Editor’s note: This article is part of an ongoing series in which Crunchbase News interviews active investors in artificial intelligence. Previous interviews and highlights with Felicis, Battery Ventures, General Catalyst, Bessemer Venture Partners, Accel, Insight Partners, Index Ventures, Sequoia Capital, Section 32, M12, Sapphire Ventures, Bain Capital Ventures, Menlo Ventures, Scale Venture Partners Read stories from 2023.
Two years into the artificial intelligence wave, it’s helpful to talk to early-stage investors in the space to get perspective on what it’s like to invest in generative AI startups from the beginning. I thought so.
To find out, we spoke with Costanoa Ventures founder Greg Sands and general partner John Cowgill. The Palo Alto, California-based firm invests in pre-seed, seed, and Series A, with a fifth fund of $275 million (its largest ever) and a third opportunity fund of $119 million to invest in winners. Both were completed in September. .
Since raising its first $100 million in 2012, the company has built a practice of enterprise cloud-based services powered by data and analytics.
Costanoa’s current focus is on applied AI and AI infrastructure, and B2B fintech. Generative AI “allows us to solve problems that we couldn’t solve two years ago,” Sands says.
Costanoa also focuses on cybersecurity – Because AI can increase threat instances and improve security operations — and national security. cultivated by the company Vannevar Institute In 2019, before defense technology and AI were as widespread as they are today.
New features never imagined before
Recent advances in AI have made written language easier to digest, consume, organize, and even reason about through natural language processing. Computer vision allows you to evaluate and understand images. When you combine these technologies, Sands said, you get very interesting and complex outputs.
go vertically
“It literally solved most of the problems that that generation of technology could solve,” Sands said of the SaaS revolution of the past 20 years.
The company believes that generative AI is now opening up vertical opportunities that may have been too small in the past.
“The challenge with vertical SaaS has always been that instead of building something that applies to all verticals, you’re limited to one vertical and the overall addressable market is smaller,” Cowgill said. .
The company is looking closely at each industry to understand how big they will become with the rise of AI.
Costanoa also supported this trend. Aquabiteprovides computer vision for fish farming. also invested in Beacon AI For pilot support in the aviation field, and force metricsa service that provides situational awareness to first responders by integrating and presenting data. He also secretly invested in a construction permitting company. That may seem narrow, but there is a huge opportunity in owning the end-to-end workflow in that space.
Bullish on native AI
The big question for investors is: Will this vertical opportunity benefit existing mid-sized portfolio companies that have embedded APIs? Or is it more like the era of cloud and mobile, where it was difficult for incumbents to reinvent their business models?
“I’m more bullish on AI-native application layer companies than I was two years ago,” said Cowgill, who has seen firsthand how difficult it is to integrate AI and where the value lies. said.
The company has several late-stage vertical SaaS companies working on AI integration. “The rate of integration of AI is going to be even slower,” Cowgill said. “They may actually end up offering different services, different products, different business models.”
“The real value of AI comes when you can own the end-to-end workflow,” he said. “This is the idea that the agent goes from just applying the AI, pointing to something and saying, ‘Search with AI, summarize with AI,’ to own the work that the AI does. It’s incredibly difficult to get an agent to work for you. ”
The nuances of picks and shovels
Costanoa has made several bets on AI infrastructure. The company invested in Delphina, the creator of an AI junior data scientist to help the team. It also supports Rerun.io, an open source visualization stack for multimodal data including audio, images, and video, and OpenPipe, which is used to fine-tune models.
Still, Sands points out that model management infrastructure in general is expected to grow 18 to 24 months in advance, in part due to the fact that companies in the applied AI space are not managing their models themselves but are purchasing them from hyperscalers. I pointed out that it hasn’t grown as much as it used to.
“What’s important”
However, some things remain the same.
In a 2018 blog post, “What is Applied AI?”, Sands said that AI does not replace the need for “good product management” and that “data is just as important as algorithms.” Ta.
Sands said that throughout the company’s 12 years, the company’s focus has been on the eternal principle of “finding outstanding talent and devoting time, energy and talent to them to maximize their chances of success.” Ta. “Choose what’s important: both important technology and important problems.”
Related Crunchbase Pro list:
Stay up to date on recent funding rounds, acquisitions, and more with Crunchbase Daily.
