Latest / The Learning Corner by Precursor / Episode #32: Is AI Coming For Your Job?, Myths About AI Native Startups, Specialist vs. Generalist Investing

Transcript

Episode #32: Is AI Coming For Your Job?, Myths About AI Native Startups, Specialist vs. Generalist Investing · The Learning Corner by Precursor

• 0:00
  My older brother is an artist.
• 0:01
  He's a glassblower in the middle of Vermont.
• 0:03
  My sister works at a massive tech company, and I work in Venture.
• 0:06
  My brother's job is probably the most protected job right now with AI.
• 0:10
  It's the week of May 12.
• 0:12
  Welcome to the Learning Corner, a Precursor Ventures podcast.
• 0:20
  Welcome to The Learning Corner by Precursor.
• 0:22
  I'm Mia Farnam, and I'm joined by our GP and managing partner, Charles Hudson, where we will
• 0:27
  be talking through the articles and Goodreads that caught our team's attention over the past
• 0:31
  week.
• 0:32
  For our first topic in a candid LinkedIn post, Fiverr CEO, Misha Kaufman, acknowledges what
• 0:38
  many in the freelance economy have been quietly fearing.
• 0:41
  AI is already displacing large portions of creative and digital gig work.
• 0:45
  Misha highlighted that services like copywriting, image editing, and translation are
• 0:49
  increasingly being handled by generative AI at lower cost and faster turnaround.
• 0:54
  Rather than resisting the shift, Fiverr claims to be leaning into it, launching AI assisted
• 0:59
  tools, creating categories specifically for AI enhanced services, and really rethink what
• 1:04
  creative work means.
• 1:06
  He frames this moment as a critical turning point that will require platforms and workers
• 1:11
  to adapt.
• 1:11
  I think it really impressed the sentiment that, quote, AI is coming for your end, my job.
• 1:17
  So investors and CEOs have all been avoidant of the topic of AI displacing knowledge workers.
• 1:21
  It was refreshing to hear Meshep speak about it so candidly, maybe because it has had an
• 1:26
  outsized effect on Fiverr's core business.
• 1:28
  Yeah.
• 1:29
  Do you think, Charles, that we'll see more CEOs begin to share their real thoughts?
• 1:33
  Or was this kind of a one off?
• 1:34
  Before that came out, I had been wrestling with the same issue.
• 1:38
  And I had written something on Substack about, like, why won't anybody say the thing?
• 1:42
  And everybody wants to talk around it and nobody wants to say that this is gonna come for
• 1:48
  knowledge workers and probably sooner than we think.
• 1:51
  And as soon as I published that, bunch of people were like, oh, you should read this
• 1:54
  Fiverr CEO's.
• 1:55
  Mhmm.
• 1:55
  I was like, oh, wow.
• 1:56
  He gets credit for saying the quiet part out loud.
• 1:58
  I think he's right.
• 1:59
  The evolution of Fiverr's business, a lot of what they were able to do was find people in
• 2:03
  low cost countries who had high skill, who could do things like copywriting and image
• 2:08
  editing that could be done across distance and time.
• 2:10
  They were able to build a marketplace around that and they were super successful.
• 2:13
  In the same way, other people did the same thing for coding, and I'm waiting for someone
• 2:17
  who runs like a BPO or coding shop to talk about what windsurf and cursor maybe mean for
• 2:22
  the sort of offshore low cost coding world.
• 2:25
  I could cut that one both ways, but I think he's right.
• 2:27
  I do think society will probably employ fewer humans doing image editing, copywriting, those
• 2:33
  type tasks, and I wonder what will replace that work.
• 2:37
  Know, the the optimist says, oh, well, like, we will find some new high value task.
• 2:42
  AI will create new jobs for these people that we didn't know of before.
• 2:45
  You know, like, my friend said, hey.
• 2:47
  You know, electric vehicles, yeah, it's not as great for mechanics, but now you have all these
• 2:50
  jobs around charging station management.
• 2:52
  I'm from Michigan.
• 2:53
  And the auto industry went through its own big upheaval.
• 2:55
  And, yes, some jobs left for other countries, but also some jobs were just replaced by robots
• 3:00
  and automation.
• 3:00
  And no human does that job anywhere in the world.
• 3:03
  It wasn't like it was shipped somewhere else.
• 3:04
  So I I appreciate the candor, and I appreciate him not trying to tell people, don't look over
• 3:09
  here.
• 3:09
  Pay no attention.
• 3:10
  The Shopify memo, the Duolingo memo, Fiverr's also a public company too.
• 3:15
  Yep.
• 3:15
  So I do again appreciate the candor of it, but then you also have to weave it of it's another
• 3:20
  company that also has to push a narrative that they are building an AI capabilities within the
• 3:23
  platform.
• 3:24
  I've used Fiverr, like, all my life, and I've found it to be the quickest, low cost way to
• 3:29
  get a good value output.
• 3:31
  If you put it on, like, a plane, there's Fiverr, which I'm asking somebody to turn
• 3:36
  something around quickly for myself.
• 3:37
  And then the other side of it is I'm just doing it by myself with one tool.
• 3:40
  I've tried that.
• 3:41
  It's not there yet.
• 3:43
  The middle, always think about, which is things that I already use that have better
• 3:46
  distribution with AI tools.
• 3:48
  So I think about it with a podcast studio today, like video editing that exists in, like,
• 3:52
  Riverside and Descript, Podcast AI, all of that.
• 3:55
  That's where I kind of wonder where the value is going to transfer.
• 3:58
  But I feel like even if you have these AI tools, I don't think it makes bad video
• 4:03
  creators top tier.
• 4:05
  You miss some of that creative flair that you would get from somebody who actually is strong.
• 4:10
  Last week, our team conversation internally touched on how AI gives generalists a
• 4:14
  meaningful edge.
• 4:15
  In Nisha's post, he encourages people to use AI to further specialize in their respective
• 4:21
  field.
• 4:21
  So from your POV, what are the limitations to this and realistic gains?
• 4:25
  I just think the problem is AI just gets better.
• 4:28
  And so if you're a specialist, every year, the AI video editing tools are gonna get better and
• 4:33
  better.
• 4:34
  Ironically, probably because they're gonna be trained on your work.
• 4:36
  And so I think there's diminishing returns.
• 4:39
  Then some people say, maybe the new job is managing all of these AI tools that do work.
• 4:43
  I'm like, oh, managing agents.
• 4:45
  They're like, yes.
• 4:46
  I'm like, well, at some point, someone's gonna build a manager agent whose whole job it is to
• 4:49
  manage.
• 4:49
  The problem I have is, like, if you take this all to the extreme, there will be very little
• 4:54
  repetitive work that isn't done by AI or machines.
• 4:58
  And most work is repetitive work, which kind of only leaves creative work for now.
• 5:02
  And of course, we will have AI, things that are creative and generate brand new novel things.
• 5:07
  But I kind of feel like maybe we'll all be artists.
• 5:09
  I don't know.
• 5:09
  I know.
• 5:10
  But I I think the danger in specializing is like you're running against a machine that can
• 5:14
  learn from the internet, never sleeps, never gets sick, never has a bad day.
• 5:19
  It's like Waymo is the world's most experienced driver.
• 5:21
  Yeah.
• 5:22
  No.
• 5:22
  It's true.
• 5:23
  It's funny.
• 5:23
  My older brother is an artist.
• 5:25
  He's a glassblower in the middle of Vermont.
• 5:26
  My sister works at a massive tech company, and I work in Venture.
• 5:29
  My brother's job is probably the most protected job right now with AI.
• 5:33
  Turns out he has the most security when everyone told him he would never have it.
• 5:37
  This is a good bridge actually into our next topic, which I'm excited to break down.
• 5:42
  Casey Newton wrote a larger piece titled The Myth of the AI Native Startup, critiquing
• 5:46
  companies that aim to run on minimal staff through AI.
• 5:49
  Despite the hype and from what we've seen, real examples typically disappoint with companies
• 5:54
  either not using AI in a meaningful way or facing reliability issues.
• 5:58
  For example, Klarna's attempt to replace workers with AI fell a bit short in a large
• 6:03
  way, forcing them to rehire humans after a large round of layoffs.
• 6:07
  We're seeing that large companies' AI initiatives often stem from a fear of missing
• 6:12
  out rather than striving for success.
• 6:14
  Recent IBM survey of over 2,000 global CEOs noted that only 25% of initiatives meet ROI
• 6:22
  expectations.
• 6:23
  Trust remains the key challenge as AI tools like customer support bots still make pretty
• 6:28
  costly mistakes.
• 6:29
  And so while one person AI startup, that dream still exists, the reality is more like SaaS
• 6:35
  businesses with experimental AI features.
• 6:38
  And so I have so many thoughts here and felt like Casey was surfacing data to back up some
• 6:43
  of the hypotheses we've had here on this podcast and just within our own team.
• 6:47
  Internally, I know you shared this article, and it confirmed a lot of things that you thought
• 6:51
  would play out with AI adoption and usage.
• 6:53
  Yeah.
• 6:53
  First of all, everyone should subscribe to Platformer.
• 6:56
  And I read this article, was like, finally, someone has connected all the dots.
• 6:59
  Because I remember when the Klarna thing came out, and they said, we're just going all in on
• 7:02
  AI, we're getting rid of people.
• 7:03
  And I was like, well, then it should show up somewhere eventually.
• 7:06
  Part of me said, is this related to their desire to go public?
• 7:09
  And is this like a marketing narrative that they want in the ecosystem as they look to go
• 7:14
  public?
• 7:14
  Is this an attention grabbing marketing story independent of going public that positions the
• 7:19
  leadership at the forefront and gets attention?
• 7:21
  Because the difference between saying we're not gonna hire anybody and we're going all in on AI
• 7:25
  versus we're gonna slow down hiring and integrate AI.
• 7:28
  You don't write a story about a company slowing down hiring.
• 7:30
  Right.
• 7:30
  Someone who says like, we're not gonna hire anybody and we're gonna fire all of our SaaS
• 7:34
  vendors, that's how you get in the press.
• 7:35
  So I won't impute any motives.
• 7:37
  I don't actually know what motivated them to do that.
• 7:40
  I will assume they did it because they thought it was the best thing for the business.
• 7:42
  And then when the financials came out, was like, well, where where is it?
• 7:45
  Shouldn't it show up in savings somewhere?
• 7:47
  It didn't feel like it had really moved the needle, I was like, well, maybe it will just
• 7:50
  take time to peter out and filter through the financials.
• 7:53
  Or maybe it's not really what's happening.
• 7:55
  And it sounds like, you know, it's not what's happening.
• 7:58
  And the two things I've believed strongly since all of this AI stuff really got momentum, CEO
• 8:03
  wants to go to their board and say, we're not doing everything we can.
• 8:07
  I went to a dinner in LA, and Robert Smith rattled off just a ton of statistics about how
• 8:12
  hard he's pushing all of the Vista CEOs to integrate AI into their organization.
• 8:18
  And I have to assume every private equity and every growth equity venture firm is pounding
• 8:23
  the table about AI, so CEOs got the memo.
• 8:26
  Which means you have to spend money.
• 8:27
  But spending money and getting results are actually not the same thing.
• 8:30
  And I weirdly think this is the best thing that could happen to AI.
• 8:35
  Is for the pressure to simply spend money on stuff that you're not sure works, to check the
• 8:41
  box or to not be perceived as a Luddite.
• 8:44
  As that pressure abates, people will actually find the stuff that works.
• 8:48
  They will actually spend money on things that make their business better, and we'll all be
• 8:51
  better off.
• 8:52
  To that point, from that IBM survey, sixty four percent of CEOs from that report claimed that
• 8:57
  basically this fear of missing out drove their decisions to invest in the tech before they had
• 9:01
  a clear understanding of what that ROI looks like.
• 9:04
  And so clearly, even though there's data to support that there hasn't been clear mountains
• 9:08
  worth of value extracted from these changes, people are still pounding the table to make
• 9:12
  that change.
• 9:12
  The article even goes in further to share that a large portion of business leaders plan to
• 9:17
  double their near term AI investment.
• 9:19
  Perhaps this sentiment is fueled by watching once strong companies, so think Chegg, get
• 9:24
  absolutely destroyed by the emergence of these tools.
• 9:27
  And at the c suite level, I just wonder what balance do you think these folks are trying to
• 9:33
  strike?
• 9:34
  I think also, if you just think about it, how much would you pay for meteorite insurance if
• 9:39
  you were a dinosaur?
• 9:40
  It's actually a low probability event that you're gonna get hit by a meteor if you're a
• 9:43
  dinosaur.
• 9:44
  But if that meteor happens to hit Earth, boy, would you wish you had, like, a plan.
• 9:48
  Sure.
• 9:49
  And I think about, like, the amount of money these companies are spending relative to the
• 9:52
  scale of their business, it's actually not that much money.
• 9:55
  Yep.
• 9:55
  So if I were running a SaaS company, and if you told me for a few million dollars a year of
• 10:00
  spend, you can get some sense of how disruptable your business is by AI, and what
• 10:06
  are your internal processes you can make more efficient.
• 10:09
  It isn't okay to have this experimental recurring revenue or this exploratory budget.
• 10:13
  Like, that's exactly what you should be doing.
• 10:15
  You should be trying to figure out what works, and I don't think we're used to the scale of
• 10:19
  investment to find out what works.
• 10:20
  This is good hygiene as much as I'm cynical about the top down pecking or as a leader in a
• 10:24
  vendor, you should wanna understand what this technology means for your business.
• 10:28
  And in the grand scheme of things, getting a ChatGPT pro license for everybody in your
• 10:32
  company isn't actually that much money to find out.
• 10:35
  The last thing I will say is, like, on the other side of the coin, I think the Financial
• 10:38
  Times released a piece that said the insurer, Lloyds of London, released a product to cover
• 10:42
  companies for the losses caused by malfunctioning AI tools.
• 10:45
  Really, there's enough issues that people are kind of asking, how do I protect myself?
• 10:49
  So you kind of shared, like, the meteorite.
• 10:51
  What if, I don't know, you overpaid for that protection in the long run, and then now you're
• 10:55
  kind of hemorrhaging out?
• 10:56
  IVR has short memory.
• 10:58
  I think for the companies that overinvest in AI and maybe go too crazy, they'll just say it was
• 11:03
  transformative tech.
• 11:04
  It was the thing of the moment.
• 11:05
  We were totally comfortable being excessive, but we weren't willing to underinvest.
• 11:10
  Right.
• 11:11
  Like, no price too high if it really does pay off in the way, which is how venture
• 11:15
  capitalists think about their best investments that maybe they overpaid for.
• 11:18
  For our last topic, in a recent LinkedIn post, Nick Moran from New Stack Ventures reflects on
• 11:23
  a conversation with an institutional LP who shared her observation.
• 11:27
  It's that specialist VC funds tend to outperform in fund one, but then see their
• 11:31
  returns fall off in subsequent vintages.
• 11:33
  The insight highlights a common pattern in venture that we've touched on before.
• 11:37
  Specialist funds often begin with strong thematic clarity and tight conviction, which
• 11:42
  helps them win competitive early deals and deliver breakout returns.
• 11:45
  But as they grow, many struggle to maintain discipline, drifting into broader mandates that
• 11:50
  dilute their original edge.
• 11:52
  So the LP's advice was a reminder that what makes fund one great, your focus, your hunger,
• 11:57
  and a clear thesis, is often exactly what get compromised as firm scales.
• 12:02
  Wanted to get your perspective on it.
• 12:04
  We talked a little bit about this as a team.
• 12:06
  It's funny.
• 12:06
  The longer I do the business, the better I'd like to think I understand why LPs make the
• 12:10
  decisions they make.
• 12:11
  And I think what Nick outlined is true.
• 12:14
  I think oftentimes there are specialist managers who understand something about a
• 12:18
  problem space or a category that others do not.
• 12:21
  The founders who are building in those categories tend to wanna seek out people who
• 12:25
  understand their industry problem set.
• 12:29
  And so if you think about it fifteen years ago, not that many people were doing Fin It was
• 12:33
  regulated.
• 12:33
  People were like, it's too hard to sell the banks, and people didn't say, well, what if you
• 12:37
  don't sell the banks?
• 12:37
  There wasn't a deep expertise of fintech in the ecosystem.
• 12:41
  So if you were a founder building a fintech company, there weren't that many people.
• 12:45
  I know Mark Goranbrough.
• 12:46
  A handful of people I know who had real expertise in that world.
• 12:50
  And so I think that's why the early people in fintech were able to get access to many of the
• 12:54
  best deals because the knowledge of fintech, regulatory, technology stack, problems was
• 13:00
  concentrated in the hands of a few people.
• 13:03
  I'd say the same is true for Forerunner when they really lean hard on CPG.
• 13:07
  I could name a bunch of other like, I think about Basiset and Zetta being relatively early.
• 13:11
  I think the problem is Venture is now such a big business that any unique insight you have
• 13:16
  that is scalable, every big multistage firm will stand up a practice to go after it.
• 13:22
  And suddenly, the ability to find an expert person in your category is more distributed.
• 13:27
  You don't have to go to one of the specialist firms to get that access.
• 13:30
  Also, the other thing is these insights get diffused over time because those CEOs go up on
• 13:34
  stage and they go to CEO dinners and tell people stuff.
• 13:38
  Right.
• 13:38
  And people figure out how the business works.
• 13:40
  The downside is I still think the firm that I think really got this right was Emergence.
• 13:45
  They figured out SaaS way before everybody else.
• 13:49
  And because Venture back then was slow in the sense that information didn't travel at the
• 13:53
  same speed, and because SaaS was like really disruptive, they had a pretty long window where
• 13:59
  they knew more about SaaS than just about an and now there's not a single big standalone
• 14:03
  firm where SaaS isn't like the main event.
• 14:06
  I'd argue crypto had some of the same elements in that it was really disruptive.
• 14:10
  It was a different way to wire your brain, and if you wanted to be really good at that, you
• 14:14
  had to not be good at other stuff.
• 14:15
  And I talked to LPs, and he would say, well, then why would an LP continue to back
• 14:19
  specialist managers if we know this is true?
• 14:21
  I'd be like, well, the optimal strategy is you back a bunch of specialists in fund one and
• 14:24
  maybe fund two, and that's it.
• 14:26
  And every LP I talk to is like, you know, just like a fund manager has a portfolio
• 14:29
  construction, we have a strategy at precursor.
• 14:31
  They're trying to pick funds, and I talk to LPs who are like, well, we like specialist
• 14:34
  managers.
• 14:35
  And I'm like, me too, but only for the first couple funds.
• 14:38
  Because if those people are gonna graduate, they have to find a way to rebrand or
• 14:42
  reposition themselves in one of two ways.
• 14:44
  They either have to become kind of a multistage, vertically integrated fund in their
• 14:48
  category, kinda like a VMG, or you have to say, our original thesis was just that, the original
• 14:53
  thesis.
• 14:54
  Now we do everything.
• 14:55
  Yeah.
• 14:56
  So you end up a generalist anyway.
• 14:58
  The SaaS example, I feel like is directly related to how I feel about AI right now.
• 15:02
  I think specialist funds that came up maybe after SAS emerged who are maybe health care
• 15:06
  focused fintech
• 15:07
  Yep.
• 15:08
  Now for the past five years are like, oh, we do SAS for health care.
• 15:11
  We'll do SAS for fintech.
• 15:12
  Now I feel like the hype cycles are almost a good way of dressing up maybe like an overdone
• 15:18
  thesis.
• 15:19
  So AI right now is that.
• 15:21
  Again, I'm picking on health care because just that's my background.
• 15:22
  If you were a health care only fund for the past x amount of years and you're seeing
• 15:26
  returns plummet on your digital health investments or tech enabled services, AI right
• 15:31
  now opens up this whole new window for you to suddenly get re excited about this industry
• 15:35
  that you've been in for so long.
• 15:37
  But to your point, does that make you still a good specialist, or are you just kind of
• 15:43
  jumping in on the hype cycles?
• 15:44
  And how long will it take for us to figure out who's good at this or not?
• 15:47
  I think a lot about the other problem you have when you start a fund is you have to market
• 15:52
  that fund to limited partners.
• 15:53
  And as someone who's run a generalist fund for ten plus years, it is a lot harder to go into a
• 15:58
  room and say, you should pick me as a generalist because I am good at this job versus
• 16:03
  I am a specialist who has specialized knowledge about a market or an industry that isn't well
• 16:08
  understood.
• 16:08
  You're all my advisors and great xenos that I know in these spaces.
• 16:13
  The advisory council and Right.
• 16:14
  All the things.
• 16:16
  And I think that's, I always ask myself what goes in the memo.
• 16:18
  Eventually, the person you're sitting across the table from is gonna have to write up, why
• 16:22
  do I wanna give money to this person?
• 16:23
  And I think the specialist memos are easier to write.
• 16:26
  Yeah.
• 16:26
  This person has deep domain expertise in this category.
• 16:29
  The category is not well understood.
• 16:30
  We think they could build the franchise fund here, etcetera, etcetera, etcetera.
• 16:34
  The generalist is basically like, I think this person is an exceptional picker.
• 16:36
  That's a hard memo to write.
• 16:37
  Well, we are all out of time today.
• 16:39
  Thank you, Charles, for calling in from Bentonville, and thank you to our listeners who
• 16:43
  are joining in.
• 16:44
  Please like and subscribe, and we will talk with you all next week.
• 16:47
  Thank you.
• 16:47
  Bye.