Is AI a job killer or job creator?

0:04Hey everyone! This week we're digging into a new report by Mercury about sentiment in the early stage startup world, including some surprising trends for AI-enabled companies when it comes to hiring. But first, SPVs are still for chumps and Anthropic is not having it. I'm Ashley and I'm joined by Helen and Sally this week. This podcast is brought to you by Mercury, the fintech used by over 200 ,000 companies to simplify their finances. We're huge Mercury fans of great chat and we use it to run this podcast. A few of us are also current customers for our funds and startups, not to mention the dozens of founders we've backed as VCs, angels, and scouts that run their businesses on Mercury.

0:46If you are launching a startup or fund, visit mercury.com to apply in 10 minutes or less. Onboarding couldn't be more intuitive or delightful. Quick note, Mercury is a financial technology company, not a bank. For more details, check our show notes. Hi, everyone. Hey. Hello.

1:10Is that not how everyone starts their meetings? 100 % Has anyone seen K-pop Demon Hunter yet? No Okay, I've seen it at least 10 times Because it is now a nightly ritual For our entire family to watch it after dinner I literally don't even know what you're talking about Oh my gosh Okay, I promise I'm not living in some bizarre corner of the universe It is definitely a thing It's like been number one on Netflix for weeks no? you haven't heard of Hunters? Hunters I've just been in like a little bit of a of a work hole so I'm glad I'm glad I have our crew to enlighten me well okay well there it's a Netflix movie but there's also like a killer soundtrack and now I mean last weekend like every movie theater in the US had like a sing-along so like you join for these like for these you know showtimes and you go and you just sing along to the songs.

2:11Okay, anyways. Well, speaking of, just a follow-up from last week was, I don't know if you guys watched the Taylor Swift, Travis Kelsey podcast. That broke the internet. I just saw all the screenshots with the sort of concurrent viewers of the live stream. I mean, they not only broke the internet, they broke the record, right? So there's a new record set for simultaneous live stream views. And we, well, I was going to say we're going to break it, but we don't even do live streams. So maybe we will. I just love love. There was a bunch of conspiracies about, because Taylor Swift sprinkles a few like hints throughout her interviews of, I think, the number 60, the number 46, which is all leading to her potentially doing.

3:07the Super Bowl halftime show. So. Oh. Anyway. Well, it's here in Santa Clara this year, Sally, in an almond town here. So that would be very exciting. The interesting headline that I read this week was that Beyond Meat is filing for Chapter 11 bankruptcy. And I think that I'm one of the only people that thought that was interesting, but I thought that the storyline of Beyond Meat versus Impossible Burger was interesting because it was like a classic sort of tops down, which was Impossible Burger kind of going through restaurants and then bottoms up beyond meat going through like the homes, right?

3:40Like retail people were, were cooking beyond meat at home. And then it was going to be this fight for the middle and this classic sort of like, which path, um, you know, builds a stronger brand. But I feel like as we discussed in the group chat, like the, like the, the concept of fake meat was like never a good one. It's like unhealthy. I think when I, when it first came out, I thought, or at least on the marketing side, I thought as a consumer that it was actually a healthier alternative to red meat, which obviously I didn't really read into the, into the details of that. So I started eating those alternative meat because I thought it was healthier for you, but it's actually a lot worse for you.

4:18So once I realized that I, yes. And it doesn't have the benefits of me, which is like, you know, a lot of protein, uh, and other, other vitamins, nutrients. I think we're in this moment where like seed oils are evil, ultra processed food is evil. And so, yeah, so anything that is like, I mean, I think the number of ingredients and the processing that goes into making non-meat taste like meat is just unpalatable in this time. I'm like a huge burger fan. Like it's a big part of my identity and personal brand. And I remember like, wow, total flashback. I don't know if it was like when the conference was still called like, you know, D9 or whatever, or when it had been re-rent code conference.

5:04But I remember many years ago, I think it was impossible that it had like a sort of stand in the little food court and like trying the first one. And like a lot of friends were like, oh my gosh, Ashley, you love burgers. You have to try this. And I literally have a video of me trying fake meat for the first time. and my thing was so sad it's so sad I'll like I'll share it in the group chat but it was just kind of like all this excitement and then like oh man and I think if you like love burgers like like I don't know there's no there's no substitute for real meat right now but my um my hope I actually like diligence to view lab grown uh meat companies not to freak out my LPs I did not make any investments.

5:51It's very outside of my wheelhouse, but from a place of curiosity. And that's the trend I'm more excited about. It's kind of amazing what you could do. It's just not commercially viable yet because it's still way too expensive to do, but that's growing actual meat cells. And it's much easier to do it with ground meat versus hard to grow a a steak. And so, you know, maybe, yeah, maybe in 10 years, like I would love, I'm an animal lover, but I'm also a meat eater. And so maybe in the future, those two things won't be in conflict. Okay. Moving on before this becomes a burger appreciation podcast.

6:35How about we talk about psychedelics? Yay. Okay. So I was sharing in the group chat that last week I went to this dinner hosted by a VC and it was like a women's, women in VC dinner. And they brought in some really interesting speakers from maps um to talk about psychedelic assisted therapy and you know i think for for folks who have been exposed to this the idea that you know it can be a real some some um of these medicines are effective in treating ptsd and um you know there's a movement to legalize it and so there's a lot of discussion on that but i thought the most the two most interesting observations i thought from the dinner were um so the room was just full of full of feces right and everyone was so like curious and dumbfounded like why the folks from maps were so um experienced with working with people specifically in bc and pe and i sort of felt like i was like on crazy pills and like because it's really expensive because because we're all really broken people that is why i mean i think that's where most people were going and i was just like um because it's extremely expensive and inaccessible and there's only a group of people that can choose it i have no idea like conception it is i mean because because it's assisted you're paying somebody hourly right and that's several hundred dollars and then there's the actual like the actual medicine itself or whatever it is um the second thing that i thought was so interesting was um because everyone there was a venture investor right and so you know i think there was generally like positive affinity and sort of like the belief that this could be very um good net net positive for society and a lot of people who are very hurt um they were sort of like they couldn't take their venture capital hat off they're like how do i make this like vc scale how do i get this to vc scale or like vc backable and that was like a really interesting tension because you know i i said specifically um psychedelic assisted therapy it's not the access to drugs you know once you go through the legal hurdles and all that, that doesn't scale, there's telehealth and all of these things, of course.

8:45Um, it's the, how do you scale the like person who was with you, you know, for the entire journey. And that was just super interesting. And then that made me wonder like, where else do we see that sort of right now in AI and the discussion around like, you know, the pieces, the services components that don't scale. I know we're all working very hard to make the services component scale, but yeah. Yeah. It's funny. I actually do have an investment in this space called Journey Clinical. And what they have built is like the clinical backend to enable therapists to incorporate psychedelic medicine into their practices.

9:21So Helen, if you saw a therapist and they're doing more traditional, you know, traditional forms of therapy, like they could work with Journey Clinical for the training, the prescriptioning, the guidance, all of that fun stuff. And it's actually really interesting, something that I've been thinking about a lot is like AI is maybe coming for like the broader profession of therapy, right? You hear about all of these like AI therapy apps, people are using chat GPT and Claude and like very therapeutic ways. And so the fact that it doesn't scale without people might actually be a really important thing for this profession like how do you incorporate um you know different practices into your into your like offering the ai can't can't take and can't automate away and so and i do think from a regulatory standpoint like maybe someday ai will be able to like guide you through you know um uh uh psychedelic assisted therapy but that's or a robot shows up at your door and they are there to monitor you and administer and all of those things I don't think that that's crazy I mean but is the robot um just there as a human form just like conceptually but you really just need the voice assisted therapy or is that not the case because like then I don't think it's far off for a voice assisted AI to help guide you through through a journey.

10:52The one thing I was thinking about while you guys were discussing this was, I think it was Sam Altman said it after the GPT-5 release that, yes, a lot of people are using chat GPT as therapy, but none of it is HIPAA compliant and none of it is confidential. And so there's a lot of guardrails that need to be put in place before all this comes out. And if that doesn't, I'm very curious how the therapy space will continue to scale. I think there are companies like Mindbloom comes to mind. Like the version of this that is like legalized ketamine. I mean, ketamine is great because it's like a very short trip, right?

11:34You can do a 30 minute ketamine session. I'm fairly certain that with Mindbloom, you don't have someone with you. It is all a virtual like you go through their process you get a kit you have a you are assisted but you're assisted virtually and so in that sense it is actually much easier to imagine we don't need a robot who's like fully able to like come to your home and sit with you like that's um that feels quite scalable and then I think like the other like MDMA and then psilocybin are much like longer journeys. And so that is where you might need that in-person assistance. But that's super fascinating, Helen.

12:15And I will say it doesn't surprise me remotely that there are people fully focused on the venture and PE customer segment. Totally. So another headline we saw was Dario at Anthropic apparently telling one of its largest investors, Menlo Ventures, that it needs to invest using its own capital and not through SPVs. And so SPVs have come up several times on Great Chat, seasons one and two. And so that sort of reminded me about like, well, you know, what specifically is he talking about? And right now we are seeing and hearing a lot more of these like super hot, you know, SPVs into very late stage, you know, big rounds of open AI and Anthropik.

13:02And it speaks to what the demand is, you know, for these companies that people want to slice. But I think the specific thing that we're seeing commentary on is that a lot of these SPVs are multilayered. And so what that means is that, you know, we're just putting more middlemen between the company issuing the shares and, you know, the person who is buying into a piece of the ownership within the SPV. And so when they say multilayered, it's like it's an SPV inside of an SPV inside of an SPV. And so you're getting people who like are starting to like not have any connection to the original, you know, sort of person who's leading the SPV and is on the cap table.

13:40So any experience with or thoughts around multilayered SPVs that are for chumps here? Yeah. Is it also true that like it's not just that you're disconnected from the actual company, but like there are stacked fees. So just like with funds, funds are normally like 2 in 20 is the norm. Some SPVs adopt that, which is pretty wild because it's just like one company that you're managing versus a full portfolio. But let's even say it's like 1 in 10. If you are in an SPV that's in an SPV that's in an SPV, your actual exposure to the underlying equity is like very diluted by this layer of fees and carry.

14:26because a portion of what you are putting into the SPV is being taken out by fees. Right. And then the carry the, the 20 part, right. It's like anything that the SPV returns, um, the person who's managing it for doing very little work, right. Gets to take the first 20 % before you get, um, you know, your returns. So that was interesting. It reminded me of my time at AngelList because, you know, when you think, when you look at structures like this, like, and I think I've said before that spvs are a great vehicle for transfers like when you're trying to transfer ownership and there is transparency you know at every stage so if the original if the founder is aware that there is this spv and they know who the person leading the spv is and who the people who are buying the spv are like that's you know transparency and then the other important point is that there isn't information asymmetry meaning like there is information flowing and that the person who, even if you're on a 10 layer SPV, like at the very end is getting information, you know, from the company and it is, you know, truthful.

15:31I think those are like ways that it can work and be fine. And just like from a utility standpoint, be absolutely fine. The problem is, I think when there is asymmetry in information and that's often what happens like with these layers and you always have to point to like who benefits here and it's the, it's the person who's leading because they're taking a fee. Right. Right. And it's also the company who's facilitating it. So I was really excited to see that AngelList, you know, they banned multi-layer SPVs earlier this year. And that was, you know, while I was at AngelList, like that was a big sort of like shadow over the brand, because I think, you know, there were some instances of SPVs, you know, gone wrong.

16:12And that's, that was something that like we were sort of constantly battling with. And even now as an emerging fund manager, when I meet LPs, they're like, oh, you worked at AngelList. You must do a lot of SPVs. And I'm like, I actually don't touch them. And so, I don't know, it just reminded me that like if you are a platform, which AngelList absolutely is, like you, you know, and people always ask me like I'm running an API company or I am building a platform. Like how do I brand this platform? Like, well, the good news is all that pressure is off. You don't have to brand your platform whatsoever because your brand is the people and the entities and the businesses that are built on top of the platform.

16:50And so companies like AngelList have to actually be very mindful of what is happening on the platform. So in the case of there being multi-layer SVVs, it's bad for brand. It's interesting. It's almost like the cost of being such an innovator and making something accessible or productizing. Like I think about early AngelList and how it was so synonymous with syndicates for me. And then like even some of those early syndicates, like I'm like, I think I thought of AngelList as like Jason Kalkanis' investment. Totally. He was an early syndicate leader and was very – The people on the platform. Yeah, yeah, yeah.

17:25And so like how you feel about Jason Kalkanis might determine how you feel about AngelList, which is like a wild thing. Helen, I'm curious, like the, the, the anthropic complaint about Menlo, do companies not know, like, like if, so Menlo must have run an SPV for like a subsequent anthropic round, which makes sense because if they invested early, like, and they have pro rata, they're going to have like a big chunk accessible to them. I like, are they, and maybe this is, maybe this is specific to the terms of that deal, but are they not required to report out on like whose money is in that SPV when they invest?

18:07Because that just feels wild to me, especially when you're an investor on the cap table and like have want to have like a great relationship with what is going to be like what may be the most important investment we've ever made. Totally. I'm not sure that the person who's leading the SPV is required to disclose everyone. I mean, the founder could ask and, you know, maybe they would disclose it. But I think like, and I don't know the details here, but what would, what would make sense to me is that in order to fill an SPV, you have to send it around. Right. And so if you're a founder, you're like, what the heck?

18:42My investor is like basically selling or marketing an opportunity to invest in this SPV. And I don't want that information out there. I don't want some random person. Right. And so I think it's probably more the activities involved in filling the SPV versus like the, the, uh, you know, actual filling of the SPV, if that makes sense. But I'm not sure. I don't know any details here, but that's like, you know, I still am subscribed to so many, uh, syndicates on AngelList just from having worked there. And so I just get emails all the time and, you know, some, some emails are great and others are just like, yikes, I can tell you have not a lot of information and the founder does not know that this is going on.

19:24And that's really hard. I did talk to an investor who got into the last OpenAI round through an SPV, and it was a multi-stage fund. And they said that all they got was a five-page memo. Luckily, that fund had a relationship with one of the senior executives, so they did a side diligence to get the team comfortable, but it was a large chunk. um and then i also like this time around on the anthropic side i um i've been even asked folks from folks asking if they could try to get into the anthropic round for 50 60 million dollar checks and it wasn't like they were asking for any information they were like hey i just want to get in is there a path then so it's just continuing to pile on and um you know it that's where when the information you don't have that access it's going to get a little bit more complicated.

20:14Yeah. Maybe in a future episode, we can talk a little bit about another path to solving this problem that some companies like Robinhood are working on, which is tokenized private shares. There was that little blip with them announcing they had OpenAI available. And it turned out, I think that it was into an SPV. And Sam had to... And they didn't even mention it. It's not access to open AI activity. So yes, the sort of gold rush FOMO mentality is leading to some crazy activity. But should we head over to our main topic and dig into this Mercury report? Let's do it. So headline sponsor of Great Chat Pod and FinTech Darling Mercury surveyed 1 ,500 early stage founders across the U.S.

21:03in May to understand how they're really navigating the current economic climate. And the report was released this week and it's titled The New Economics of Starting Up. And it showed us that despite excessive capital, AI anxiety and macro uncertainty, founders are charging ahead. They're optimistic, spending more and rewriting the startup playbook in ways that challenged some conventional narratives. So we got a chance to get an early preview of this report and we picked out some of the things that we thought were most interesting. And so I'll kick us off. One of the most surprising findings was around AI as job creators versus job killers.

21:38I think that we've heard a lot of discourse around AI is coming for our jobs, you know, entry level roles, sort of like repetitive tasks. But actually, this study found that 79 % of AI adopting companies with significant AI adoption are hiring more because of AI, not less. So what do you think of that? I mean, it tracks to what I'm seeing in my own portfolio, which, of course, you know, is not 1 ,500 startups like Mercury's survey group. But yeah, it absolutely makes sense. I think the AI companies, well, companies that are growing faster are hiring faster. Like teams are still scaling with businesses.

22:22and it's funny I was reading the Mercury report right before a conversation I had earlier this week with the founder of an AI native startup I won't share too many details but this is a company that launched two years ago has really taken off they're up to 50 employees they're hiring like crazy like a lot of their go-to-market has to have a ground game and so So there was, you know, while AI has streamlined parts of their business, like to be able to meet the demand and meet the opportunity they're seeing, like people are still really essential. And so I kind of love that as a counter narrative.

23:02And I think it's a counter narrative that exists primarily in like the earlier and maybe growth stages. I mean, at the largest company level, we're seeing layoffs, right? I mean, Microsoft being a recent example of this. But yeah, it totally tracks with what I'm seeing in the early stage worlds. How about you guys? I mean, I thought about like marketing, right? Where, you know, if you had asked me 10 years ago, like what roles within a marketing function would I think would be replaced by AI? I would have very, very clear and immediate answers. It would be like marketing ops related because they're very repetitive tasks.

23:37And I sort of, as somebody with like, you know, always finite amount of headcount, you're making these trade-offs. Okay, well, I have to have a marketing ops person because our operation, you know, our growth team is growing at this point, which means I have to sacrifice maybe somebody on the brand side or another product marketer. And my thought is that now with AI, it's not like marketing teams are like, nope, I don't need the headcount. They're just going to find like other ways to fill the headcount. And almost like, well, now that AI is doing that role and providing us with more insights, right?

24:06Like it can actually make other people on the team better at their jobs. And so let's increase our goals, right? Let's like raise the bar and, you know, invest here accordingly. Because like, you know, typically you don't take venture capital unless you are in investment growth mode. And marketing is one of those functions that like you have to pour money into, right? In order to like sort of get that hockey stick growth. And so, you know, this to me tracks as well where, you know, just because you automate some functions, it doesn't mean that great, cool it's you know you're gonna have these standalone like single person teams yeah i was talking to a founder a few days ago who was telling me that the days of these like large sass contracts are over so looking at salesforce or workday they're like look these companies are not catering towards smaller stage companies like the one that the founder is at and so they're like the the the ceo brought in three junior engineers to sort of vibe code something that was more suitable for this like uh sass in a box for that company and he was like look the three junior engineers were much more efficient and cost efficient than you know the large contract that i would have had to purchase for a monthly sass fee and i think we might see more of that too um that's going to be more customized for smaller scale and rather than these bigger companies that are trying to fit into to sell into the smaller funds well that's so interesting like the build versus buy conversation.

25:40Like we've been in a buy mode for so long. And if you're in a build mode, like you need more people who are like concepting those products, like vibe coding those products, making sure data is like secure with those products. Maybe you have like more touch points and so you need more marketing. And so it's really interesting to think about, yeah, If you're not like sort of solving poor business problems with like these sort of more generalized SaaS platforms, maybe you do need more people on your team and that that spend is going to towards humans instead. This is like not the way that I thought about this before, but it's super interesting.

26:19yeah like if you're a founder that doesn't have the baggage or the overhead from like you know the pre-ai era right and you're starting a company from scratch like right now today you know how do you think about ai um especially when the technology is moving so quickly but like when you're thinking about building the team like that's something i'm you know talking to founders about all the time but like curious if you've seen you know specific sort of plans i invest so early that they're not at like headcount planning stages but like um Ashley maybe you see in sort of like the way that they're forecasting um at least like company growth yeah it's so interesting I actually I had a conversation at lunch yesterday where we were talking about like how the roles are changing and um especially as as sort of these like coding technologies get really good, like a swing towards more generalist over specialist.

27:14So like as you think about building out your tech team, instead of like, oh, okay, I need like front end engineers and back end engineers and like infrastructure people, but like you actually maybe just need technical product minded, design minded people who can like cover a lot of those areas. And so I haven't seen that necessarily show up in my companies yet, but it's going to be really interesting to see if the actual roles people hire for are scoped very differently than they have been in the past. I think it's the more traditional software engineering entry level positions are the ones that we've sort of seen reports of like those people not being able to find work.

27:59And so what is, if you are, there's a question of like, if you're an early stage founder, how are you building your team? There's also a question of if you are like a student today and thinking about like, okay, like how do I start to like assemble the building blocks of my skills to meet this world? It might look very different than, you know, than sort of the major or the path that you have picked even five years ago. This is like in direct contradiction to the mythology of everyone's been talking about, ooh, who will be the first like billion dollar revenue one person startup. And like, that is still very much like a myth that, that, that hasn't been delivered on yet.

28:37Um, there's a Helen, you were talking in the group chat about like the seed strapping concept, um, which is also like very catchy, very clicky. Right. Um, but it's not actually done very much in practice because like seed strapping makes sense. I think when you're like the only player in the market you're growing explosively and nobody knows you exist because if you were you know like growing in public and people like you'd have people flocking to the category to build the same thing and the to satisfy all three of those criteria is very hard um and so i think people are still opting to take venture dollars and sort of like you know get that big push in the beginning and and you know shoot their shot i have i have one of those companies in my Fundrun portfolio.

29:24And like, I won't name them because as you said, being under the radar is part of the move. And like, they are not hiring. They will not take more capital. They are growing beautifully. And I'm like trying to get them to take more. And I respect it. They keep saying no. And actually this brings us to our next question. I think the seed funding was enough to get them started. And now there are other sources of capital that are available to them that help them scale their businesses. of debt financing, et cetera, and so they can grow in non-dilutive ways. Maybe I'll take us to the next finding that we found really interesting, which was around more diversified funding strategies, so not just venture capital, and that companies that had more than four funding sources and were raising venture were 40 % more likely to raise rounds of$5 million plus.

30:20So founders who are more creative on the sort of fundraising and capitalizing side actually are more successful in raising venture. Any thoughts there? Maybe just a note, when we say diversify funding on the list is self-funding, friends and family, angel investments, VC, revenue-based financing, business loans. And that makes sense, right? If you're a founder who has a network that can get you to your angel round or have friends and family who can put in capital, you're probably just generally going to be better networked and better positioned to raise that venture capital, especially if you are able to raise a smaller amount to get started.

31:02It can be easier to get people's attention if you have that early traction. So that really tracks. I mean, I saw an interesting tweet this week that was around, like, who is venture for, which I think this, like, kind of gets at, right? And the person was saying, like, you know, historically, really only, like, 1 % of all startups should even raise venture capital. And now with AI and being able to build so much more efficiently, it's, like, 0.1 % of startups should be able to raise venture. Because maybe you can get to$1 to$5 million in revenue just on that friends and family round or on that angel round.