00:24:38 · Guillaume Moubeche
[00:24:38] S1: Yeah. I agree 100%. I think like if you ask people interview questions and it gets too much about, you know, like, presenting something that they could do with AI, it's a bit annoying. However, if you ask them, like, what's one of the toughest problem you've solved recently? And you try to dig in, you see usually a lot more, and if they have built a project, like, even better. To your point, I was... I wanted to go back a little bit, like, on on on AI and how it's evolving. Because for me, you know, when you look at ChatGPT at first, you had, you know, all these tutorials about, like, prompt engineering, and everyone was setting course about this and trying to make money on it, but it felt even like it was gonna be like a new job, and people were hiring prompt engineer at, I don't know, like 500 k per year or something like this. And eventually, it totally disappeared because it was like a bug and not a feature. I feel like for the the models, like, it's the same whenever you you look at the the quality of intelligence that you select. What do you think is gonna change a lot in AI, like, and also with agents in the coming years that we're not seeing right now?
00:25:46 · George Bonaci
[00:25:46] S0: It's a good question. I think we all kind of see the end state, but how we get there is is very is very uncertain. Yeah. Kinda going back to like my Internet example from earlier. I can share with you what I what I think at least. I think that most people's roles that are very dependent on high trust relationships or creativity, like deep deep like subject matter expertise that does not have a clear optimization function. Those are going to stay like with people. I think everything else, the actual tactical work of like clicking buttons in a platform or whatever it might be, gets commoditized with AI. So if that's the case, it's like most people's jobs, and I obviously think about this like a lot more for for marketing than other rules, but it's probably applicable. So then most people's jobs become like, how do you find an edge? How do you find an advantage? How do you be creative? How do I build a personal relationship with someone? I think that's probably actually the work that most people want to be doing. I don't think there's anyone on my team where I'm just like, hey, go do the same task in a platform clicking the same series of buttons every day for like the next three years. They'd be like, yeah, sounds great. Like, no one wants to be doing like the tasks. They want to be like doing like intellectually interesting work. And so I think that's what AI frees up a lot. Going back to the idea that it really just like amplifies the the amount of information that you give it. I think it also like amplifies whatever creativity you may have or amplifies whatever reach you might wanna get with tactics that you had an idea about. So it's a giant force force multiplier. It's a it's a giant lever. So I think everyone's roles change, but exactly how they change, I can't quite say. I all... I was one of the folks that thought it's like, okay, we'll build like this agent operator function and they're gonna build a bunch of automations and then people can move on and do something completely different. And it turns out that like, yes, you can do that. But that doesn't mean that like, oh, the person that's running paid search all of a sudden can go and do like paid social. It means like, oh, now I can do this long tail of crazy stuff that I was never going to prioritize within paid search and they can just go deeper and deeper and deeper.
00:27:31 · George Bonaci, Guillaume Moubeche
And I actually think that's probably like more accretive to the business because it means then that like there's lower risk. That person doesn't need to go learn like a completely new skill set or go try something that is incredibly high risk. It means that they can actually become a deeper subject matter expert and actually get like insanely higher alpha from like this channel that the average company or the average person running that channel would not have time to do because they're still clicking buttons in a platform. So So I think it probably makes subject matter experts... Deeper subject matter experts, and that's probably where you distinguish great talent from okay talent. So again, I think I think AI becomes like an amplifier. It's a big lever in every way. [00:28:06] S1: Yeah. I agree. The only question I have for you is on... To build up on that topic is whenever you look at entry level jobs or like people, you know, who are just getting starts, because I feel like not domain experts can really like dig in a lot more on what they know and they can become, you know, like the 10x profile are becoming 100x. However, for a 10x profile to exist, you know, they had to start eventually at a job and become a 10x profile, and it takes time, anything. And I feel like the entry level job, to be honest, most of the time can be done by an AI or an agent. So how do you think that over time, the... We're we're gonna be able to actually, like, keep having the entry level job, train them, and not just replace all the the task with the agents?
00:28:54 · George Bonaci
[00:28:54] S0: Yeah. The answer is like, I don't really know. I've flip flopped on this quite a bit. I'll I'll give you both points of view, but, like, I have flip flopped. And the the only thing I really can say is is Ramp published some really interesting research maybe a couple months ago that showed that the top adopters of of AI in terms of companies hire more junior people than companies that adopted AI less. And and that's very counterintuitive. It's probably a different flavor of Jevan's paradox, to be honest, but I kind of buy it. Like, you want for... And and it's probably for a couple reasons. It's probably because number one, they're more open minded, and they're more able to adapt as the technology is changing so so quickly. I think we've all felt like we're falling behind because things change, like, weekly. So I think if you're early in career, you have, like, fewer frameworks that you need to break or or or fewer priors that you need to overcome. And so, like, you're probably, like, able to adapt to these changes more easily. So maybe that's one reason why these companies are hiring more early in career folks. I think number two is you're probably more likely to be AI native if you're relatively early in career versus like late in your career. It's just like you grew up with it. So maybe that's number two. Now having said all of that, I I... I've kind of always had the philosophy that like you really shouldn't fully automate with AI or otherwise. Like you shouldn't automate something that is a path for a junior person to become like a mid level person and a senior person. Like you need to have build those talent pipelines and those learning pipelines. I shouldn't even call it talent pipelines. There's learning pipelines. Otherwise, like, you're really just screwing yourself over as a business in the future. And that is both specific to like these subject matters. Like, I'm not going to fully automate Google Ads because I need a Google Ads expert in the future. And we need to understand how the systems work and we need to understand like where they break and what their limitations are. But I'll automate the things that are not a competitive advantage and do not need to be learned. Like, there's still a role for mental math even though we all have calculators.
00:30:36 · Guillaume Moubeche
[00:30:36] S1: Yeah. Yeah. I agree. I agree 100%. Something I I wanted to to know also because you've looked a lot at, you know, like, where people are spending time and money and how exactly it splits and evolves. A few months ago, like, mean, earlier this year, a lot of people started talking about Saaspocalypse, like, the end of software as we know it. We've seen, you know, like all these public companies stock price, like, going down, getting, like, shrank, etcetera. So from your perspective and from what you're seeing at Ramp, like, do you see companies actually spending less on software, or do you see them still spending the exact same amount of money?
00:31:12 · George Bonaci
[00:31:12] S0: That's a great question for Ara, our economist, actually. He should he should write a post on this if he hasn't already. But the best example is is I remember when Claude Design first launched and the markets, I think, massively overreacted against, like, Figma as a stock. Yeah. And I was like, this is crazy. Anyone that's used Figma knows that, like, the, like, the overlap between these two concepts is very low. So I bought a bunch of Figma stock. I was like, this is... This doesn't make any sense to me. And I think that's true in general. I think there's a lot of hype around, like, how AI is going to destroy jobs, destroy industries, destroy companies, whatever it might be. But the reality is is like number one, I think the technology diffuses much more slowly than we expect. We are all like... The fact we're having this conversation demonstrates how early we actually are to this is I think number one. Then number two, I think like, especially with these really big paradigm shifts, you really don't know how it's going to play out. Like the Cloud Design Figma, like interact... Like micro example is is a great one because like the market was wrong. And I think the markets are generally very right when there is like a steady state of historical data that you can look at and like make some informed decisions on. And that that is not the case right now because the technology is changing so quickly and this is not something that most of us have been through in our lifetimes. And so humans are bad at forecasting is maybe what I would say. They're bad at predicting the future. I So wanna try to predict the future, but I do not think that things will go the way we think they will go.