Invest Like the Best with Patrick O'Shaughnessy
00:10:33 · Patrick O'Shaughnessy, Christina Cacioppo
[00:10:33] S0: We're just taking lower margin. [00:10:34] S1: We're taking lower margin and really concentrate on usage. That experience that I just talked through where the agent is going off and doing things, we just fundamentally believe that is a much better user experience. And we've got 25% of all of our customers in that sort of flow right now. We can get that to 75%. One, I think we'll have like much better retention, much better customer experience. And then if we need to go sort out pricing later, we should go sort out pricing later. This is a usage phase and like a deployment phase of this technology and just go nail that and you can figure out commercialization after. [00:11:09] S0: I'm also so curious, you said marking up a CLOUD token, what about marking up like a Kimi token or like something that's much cheaper? What's it been like deciding what to use, what's good enough versus cost?
00:11:09 · Patrick O'Shaughnessy, Christina Cacioppo
[00:11:09] S0: I'm also so curious, you said marking up a CLOUD token, what about marking up like a Kimi token or like something that's much cheaper? What's it been like deciding what to use, what's good enough versus cost? [00:11:20] S1: This was something we started concentrating on at the beginning of the year. Was basically evals, task specific or agent specific emails. From a product perspective, we call it all the Vanta agent and it's one thing. Of course, behind the scene, it's like hundreds of agents that all have specific tasks and use tools and have computers and that whole setup that I think everyone's converging on. But in the beginning of the year, really invested in using our subject matter experts, are the folks who like ran security and compliance programs at other companies, to write the eval frameworks and then eval their evals and using tooling for that. When we think about model swapping, it is mostly quality. It's not cost. I also think some of it to this is because we don't use as many tokens as coding. Coding is, and coding agents, lots of token use, maybe legal and like document drafting next. And then when you get to more of the decision action world, which is where we are, relative to coding, we just don't use that many tokens. So that lets us optimize for quality versus being as concerned about cost as some of the other companies. [00:12:23] S0: When you talk to other founders, of your era and of this new era, what are they doing here? Are they also eating the cost and just charging a software fee? What do you think the end state's going to be?
00:12:32 · Christina Cacioppo
[00:12:32] S1: I think everyone wants to get to outcome pricing. Everyone wants to get to that. In another life, was a microeconomics professor, like, I love this conceptually. You would get paid for outcomes and the alignment of incentives. Love it. The hard part is what are the outcomes? Support ticket closed makes a lot of sense. Actually, to that, maybe differentiate outcome and output as I understand it. In support, the ticket closed as an output. The outcome is actually my Sonos networks. But even like the ticket closed, that is relatively unobjectionable. I think a lot of companies, Ranta included, are trying to figure out what's that like neat unit of work that we can all agree on. And you can say it's, oh, it's a document drafted. It's an alert closed. It's an account off boarded. And you can kind of do that, but then there's just so much cognitive overhead because you're like, I just named three things. I could name another 47. And then our user has to be like, oh, well, if I do this with the agent, I paid this. But if I do it, know, and you're just like, no one wants that. Everyone I talked to wants to get to outcome output pricing. But it's really not clear how you get there. Charging tokens themselves really puts you up against Claude. Then people try to obscure that by charging credits, having some like credit token exchange rate y thing, which I think again, user are smart and they're like, I see what you're doing and I see you've obscured it. Fine.
00:13:54 · Patrick O'Shaughnessy, Christina Cacioppo
[00:13:54] S0: Are you similarly far along in if you compared yourself just as you did just now to companies and how you're building product and the business using these tools? What are you doing that's interesting there? [00:14:04] S1: In the vein of eight year old company and eight year old code base, it is so much easier to throw the coding agents at a small code base than a large one. And that's like a token input thing. And they're getting better and better. But one thing we've spent a lot of time also investing in in the last year, and I know Notion has spent a bunch of time in and they were like nine months ahead of us on, is modularizing parts of the code base. So that the agent can just understand smaller parts of it and understand it fully and then be really effective. Whereas if you chuck it at, again, eight year code base site unseen, It just doesn't have the context window to understand everything and will do silly things. Then you need a high context engineer to babysit it, which is helpful, but sort of invalidates some of the point. Companies have always had this problem. They've just talked about it as, oh, when we onboard a new engineer, it takes them a while to figure out the code base. Agents have made it like, great. Now we want to onboard 9,000 new engineers tomorrow. So it's just like heightened that problem. I give or feel the lower grade on that. We are like in the process of using agents and people to make the code base easier to understand too.
00:14:04 · Christina Cacioppo, Patrick O'Shaughnessy
[00:14:04] S1: In the vein of eight year old company and eight year old code base, it is so much easier to throw the coding agents at a small code base than a large one. And that's like a token input thing. And they're getting better and better. But one thing we've spent a lot of time also investing in in the last year, and I know Notion has spent a bunch of time in and they were like nine months ahead of us on, is modularizing parts of the code base. So that the agent can just understand smaller parts of it and understand it fully and then be really effective. Whereas if you chuck it at, again, eight year code base site unseen, It just doesn't have the context window to understand everything and will do silly things. Then you need a high context engineer to babysit it, which is helpful, but sort of invalidates some of the point. Companies have always had this problem. They've just talked about it as, oh, when we onboard a new engineer, it takes them a while to figure out the code base. Agents have made it like, great. Now we want to onboard 9,000 new engineers tomorrow. So it's just like heightened that problem. I give or feel the lower grade on that. We are like in the process of using agents and people to make the code base easier to understand too. [00:15:09] S0: If you to grade yourself one through 10 about, wow, this changes everything. Building this company from this point forward is going to be totally different. Is it 10? [00:15:16] S1: Oh, yeah. Like nine. Yeah. [00:15:18] S0: You are like a 10?
00:15:18 · Patrick O'Shaughnessy, Christina Cacioppo
[00:15:18] S0: You are like a 10? [00:15:19] S1: Eight, nine. I mean, I think the like principles are still the same. Everything is hard. People matter. All the tropes are still the tropes, but the way you do stuff, it should be quite different. [00:15:28] S0: Do you feel like there's a separation happening? Again, if I took your cohort of the successful a year ago era companies, do you feel there is a bifurcation where some have gotten the memo and are doing really and some are just going to Kind Bending [00:15:42] S1: of. I think there are just structural parts of some markets and I think Bending Spoons are very good at identifying these where you can put any entrepreneur in the world in that structure. This is kind of like the, like this market or founder win, Andreessen Horowitz thing, and you're like market wins. I think that is still true. And I think this AI technology shift has made what some markets that used to look really good look really bad. The best person in the world can run-in that must still run into brick wall. And I think there is a founder motivation piece where it's like you got to be in it and like push in this era. And if you don't want to do that, totally fine, but the rest of the company will not self organize to do so.