company. And maybe you're going to come in at a more attractive valuation. There's still a big market need. And yeah, we might use some AI, but that's not the defining kind of nature of the company.

21:06That makes a ton of sense, like actually getting to the bottom of, okay, there is maybe AI incorporating your company, but how is it actually beneficial just like saying you know previous and maybe previously last year or two years ago oh we use blockchain technology well do you actually need to use blockchain technology you know what i mean because it seems like you're forcing the ai into the pitch to your point and it doesn't seem like you really need it in there people kind of don't get it especially you're like hey we're just using some open source or off the shelf tools right so we didn't develop any of the tech we're not using it in any proprietary way then people say like okay that's great they're using some new tool, but what, you know, are you really an AI first company?

21:47Totally, totally. You know, and as I think about, you know, generative AI in like this period, you know, I was asked by another investor who I, who's been on the show, love him, Daniel Gulati, about what are some of like the consumer implications for, for AI and especially what you might have, for example, synthetic podcasters or music that's, you know, AI this but like back in 2012 2013 when like the tupac hologram was um at coachella i totally remember this watershed like yeah it's like it's super super well it just kind of reminded me of that moment because you know at the time i was working at wme and which is one of if not the like biggest you know entertainment this is this tech podcast people might not be as familiar with with wme but but it's one of the largest um entertainment agencies in the world and i was working in the music department and mark geiger who's the head of who is head of music at time and he's like an absolute music industry legend highly recommend people go and and kind of check him out but um he would have these quarter quarterly meetings which were really actually like looking back pretty incredible that he made that time for assistance i mean this guy's a super busy guy and one of the meetings like after it was right after coachella one of the assistants like asked the question i mean i was an assistant but one of my fellow assistants asked the question like hey how is this tupac hologram going to change music and live music like are we at a point where people are going to want holograms and people are no longer really going to value seeing actually like their favorite acts in person that holograms will kind of do the trick and that we're actually not going to see touring quite what it used to be um is there already going to be like a technology component.

23:33And what Mark said is that, you know, at the end of the day, people love their stars. They want to see their favorite acts live in person. They don't, I don't think at the end of the day, they don't want to see like a hologram. It's a cool, like novel type thing, but you know, overall, I don't think it's really going to impact, but I, I, I, I bring this up because I was talking with Daniel and I thought about this, how it relates to podcasting. And they thought of me too. And maybe this is like the moment that we've had here, I wonder if you agree, maybe similar to the Tupac one, but the Joe Rogan, Steve Jobs podcast interview where it was, you know, generative AI, Joe Rogan and Steve Jobs kind of maybe similar to Tupac in the Tupac hologram.

24:17And will we actually see, you know, podcasting be disrupted, for example, with synthetic, with AI podcasters or, and I'd love to ask you this question of how do you how do you think media in general do you think will change or won't change well so first of all by the way this tupac thing so a uh mark uh you know your former uh our co-worker he was right right like it so far at least you know since then that was a while ago in the last decade there wasn't this big boom of holographic concerts that said at the time was amazing there was also as context for people who don't remember there was a conspiracy theory kind of like Elvis that Tupac was still alive, kind of wishful thinking people have.

24:59So some people who weren't at the concert, even people who were there, thought that was the real Tupac. It was so realistic. It did kind of blow people's minds how good it was. I think because it was nighttime, the way it lit up, people thought it looked very realistic. But then it didn't take off. So that was interesting. But fast forwarding to what you're talking about now. I think you will start to see, you know, there's A, this concept of like personalized media, right? So even there's kind of fiction and nonfiction, even in fiction, people think maybe there's going to be, you know, you're seeing with ChatGPT personalized stories, pretty soon there's going to be like, let me create a movie.

25:34And pretty soon, like, can the algorithm create storylines? You're back to the Simpsons example we were talking about earlier, right? Can you get personalized Simpsons episodes or entire personalized shows or comedies? You know, I like some of these political thrillers like the show Homeland, and there's a new one that I've been watching, it just finished. Can you generate new shows on the fly for that? For me, is Netflix going to get into that? I mean, that's pretty out there. There's a ways, I think, a lot of work to get there. But much simpler to your point, for now, can you just get a Q &A going where one or both parties are actually AI?