Conviction's Sarah Guo on the SaaSpocalypse, AI reskilling as a venture opportunity, and the coming ChatGPT moment for robotics

of uh the de-industrialization process that we've been suffering through for the last few decades. But it was just interesting to see it in like a viral uh Instagram reel. Anyway, um we have our next guest Sarah Guo from Conviction in the Restream. Let's bring her into the TV in Ultra.

Sarah, how are you doing? Welcome to the show.

I'm great. Hey guys,

what's happening? Great to see you.

Uh we we we like the hot drop. We like bringing you in just as soon as you're ready. But great to see you. How is your Hill and Valley going?

Uh, it's amazing. It is a much bigger than it's been in prior years, but there's a lot of work to be done, and it's a great group that they've assembled here.

Does everyone understand what a Neolab is? You introduced me to this uh during our Christmas episode, which you uh had a fantastic costume for. Uh, and I and I and I serious person. Yeah, it was it was amazing. All very serious people. But but have has has have Neolabs as a concept gone mainstream? I don't know that everybody here is paying attention to that, but the impact of the capability improvement from the labs and um lots of other projects in AI, I think Washington's definitely paying attention to.

And and what specifically? I mean, we talked uh we we we talked previously about some of the admins uh work on uh you know, protecting children and and giving parental controls, but what are the other areas? There's obviously the rateayer protection pledge, there's energy. What are the other conversations around AI that might actually ripple down to uh a startup founder, somebody in your portfolio, a neolab or even a lab lead? Yeah, I I think one of the uh areas that has come to the four is this idea that we need to have more industrial capacity in the United States to manufacture things from munitions to chips to uh parts for robots and and there are a lot of different strategic areas that America's given up in the supply chain and in order to reclaim them the path has to be automation right just in terms of cost of labor and I think there's growing recognition amongst both um people in the administration, policy makers and private companies working on it now.

And and what's at the top of the list? There's there's like zoning requirements, funding bills, tax incentives. There's so many levers that you can pull. What are people actually excited about?

Uh I think one of the uh basic blockers is just going to be the ability to build faster in areas that have been traditionally very regulated, right? And that could be generation, it could be transmission lines. A lot of it is about energy actually, but some of it is about labor. Let's talk about that too. Brad from OpenAI was pointing out that the bottlenecks to increased AI capacity in the United States are very real world. Uh and then they're very human like they're laborcentric and the ability to unblock let's say um more electricians or more folks standing up data centers or more power generation is actually something that Washington has control over. So it's something we have to work on together. Yeah, we were talking with Chase at Crusoe about this about retraining, upskilling, and I'm wondering like do you see that as a venture scale opportunity if somebody came to you and said, "I want to build a a trade school or even just like

55 electricians, 55 plumbers."

Yeah. Or even just like a a website, an app, an online course. I don't even know the shape of it. But is that a startup opportunity or is that something that like we firmly need to put back on the university system or the trade schools or the government?

Personally, I want somebody who built a startup previously to go work for the government to do something like that. Bring bring the speed. But if you'll but

what do you think?

I think it's all of the above. It's funny you should ask. I didn't plant this question, but I am in fact a longtime investor in a company called Uplimmit that focuses on reskilling and they work with lots of great companies on, you know, skilling for the AI age as well, mostly digital today versus um the trades. But I I think it has to be a combination of these things. And I would say um even if you know reskilling is something we should clearly do and make a huge investment in from both the public and the private front, it's not enough, right? When you think about the demand in this race where uh as congressman Molinar said like we cannot lose because it represents the future productivity of the global economy uh it is happening very quickly right and so even if we invest in upskilling which we should do uh with startups uh with universities with government programs we need like 10 times as many electricians like next year right if you just think about the capacity growth and so there's some very real limits to how quickly you can move u capability in humans. And so I I think we we need to go to automation for parts of this.

Yeah. Uh Jordy,

I'm curious if you've picked up on on the Hill's information diet. Like how are they learn? Hopefully they're they're you know watching every episode of No Priors, you know, indulging cash, but like where are they getting their where are they getting their information? Obviously the event is a great place to kind of like understand these different categories, but like where like what what what narratives are they

are they tracking and who is most influential from a

a you know kind of an information standpoint.

Well, I uh I hope they're listening to TVN all the time as well. That would educate them. I I think uh the uh one of the things that is always going to be clear is in order to take the risks that you need to uh in terms of understanding what's going on at the frontier and listening to different audiences that are not always going to be super open nor is everyone of the narratives of these different audiences supported by the mainstream media. You have to you have to go talk to the people, right? And and the thing I'm really encouraged by with something like Hillen Valley or with, you know, I've been in DC like four times over the last six months, which is more than five years ago by by far, but you also see like the ecosystem coming here and increasingly like we see lawmakers come visit us when they're out in the Bay Area. And so I think a lot of it is in person and I think that's like excellent in terms of education. How have you been talking to founders in your portfolio, founders you work with about the SAS apocalypse? If they're in tech, but they're not a foundation lab, a neol lab, and they're watching maybe the dinosaur that they were eating one bite at a time, the multiples compressing over there, and they're sort of thinking about their strategy, how AI native they should be, what they should do, how they should think about their valuation. What what what type of conversations are you having around that? One of the most useful things a board member can do is provide market context and like actually sound the um the red alarm. Yeah.

Uh once in a very long while, right? Uh and so here I I think if if people have not dramatically changed their strategy according to what's happening in the market with capability in software over the last three or four years like uh the vast majority of them have a real problem, right? And so I would like to think that the companies that we work with know that and a lot of really smart founders and CEOs know that. Uh but it I I one thing one distinction I would draw is I don't think the the entire world collapses into a few labs.

I I think the world is very big in terms of capabilities we need to go work on from getting people their you know prescription drug treatments with companies like Leon Health to work in AI or contact center like Harvey and Sierra. uh it is not clear at all to me that that is in the pathway of the research work that the labs want to do and the largely productivity and code focused products that they've built.

Yeah. uh how are you thinking about actual deployment of just like AI tools being AI first if you're uh sort of a growth stage CEO five or 10 years into your journey and you need to go and sort of revitalize the company. Are you more of a fan of, you know, crack like labs type team that has the authority to go and rewrite systems and vibe code tools and try and speed things up within the legacy structure or is it sort of the all hands meeting and everyone is going to be onboarding to AI tools today or they're going to get left behind? How do you think about the trade-off between like verticalization versus horizontalization within a company? I think that there's there's probably like three elements you need to really upgrade a software company for the AI age, right? You need first of all the urgency, which is if if you were a um enterprise software company in the last decade before chat and you had kind of one-year market in terms of adoption or lighthouse customers at let's say scale of more than 100 million on run rate, you felt pretty good about the future, right? barring you know limits to your market size at some point uh I I think everything has become much more challenging for workflow software companies and here's where the uh willingness to back founder-led companies and you see founders coming back in and getting much more active about this at companies at at technology companies of all scales where they have the moral authority and the urgency to say we need to operate very differently and so I think that's element one the second point is what you said which is uh given it's a general capability then you actually want every you know great companies it's not just like one function is good like often you are good at everything right you are good at finance and GNA you are great at sales and marketing you are great at building product you you know are really smart about strategy and so I think in every dimension of the company maybe it's the all hands you describe them maybe it's hackathon maybe it's just like allowing for internal use and experimentation and making that a norm but you need everybody to do it and then I think the third piece is like how do you create this compounding motion where you at a huge part of the business to be driven by AI such that you can make these very um you know kind of innovators dilemma decisions about where to invest and even how to price. Maybe that is a labs team. Maybe it is a founder with the moral authority to do things like Simon at Notion. Maybe it's an acquisition if the company is lacking the talent. But I I I do think you have to do all three of these things at once.

Chat GBT moment in robotics. Has it already happened with Whimo or do you think it'll look like something else? Do you think it can look like that given that physical AI can't just be instantly distributed through every iPhone? How are how are you thinking about you know physical AI broadly and kind of breakout moments?

I think we will see it in the next two years. uh and and Whimo uh and robo taxis. These are absolutely the first version of autonomous robotics that most people will see, but it's a very large physical world. Uh and we still don't like I I think the in certain communities in, you know, big cities in California, the acceptance speed, uh consumers are hilarious, right? They're like, I am never going to get in a Whimo. That looks unsafe. And then the second time they sit in the Whimo, they're like, "Oh, you know, maybe I'll like pull out my phone because I'm already distracted." Right? And the third time you get in the Whimo, you're like, I can't believe that this thing doesn't go any faster and like it went to this point in my driveway instead of right here. And so I I I do think we're going to have that transition very quickly. And it's a huge opportunity the way, you know, the 900 million users uh are now on Chhat GPT or more. Um but uh but I think we're going to see it in different functional areas, right? And so I sit on the board of a company called Sunday Robotics. You guys have met Tony. Uh and so whether it's them or somebody else, we're going to see them. I think we're going to see robots in the home. Uh and that'll be the first time people believe that it's possible for a robot to do general work. Uh as well as in in manufacturing and industrial uh environments as well.

That's very cool. Um, another sort of prediction, uh, Ben Thompson called Chat GBT or OpenAI like the accidental consumer company. They were very much like a Neo lab in the sense they were just doing research and then they just wound up with a consumer product. Do you think it's possible that we'll see another consumer application emerge from a Neol or something like it?

Yeah, I I'm I'm betting on it. Uh and so uh I I absolutely think that the uh chat box the ability to interact with information in sort of mostly a single turn but sometimes a conversational way uh where yes you increasingly add data sources actively is one way that consumers will interact with AI obviously but

the uh open claw was this very interesting phenomenon where the it it was a combination of a a bunch of really interesting ideas that still hasn't penetrated the mainstream consumer even though tech audiences are very excited about it. And so uh I I think that's just you know one signal of what's to come when you have longunning agents with some idea of memory and persistence and ability to do things on people's