I think increasingly we're finding ourselves betting on people who are just great at building products and kind of trusting that maybe there's an opportunity that we can't see, that this person can through the opportunity of AI.
The hardest thing with consumer is not only identifying the trend and the team, but getting the timing right.
Like when is this thing actually going to tap into the culture and be relevant with the culture?
And that's like almost an impossible thing to predict.
I just think it's a great time to be betting in any category.
I'm thrilled to be sitting down with Mike Mignano, a partner at Lightspeed Ventures.
Before Lightspeed, Mike founded Anchor, a podcast platform that was acquired by Spotify in 2019.
At Lightspeed, Mike's invested in some of the most legend companies in consumer and in tech overall Neuralink XAI, Suno and Granola.
In addition to his work as a VC, Mike recently co-founded Oboe Labs, an AI-powered learning platform.
Mike has a ton of thoughts about consumer startups and how AI is reshaping media and what founders should be paying attention to right now.
Mike, thanks for being here.
Thanks for having me.
Actually, love to start off with your founder experience.
I mean, we're doing a podcast now got involved and built one of the major pieces of infra that you know everyone uses.
What was that like?
Never intended to build a podcast infra company or even really a podcasting company overall.
My co-founder, Nir Zickerman, and I.
When we first started building Anchor, we were actually trying to make a social audio platform.
We had both kind of fallen in love with podcasts.
This was back in 2014, 2015.
Kind of like when, if you remember, Serial and Grantland and all that stuff.
We got into it and we realized that it was really hard to make them.
And so we had just built all of this photo editing technology for mobile phones in this company, Aviary.
And we thought wait, could we do the same thing for audio and podcasts?
At the time uh, if you remember, like everything that was launching on product time or the app store around the time it was all social networks.
So it was like oh, we need to build the social version of audio, as if uh, as if that hadn't been tried before and um, through a bunch of uh, pivots and starts and stops and sort of near-death moments, we ended up landing on the easiest podcast creation platform, basically a mobile podcast recording studio like we have here, but in your pocket, and easy enough to just tap a button to end up on Spotify.
Apple Podcasts start monetizing, you know right away.
Yeah, that was sort of the 2014 era was sort of that.
Right at that moment when people went from, almost all startups were consumer and then they became all sort of prosumer and B2B.
It sounds like that was actually part of your story as well.
It's like you started off consumer like quite a lot of other people.
And then that was also sort of the moment when the social platforms were sort of coalescing.
Totally.
And then the platforms were closing down a little bit.
Explosion in consumer in sort of the 2008 to 2012, 2014 period.
That was an opening up of platforms.
And then, as the consolidation happened, distribution closed down, which sort of spurred this move over to B2B.
Yeah and, and i think, like our, our thesis was, you know, similar to how instagram had made it really really easy to create and share photos, but to do it in a, in a format that kind of existed uniquely, just in their own platform.
We wanted to do the same thing for audio.
We wanted to make it such that you know, on on your phone, you could tap a button or you could hold the phone right up to your ear talk, maybe do some light editing, And then people would just come into this app to listen and interact with this content just inside of Anchor.
But what we found was increasingly, over over many iterations, people were good to create content on anchor, but when it came to listening it was like why would i listen to this like not great quality audio in this other random app when i have all these podcasts, like every podcast in the world uh, on apple podcast and spotify and i can listen over here?
I was looking at sort of the ranking listings of all the top consumer AI startups today.
And you have sort of two of the absolute top.
Yeah.
And I think actually Suno is an interesting one, because I think the thesis of Suno is actually quite similar to the thesis that I just described for Anchor.
And again, for a number of these kind of media platforms that came up in the years you mentioned maybe 2012 2013 2014, You know, if you go back to that time, or if you go back even earlier and it's probably connects to some of the work you did with your startup, it's easy to forget just how hard it was to publish content.
Right.
How to write content and publish it on the Internet, or to take a photo and put it in a feed that millions and millions of people are scrolling through, or take a video and upload it to YouTube.
Like.
Maybe you know we'll just use Instagram as an example that made it not only really really easy to create something beautiful with minimal effort, but then find a distribution channel where you can start to get it out there and build an audience.
If you look back at the history of all the products that have done this in one form or another over the past 25 years, nobody's done it for music.
The reason we believe is that technology up until AI did not make music creation easier, right?
The camera made photo taking easier, the camera phone made photo taking easier, the microphone made podcasting easier, the camera made video easier, but We never really had a technology that you know democratized music creation.
And so the thesis with Suno from the beginning was, well, now with AI, anyone can make music.
You know, you have like the most popular format maybe in the world.
Everyone listens to it.
What happens if you can get everyone creating it as well?
Has that one sort of evolved?
I mean, initially, I imagine it's, you know, maybe they were going after There's a ready-made almost like prosumer B2B audience.
You know, when did they get started?
Yeah.
I mean, it's still relatively new.
The company's more or less like two years old.
Oh, my God.
Yeah.
So even two years ago.
Yeah.
I imagine like you have this tech and you're like, how are people going to buy this or use it?
What kind of business are we going to build?
And, you know, I imagine the obvious start would be let's go after prosumer creators.
But, you know, that's not sort of... how it worked out, actually.
You should talk to Mikey, the CEO.
But you know, if he were sitting here, what I think he would say was the vision has always been to enable more and more people to experience kind of that joy of making music that previously only professionals could.
I think early on, when we were evaluating it, a lot of the behavior we were seeing it felt a little bit like a novelty.
I mean, ChatGPT felt like that.
Yeah, totally.
But increasingly, I think what we've seen and what we've observed is that more and more people are using it to make something that's meaningful to them, something that they're serious about.
They're becoming creators.
I think the other behavior we saw as we dug in which was super fascinating, was people were making music for themselves.
They're creating the music that then they will go listen to, which I think is super interesting.
I can't really think of a behavior we've seen like that in any other format.
People don't write to read their own stuff.
We're just such an early stage.
I think both of us.
While you're doing it, you're actually finding some of the biggest new categories.
I mean, you guys are doing it too.
Yeah.
I mean, we're working on it.
I think this last batch I funded enough to have an entire section.
It's about six startups that are all consumer-based.
Oh, wow.
Yeah.
So you're seeing a lot more consumer then.
Yeah, exactly.
That's awesome.
But I think we're a little bit in the minority.
Like, why do you think that is?
About consumer?
Yeah.
I mean, B2B, like, sort of ate everything over the last 10-ish years.
I think that's it.
I mean, I think the reality is there was a Especially pre-AI.
There was kind of like a playbook for B2B and SaaS.
And if you have the right team and the right wedge, like you could much more easily, I would say.
I don't want to trivialize any of this stuff, but it was more straightforward, I think.
But consumer has kind of always been a little bit more lightning in a bottle, right?
And I think the hardest thing with consumer is not only identifying the trend and the team but getting the timing right, because so many things are attached to cultural moments.
Like when is this thing actually going to tap into the culture and be relevant with the culture?
And that's like almost an impossible thing to predict.
You know, with Suno, I feel like we kind of got lucky that we met the team at a certain time and were able to invest like kind of just as it was inflecting.
Yeah, sometimes you just get lucky and maybe overpay just to be able to do it.
But no, I mean, there are obviously so many great consumer investors out there that you look back at some of these bets they made and you just think wow, like The timing was impeccable.
But the bottom line is we have AI now and we have all these new opportunities and things we can create that we previously couldn't.
So I just think it's a great it's a great time to be betting in any category.
I think increasingly we're finding ourselves betting on people who are just great at building products and kind of trusting that Maybe there's an opportunity that we can't see, that this person can through the opportunity of AI.
Yeah, that makes sense.
I guess the framework that I've been using that I hope turns out to be true, we'll see, is...
AI will actually increase retention, but you're still subjectunless.
You havethere aren't new types of distribution yet.
Classically, the death of consumer in the 2013 era was about the platforms collapsing in APIs closing down.
You have to bring your own distribution, but the platform won't give you more distribution.
There's this one-time opening and then a shutting, and then suddenly you have to charge, you have to have a subscription.
Consumer service.
AI, interestingly, is expensive.
On the one hand, people are willing to pay.
On the other hand, it's $20 a month or up to $200 a month if you're doing work with it.
Retention will go up.
So certain paid models will be possible now that weren't possible earlier.
But then you still got to solve the distribution problem.
That hasn't changed.
That hasn't gone away.
I do think there will be new distribution opportunities as AI becomes ubiquitous.
I mean, we're seeing this now.
Through AI, new distribution channels will emerge that people will exploit and find opportunities in.
But you're right for like a SaaS product or a pure net new consumer product today, you still have to just go and build the distribution yourself, which is arguably the hardest thing.
Yeah, it's funny.
I was hanging out with Eugenia from Replica.
She has a new startup.
Yeah, of course.
I was asking her about for a lot of these consumer startups.
How do you hire your head of growth for a consumer startup?
And she's like, actually, you can't hire them in the United States anymore or in the West.
They're all in Eastern Europe.
Yeah, I've heard that a little bit as well.
Which is interesting.
It became like a lost art.
We forgot how to do consumer distribution and we forgot how to create the SR-71.
It's been so long.
I think that her product as an example I think it's a great example of You could see her and that type of product, borrowing from the previous playbooks of Instagram and Twitter and maybe trying to leverage existing channels distribution to get that out there.
Like I don't know what are the ways that you can distribute an app on like an existing channel.
I think it's going to be well.
Basically it's like toy apps right now, but it's very well done, totally.
I mean, it's kind of the kind of stuff you could do in code gen tools today, but having it be entirely contained to your mobile phone yeah, i mean, that's like an instagram moment right there.
100 yeah, that's right, yeah.
And then i think like Basically e-groups, Yahoo groups, all of these things were trying to scratch an itch and you're just not willing to go through a customization step and group software is just the perennial problem.
But this is a perfect example of what we were talking about earlier.
Without AI and Code Gen, You're like, why would I invest in group chats?
Why am I so lucky to work on that right now?
Yeah, but now because of AI and CodeGen, it's like, oh my God, maybe there's actually a really interesting opportunity here.
Maybe you and I have missed the browser opportunity now that we have DIA and Comet, but I think back to two, three years ago.
I saw these browsers and I just I just passed on the opportunity because I'm like oh, you can't compete with Chrome or Safari because they're embedded in the OS.
And it's like no, actually AI creates like a really interesting opportunity for the browser to be an investable service.
But if I were a consumer founder going through YC right now, I might be thinking to myself What are all the opportunities that have sort of been written off that I can inject AI into inside a consumer?
What's funny is you should try all the things and and then you're only limited by distribution.
And then the funny thing is like, you know, what do you?
What's your take on?
You can access, i mean hundreds to at least maybe, maybe ten thousand, maybe tens of thousands of early adopters who are uh, anons on x yeah, so that's turned out to be uh, a whole way to, you know, get online and get customers.
I actually was telling your, your team, before the show that um, One of PG's essays was like a big.
It was like a big, big inspiration to how Anchor actually survived and not just survived but like ended up figuring it out.
And we were dying.
We were running out of money.
We had like three months before we shut it down.
We told the team, we're like, this isn't going to work.
This product is just kind of bouncing along.
Like it's growing, but it's not growing great.
And then we implemented this framework where like, we have to hit 15 week over week growth every single week from here on out for the next three months.
And it forced us to challenge our assumptions and it forced us to pivot to something that we didn't want to pivot to, but the users were asking for.
Oh, so there was pull from the market for, you know, sort of going in that anchor direction.
Yeah.
So we wanted to do the social thing.
We want it to be the platform because again, that would have been a bigger opportunity.
But we were hearing from the users is hey, we love your tools.
We don't want we don't want to bring listeners over here.
We want them to be able to listen where they're already listening.
Help us get our content on spotify and apple podcasts, and these these things didn't have apis at the time.
There was no way for us to port that content over, and so we actually we built out a whole framework where We literally had physical human beings manually creating RSS feeds and submitting them to the Apple Podcast Store on our user's behalf when they tapped a button in the app.
User didn't know it.
But they would tap the button.
Do something unscalable.
Yeah, do something that doesn't scale and then scale it.
Exactly.
And that worked.
And we only got to that because we had this rule that we had to hit 15 week over week growth every single week.
And it was startups equals growth.
And then otherwise, don't do it or do something else.
Yeah.
Yeah.
And every week it was something else.
Like, what are we going to do this week to hit the growth?
What do you think changed?
I mean, I guess, before implementing that, the default for startups is well, you have a bunch of money in the bank, you have some users.
And then the classic trap sometimes is, oh, well, we have some technical debt.
We're going to take a month just to like work down the technical debt, which is basically, like you know, treading water or like slowly sinking into the water.
Yeah.
And then you're drowning.
I think I think you don't realize you're drowning until you kind of like, look at the calendar and you're like okay, this is the date where, if we don't figure it out, it's over.
And like really staring that in the face and being like okay, we actually have three months to figure this out.
And yeah, i think there is this trap that happens often, especially now, where startups get over capitalized and they never, never feel that pressure and they feel that they can just keep going forever.
Yeah, there's this other aspect how many people were you at when we were about like eight?
Okay yeah, So we can get everyone in the room and we can be like this is going to happen.
We are going to fail unless we don't figure this out.
But if you had raised, like you know, today's seed and you had three times more money and maybe you had, you know, hired up and had like twice the size or three times the size of team and you couldn't fit them in a room.
Yeah, that would be harder.
Although maybe we would be spending it differently because of AI.
Maybe we would be, maybe we would still have a small team.
Maybe we would be investing more in marketing and distribution, TikTok influencers or something.
I think our burn would probably be higher, but we wouldn't necessarily be a bigger team.
Yeah, I'm hoping that some of the labs start experimenting even more with distribution.
You saw that with the GPT store at OpenAI a little bit, but I think that was a little bit of a failed experiment.
I think they might have even removed the GPT store or maybe it's still in there.
It doesn't feel like a super natural way to make an app.
Like a GPT doesn't really feel like an app in some sense.
Clearly they still want to chase this opportunity though, right?
Like it feels like multiple times they've announced partnerships with companies or, like recently, didn't they announce integrations with Spotify and all these other products that you could invoke right within the command line?
You could imagine a world that that turns into a platform and anyone can integrate.
I don't know how you would solve discovery in that sense, but Yeah, I think there's something there around MCP, but it's such a mess of an ecosystem.
The integrations just barely work.
I'm sure I will stop looking at it at some point in the next couple months and then magically in six to nine months when I'm not watching.
It's actually going to work really, really well.
Yeah, probably.
I mean, some of the consumer things that we're seeing are literally taking large data sets that are somewhat hard to get at and then just plugging them into LLMs.
What's an example of that?
I mean, there's a company called Nori in the current batch.
It's literally Apple Health.
Oh, right.
Dropped in.
Yeah.
Yeah.
Dave's a fellow podcast.
Yeah.
He did Chartable and sold that to Spotify.
So I think that's really interesting, actually.
I think I think I think that's an opportunity.
We actually invested in a company called Doctronic.
Which is right now doing, you know, kind of medical triage similar to what you can do in ChatGPT today.
But you know they have a model that's trained on a gigantic corpus of health information and data and you know medical research.
And you can imagine a world in which, Similarly like, you're getting your medical records in there and you're able to chat with it.
I'm doing this today just in like a cloud project, but it's super manual, right?
Like uploading all this stuff.
So I think Nori sounds super interesting.
My dad went to the hospital and he was fine, but he passed out.
He'd never done that in a restaurant.
We went, you know, hurried over to the hospital and you know I logged in with his Kaiser login to get all of his
Oh, wow.
I could see the labs come in on the website.
So I downloaded them PDF, uploaded a chat GPT, and then it just told me exactly what was going on, even though, like the emergency room doc hadn't come by.
But the funniest thing was like.
When he did come, I asked my three questions and then actually helped get him a better standard of care from that.
That's amazing.
Because I was like, well, had you thought about this?
And I half expected him to be like, don't use chat GPT on me.
Oh, the doctor appreciated it?
The doctor was like, oh yeah, that's a good idea.
We should check for that.
I'm sure they're seeing this all the time now.
Yeah, you have to wonder are they welcoming it or are they like Come on please, enough with the chat.
But I mean, I think it's that good.
I think it's all in distribution now.
One area I'm thinking a lot about is social.
Interesting to see Sora, at least.
Sora, I think, is really interesting.
I kind of think of it as... in a way, the end of social media or the last phase of social media.
I currently bucket social media into three phases.
The first is true social media, where companies are building up social graphs.
People are following each other.
Content is being distributed based on who you follow.
And so you follow your friends and maybe some random influencers.
And then when they make content, you get served that content.
And it's like pretty efficient.
Some of the content is relevant to you.
Some of it isn't, but it kind of works.
And then the second phase is like the TikTok kind of I call it recommendation media, where they figure out what you like and creators make content and then they program that content against your interests.
Sora, to me, feels like this third, the start of this third phase, where eventually they don't really need creators to make content right yes, today people are prompting, but you could very easily imagine a world in which you're just coming into the feed and the content is just being created immediately dynamically, on behalf of you, and i think that's really interesting.
Uh, interesting in a couple scary ways, but obviously there are also some interesting opportunities.
The thing that i like to think about in this last phase is what is the?
What is the role of the human to help kind of shape that experience?
Obviously, one of the roles is just to consume and let the model train on your interest.
But maybe the more interesting thing and we've sort of seen hints of this with sora i think sam published a blog post that kind of hinted at this is kind of the, the creation and distribution, potentially monetization of name and likeness and uniqueness.
Oh yeah, it's like when you cameo somebody on Sora.
Maybe that person is getting compensated some way.
Maybe brands are servicing as cameo-able.
You talked about distribution earlier, like that could be a new.
That maybe is the new form of distribution in social, where you're sort of like injecting some uniqueness or personality or likeness into a model that will then get distributed through no sort of manual human creation at all.
That feels like the next phase.
It's kind of scary because there's probably no actual pure human creation in that model.
There's the prompting.
There's some art in the prompting.
Maybe, but does that just go away?
Eventually, why do you need that?
They just know what Gary wants to... It's just going to be the auto AI slot machine.
So I'm a little bit terrified of that, but I also think that there's probably going to be some interesting opportunities.
And I'm totally expecting that TikTok and Instagram will start having pure AI generated content and eventually get to this place as well.
But all in all I do wonder if we're in kind of like this third and final phase of, like human created media.
Yeah, it's a little terrifying.
If you use Sora, it's a very promising, super funny but also extremely frustrating, like 80 of the time.
You mean on the output when you get out?
Yeah.
That'll get better.
Yeah, it'll get better.
Also, the mobile app itself, it's clear that they're experiencing crazy GPU scaling issues.
Yeah.
I think it's still number one in the app store.
It's crazy.
Yeah.
It is funny to see XAI and Meta sort of struggle to try to capture that vibe.
I think Cameo was the feature that really... did it for them.
But I also noticed that they didn't really invest a lot.
Maybe it was intentional.
Maybe it was just prioritization in some of the graph stuff we mentioned.
Like it almost seems like that's not going to be important.
You know, it's more going to be about, we just know what to program for each user.
Right.
It's like this person seems to click like on all of the Shaq cameos for some reason.
That's me.
The Shaq cameos are good.
Everyone's cameoing Shaq.
So yeah, I don't know.
I don't know what happens to social media and I'm not totally sure yet what the opportunities will be for builders there.
Yeah.
Distribution in the end.
I mean, even TikTok.
I'm hopeful that Chachi PD comes around and realizes like that was one of the most amazing things that TikTok did.
And that's what the YouTube algorithm does today.
Those are the most potent places to get distribution.
Right right, i mean x.
I think x is there like thanks nikita, he's doing a great job, he's doing it, he's changing up creating the web view.
Yeah yeah, gotta love the new web view.
Yeah yeah yeah, maybe the model is the distribution.
Then, like we're saying, and and maybe similarly how, for tick tock, it became about obviously, the videos and it wouldn't create a video, but you could also like you could let uh creators pull from the song the, the song catalog.
You know, instagram has things like stickers and filters and things like i don't know.
Maybe the model is a new place for some form of creator distribution.
It's not your video.
It's something else.
It's your likeness.
It's your brand.
It's a meme, right, that others can then invoke through the model.
So we talked a little bit about media and then a lot of people watching who might be just starting their builder career.
Is that like sort of the model for consumer founders in the future?
Should people be creators?
I don't know whether or not people should be creators.
I mean, I'm not saying they shouldn't, but it is also, it's time consuming.
I mean, you're sitting here in the set.
I don't know how many of these you do a week, but I'm sure it takes up a chunk of your time.
And I do think there is a new consumer playbook for distribution that I previously maybe a couple of years ago, kind of maybe was too dismissive of.
And now I almost feel like it's table stakes, which is leveraging creators.
Maybe it's TikTok influencers, you know, Reels influencers, whatever kind of creators you can tap into to reach some massive scale of distribution.
It almost feels like you have to be doing that now.
And what it can drive in terms of downloads or installs or signups is crazy compared to the early growth we saw from consumer startups five years ago.
Yeah.
I mean, this is literally the definition of organic then.
When you say organic like ultimately, the most potent form of organic is X feed, YouTube feed, TikTok feed.
TikTok.
Yeah.
And it's actually not organic if you think about it, right?
A lot of these companies.
They're figuring out kind of what works on TikTok and then the TikTok algorithm takes over and does the work and puts it in front of a million people.
It's not really organic, right?
Like organic, I would say, is you build some incredible product and there's this word of mouth dynamic where everyone and their friends are talking about it's just growing.
But just because it's not purely organic doesn't mean you shouldn't do it.
I think everyone- Correction.
It's not organic.
It's non-paid.
It's non-paid.
Well, you still pay.
You pay for it in your time and your mental space.
Yeah.
Or you pay the influencers or the creators or whatever.
When you're paying someone else.
The reason why that's interesting for consumer products is it's a mispriced asset generally.
Still,
Exactly.
I mean, Mr. Beast is not a mispriced asset.
He's getting his value from it.
But it does seem like the mispriced assets are sort of the creators with like 1000 to 10000 followers.
Right.
And if you can wrangle up enough of those, you can get some real scale.
Yeah.
I mean, I think I used to be dismissive of this as a tactic because it felt inorganic to me, but now i think it's table stakes.
Yep, i mean, all of the best consumer startups that are pitching us.
They have these crazy growth charts and they're all doing exactly this to do it now.
There's obviously then the question of what's the retention like?
Is anyone paying?
What's the funnel look like?
And if the funnel is trash?
So you need both.
It might not be a viable investment.
If you're a bad builder, but you're a good troll, then you still build nothing.
But if you're enough of a troll marketer who's actually got the goods, then you could build something pretty big.
Huge, I think.
And I think the reason I was previously dismissive of it is because I just assumed that at some point it would go away.
It's like if you're tapping into this inorganic channel that you don't control, at some point it's going to turn off or it's going to do something else.
But it's been years now and it seems as reliable as ever.
So back to the original question, maybe you should be a creator, maybe you shouldn't.
But you definitely should be leveraging creators and these distribution channels that creators tap into to get a crazy amount of eyeballs on things.
You have to.
The big question I see founders asking a lot and I'd love to get your perspective on this is should you wait until you have product market fit or some level of stable retention before you start doing the inorganic social distribution play?
So would you encourage a founder to seek distribution before they've found product-market fit?
I mean, maybe as a side project it might be useful to you know.
Start an Anon account thatjust to learn like How do other people think?
What do people click on?
Oh, that's interesting.
Then it's like a background process that you can call on later.
A lot of the time we spend with, especially consumer founders at YC is helping them with their launch, and how do they talk about it.
What's their launch video when they meet someone in a in-person setting, or an investor or a potential user?
Like, what do you say literally in the first 10 or 15 seconds?
It's like first you have to let someone know what the heck it is.
And then right after that you have to make sure that they know that what you're doing is awesome in some way, like that you're worth spending.
Like you know, that was 10-15 seconds and then it's worth at least a minute conversation and if you can make it to a minute you might have, like this amazing, like 10-20 minute conversation.
They might Try your product.
They might tell their friends they might, you know, They might invest.
Like all these different things happen and it sort of comes out of being Like perceptive good communicator, like kind of funny.
Yeah, that's pretty interesting and and the idea of doing it, at doing that or doing some of that, learning some of this as an anon, is really interesting, because you just lower the stakes.
You're not really like burning anything.
I'm, you know, thinking about um, how many times have you heard like, oh don't, don't do your big marketing push or your press announcement until you're ready, because you'll never get these users again.
You know you'll like.
You'll like burn the opportunity.
But if you're doing things like testing from an anonymous account or, you know, testing a tick tock influencer strategy even that, like there's always going to be more eyeballs on twitter, you can get or on tick tock, you can get in front of, and so Maybe the lesson is you just should be practicing that.
You should find a way to be practicing this stuff with lower stakes, even before you're ready to launch or you're ready to, like you know, blow the thing up.
Because figuring this stuff out is going to be hard.
Yeah.
So taste matters.
And there are lots of ways to develop that taste.
And, like it or not, you've sort of got to put on your fighting gloves if you're going to especially do the X thing.
And it might actually be worth it.
I feel like there was a lot of talk maybe six months ago, a year ago of, kind of taste as a moat.
It's becoming easier to build products because of AI.
So whatever you build, you got to have great taste and great, great craft to be able to stand out.
You know, I think granola was a great example of this.
I think the question is to what extent is that now a durable asset, given we've seen how aggressive some of the labs are?
Like, can taste really stand the test of time?
Or is it really just a thing to give you a first mover advantage?
Yeah, I guess I don't know yet.
I know.
Every version of the model, it has bigger and bigger model energy.
Yeah, it's got some bigger and it's got some BME.
These days, you know, like Opus 41 is It like feels vast.
Going back to Sora, I mean touching on this, this topic, a little bit like Sora, kind of surprised me a little bit Because it's a new product, a new app and it's good.
Yeah, you know, I think there was a belief among some startups and founders that as long as you're not doing exactly what chat gbt is doing or exactly what claude's doing like, you'll be okay, because that's where they're focusing their energy into these products.
But sora is kind of proof that no, these labs like, they have the taste and capability and the horsepower and the execution to build and ship net new products that also might run you over.
I think the point is though, all the more reason why, like you got to have taste, you got to be willing to put your product out there.
But you have to, like we're in this environment that is just so hyper competitive.
Like you have to.
You have to move fast, you have to be aggressive.
You can't just kind of like sit back and iterate and like wait for your moment.
I mean, the machines don't quite have taste of their own yet.
No.
Like that's what the evals are for.
Most people are in these consumer scenarios.
Even they're just, they're writing prompts and then they're trying to give a certain experience to the end user.
And then there's still a craft to that.
Totally.
So Mike, you're one of the most legendary consumer AI investors and I think a lot of people out there would love to know like, what are you seeing?
What would impress you?
What would get over the line so that they could get the chance to work with you?
There's a huge opportunity right now to re-examine opportunities that have been previously overlooked.
I mentioned earlier that we recently invested in a mail app, a category that I think we previously would have ignored, because it's been a graveyard typically.
But AI has presented brand new opportunities.
And I think we're seeing spaces like that over and over again, like these, these services that maybe got a lot of investment and building early on in terms of like the Internet or maybe like advertising and things like that that we've we've just like moved on from as an industry because we consider them baked and done.
And I think AI is presenting an opportunity to just completely like rebuild a lot of the stack that we've already built.
I would also encourage people to think about what are the large data sets that exist, either sort of, that are publicly accessible or that are private and personal to someone, that if you layer an LLM on top of it, or maybe a photo model or an image model or a video model or a music model, what are interesting things that you can do with these data sets that have been untouched?
We talked a little bit about health data and going into the doctor's office.
That's a huge opportunity.
Obviously, a lot of people are building for that right now.
But what are some of the other data sets that no one has really put AI around and given you access and insight to?
That you haven't really thought of before?
That kind of gives me an idea for a request for startup a little bit.
I mean there's just a lot of data and it's in your phone and it's I mean, frankly in your medical records.
It's in Apple Health.
Simultaneous to that, there are great startups like Memzero that are kind of trying to be like a memory layer.
But inside the system of record, inside someone else's app, it feels like there's some space in here around like an enabling tech, where you want a layer that basically knows everything about a given person.
I mean, and then you could even break that down even further, right?
Like, there's probably a bunch of really cool things to be built on top of your camera roll.
I don't know what they are.
And maybe they're social experiences, right?
But if we used AI to kind of look at your photos, the things that are in them, maybe the places you've been, maybe your geolocations, you could build an interesting thing on top of that.
I think it would know who you hang out with, who you spend time with.
Do you spend a lot of time with your family?
Are you an archer?
What your favorite beer is?
There's probably all kinds of stuff that's implicit to that.
Yeah, Dennis Crowley just launched something.
Dennis Crowley is the founder of Foursquare.
I still use Swarm.
Yeah, amazing consumer, product builder, thinker, super creative.
He just launched something where, when you put your AirPods in, kind of AI goes to work, based on your geolocation.
So you know, it might know, because you've done it a bunch of times that you like really good coffee or margaritas.
And then if you're near something, it's gonna tell you, it's just gonna like chime into your ear.
Based on your previous history.
So, like I just think, there's a lot of opportunity to take this, this memory, this sort of like personal information that you're talking about, and break it down on a bunch of different levels and run it through AI and create new consumer experiences.
Aside from that, you and your co-founders are starting a new company as well called Oboe.
How did you decide to work on that?
And, you know, how did you pick that idea?
Yeah.
So I think again, this is another opportunity that AI I think presents, and that is education.
I mean the highest level.
Premise is that We've spent billions, maybe trillions of dollars to invent artificial intelligence.
And obviously there's amazing opportunities that are coming out of that.
But what if we sort of took that artificial intelligence and invested it into human intelligence?
And obviously a lot of people have been talking about this for a while. ai is going to be a great tutor it's going to be great at teaching you things but nobody's really gone in and built the product for that yet and so so oboe and it's available now obo.fyi it is a product that anything you want to learn it will magically create a course for you on that subject in any format you want podcasts you know long form lecture It'll create the study materials for it.
And over time, more to the point that we're talking about with personalization and data.
It's obviously going to know how you like to learn in terms of the format, but it's going to know what you already know.
And so each subsequent lesson is going to get more and more personalized.
And then therefore more efficient at teaching you, because obviously the way in which we all learn right now it's like extremely one size fits all.
All the content is the same.
It's kind of a blunt instrument that's just kind of like forced upon you, whether it's reading a Wikipedia article or, you know, going to college or whatever.
It's just one size fits all.
But AI presents an opportunity to highly, highly personalize education and get better and better and smarter and smarter the more you learn.
And so the hope is that we can literally make humanity smarter through AI.
In my opinion and in my experience, the way to build startups it's to take these large surface areas where there's kind of obvious opportunity.
And yes, you start with a point of view, but like your point of view may be wrong.
And then you iterate, right?
You get punched in the face face exactly and then you alter your plan.
Yeah, you're still like kind of pointed at the same north star, but you're just, you're just taking a different path.
I mean, that's what we did with anchor.
We're like we want to democratize audio.
We think it's through social audio and short form voice notes.
Oh nope, it's not like.
Let's add more tools.
Oh, that wasn't it.
Let's add distribution to spotify, and I think that that is a recipe for a successful startup.
Take a really ambitious space Point in a certain direction and just keep iterating to find the best path until you get there.
Amazing.
Mike, do you want to give a brief plug for your podcast?
Yeah.
So when AI kind of exploded a few years ago, we at Lightspeed did the thing that every VC does.
And we started a podcast and the whole idea was to talk to people that are building in the space.
Then after a few months or whatever, we realized that, frankly, I got bored.
I was like, I'm having the same types of conversations.
And then so a friend of mine, Samil Shah, was like why don't we just go outside and record a podcast and we'll make it like anthony bourdain parts unknown or comedians in cars.
So we tried it out and we had a blast.
The audience loved it as well.
People were like oh, this is cool, this is different.
And so uh, now we've launched a new podcast called out of office.
It's gonna be fun and you're gonna come on right.
Yes, all right, can't wait.
Watch out for that one real soon.
Yes well, we're out of time but mike, thank you so much for hanging out with us.
Really appreciate it.
Yeah, thank you, this is fun.
