AI Survey Shares Insights into the VC Realm – Techstrong AI Podcast EP47

Hello and welcome to the Techstrong AI Podcast. I'm Amanda Razani. I'm excited to be here today with a partner of Costanoa Ventures.

Martina. Lao. How are you doing today?

I'm Amanda. I'm doing great. Wonderful.

Well, we're here to talk a little bit about a survey that you put out, but first tell me a little bit about Costa No Ventures. What do y'all do? So, Costa No Ventures is an early stage VC firm based in San Francisco, and we focus on B2B investments largely in the infrastructure that help companies transform.

So for example, if you are an insurance company and you wanna get more intelligent in how you're doing claims, not just more efficient, we invest in companies that build that infrastructure. Or similarly on the IT side, if you know, all of a sudden your data's been hijacked and you're needing to respond in 30 minutes and you call it department to try and get their help, how does a human insert themselves in that as opposed to just having the typical response and going through a, a chain a, uh, the typical chain of things you have get to, to actually talk to a human. How do we build technology and infrastructure, AI enabled infrastructure that helps us be way better at dealing and responding to circumstances like that?

All right, wonderful. Thank you for sharing. Well, so you recently did put out a, an AI survey.

Can you share a little bit about the survey? What, who were you surveying and then a little bit about what information you were trying to gather from this survey? Yeah, Well we are definitely in this whole new era of what we are calling AI native companies.

So they've been built in this era of large language models, which means they're building their companies differently. They're whole product backend is different. So we were trying to get a sense of, in this new era, how are companies different and how do they build differently?

How do they think differently? So an example of, one of the things we were trying to understand is, well, how many models do they use? 'cause everyone thinks, oh yeah, they just use, they use large English models.

Well, one of the interesting findings was over half of the surveyed companies, and these are all early stage, so seed, uh, seed and series A. So quite early in their, in their lifespans, over half of them were using four or more models. And so that was a really interesting result as an example.

Okay. So, um, what should business leaders take from this survey? Were there any key, um, details you wanna share from the results?

Any information that's interesting to, to business leaders? Yeah. Well, I think what is interesting is a business leader's gonna have to make a decision.

I mean, the, the entire application, what we call the application stack is being rewritten with by these AI native companies. And so business leaders will have to make decisions about companies they feel comfortable working with. So if you don't feel really savvy in this whole new world, what are the types of questions you should be asking yourselves about?

Is this a company I should trust? Are these innovators that are really leaning in and are keeping up with how much has changed? So an example of how you might use some of the data we're finding is how many models are they using in how they build their product?

It doesn't mean that there's a a set minimum, it just indicates the sophistication of the organization you might be doing business with. So an unsophisticated one will be using one model, A more sophisticated one will be using multiple because it makes them more computationally efficient, which means you get better costs, uh, or pricing. So these are things that would be invisible to you if you didn't know what was behind it.

But they're the types of questions you can ask when you're trying to make a decision. Do these guys really have good technology? Absolutely.

And were there any, um, key stats that, that you have access to that you could share from the report? Yeah, well, uh, one that I showed earlier being that more than half of those that we surveyed were using four or more models. I think what was interesting there was how many, so about 40% were using both Claude and Llama, if that's, uh, you don't know what that is.

Those are the, the new models from Meta that is open source. And Claude is, is developed by Enro. Anthropic, very well funded, but the fact that they are used by so many I think is interesting and might be interesting to you as you're trying to get savvy on how do I assess the businesses I might be working with.

And I'd say also where they are based has an impact on the talent pools they have access to. I think we were surprised, not surprised by the fact that over half of the companies surveyed chose the Bay Area as the place to to be. But I think what was interesting there was that 33 per that second time founders were 33% more likely to be based in the area area citing talent and access to investors as the primary driver.

And so again, it's, there is this whole new wave of technology and how we are building that is sweeping the startup community and for business leaders understanding how do I assess these companies and know that they are really strong in what they claim to do. These would be the kinds of things that could be indicators for people that don't live in the world of technology. Absolutely.

So this, of course, we've seen this AI technology advance very rapidly only over about the last two years. What do you see for the future of this technology as quickly as it is advancing? Well, I think that's the big thing, is we have to make sure that the companies that we are working with or that we invest in know how to keep leveraging and take advantage of and keep up with how much things change.

So it requires an amount of humility that I think many people aren't used to having in the technology world. Like I know how the te where the technology is going and I know what I'm gonna do. We're surprised all of us by how quickly things continue to evolve.

And so we just have to be ready to dive in and and adapt. And I think that's one of the biggest shifts. Like I've been in technology for over 30 years, this is a, this is a place where nobody is an expert and we're all figuring it out together.

And the people that are figuring out the fastest are those that just dive in, put their hands on the technology and try and figure it out. 'cause things are just changing so quickly. Uh, from your experience, is there an issue with, um, that kind of fear of AI from people?

Um, they're, they're a little hesitant to dive in because they don't necessarily trust AI still? I, I think it's merited. I think we get different results at different times and that that makes us decide how much we trust what we get in response.

I will say that is one shift that I have noticed in the last year is the consistent quality of the answers. I'm getting outta large language models and products that use them. That evolves very quickly.

So what I would say is it is absolutely appropriate to be skeptical. Uh, but we can't stick our heads in the sand on this one. This is one where everyone's like, even if you tried it six months ago, try it again because the models have improved since then.

And you might be surprised by what you see and learn and how quickly the models are getting better. So I would just encourage everyone, no matter how skeptical or concerned you are, to dive in and be unafraid of trying and ex and having your own experiences so that you know how to judge this technology yourself. Wonderful.

Well, if there was one key takeaway you could give our audience today, what would that be? I would say be curious. Be curious.

Don't trust others to tell you what is the best, what is the most interesting. Try it for yourself and know that everybody's learning at the same time, which is a very unique moment in technology history for all of us. All right, wonderful.

Well thanks for coming on and sharing your insights and some of the stats from your recent survey. And thank you again. I look forward to speaking with you in the future.

I look forward to it as well. Thank you so much. Alright.

And also thank you our audience for tuning in every week. Stay tuned. There's more.