Large language model companies like OpenAI and Anthropic have largely stolen the show among investors this year, but, as you may have noticed, VCs have increasingly been looking into more niche, so-called AI copilots that use AI to make some processes more efficient for humans—be it in healthcare or banking.
As a small sampling: Corti, a Copenhagen-based startup using AI to assist healthcare providers in assessing patients, just raised a $60 million round in late September; Truewind, an AI assistant for companies’ accounting and CFO functions, raised $3 million in seed funding around the same time last month; and atla recently pitched itself to investors during Y Combinator’s summer Demo Day as a startup building AI assistants to help in-house lawyers answer legal questions. Other more established companies, like PathAI, are essentially doing the same thing: PathAI’s cofounder and CEO Andy Beck told me earlier this year that their AI technology is able to analyze and interpret tissue samples and assist pathologists. For the human pathologists, in the future “their job is going to be to use their judgment, which will not be replaced by these systems, to interpret the output of the AI system.” And there are plenty of other examples.
It’s a theme that Samir Kumar, a cofounder and general partner at the new VC firm Touring Capital, is particularly looking to invest in. He recently told me he’s eyeing companies that are building the “copilot for X.” “These are wholly new companies that are reimagining a user experience or workflow from the ground up with the notion that the fundamental ethos is: Human plus AI assist working together to make a workflow more efficient,” Kumar said. It’s “more [generative] AI native, but still focused on solving a particular use case.” Touring is currently raising their first fund, targeting $300 million, which will focus on software and AI companies.
One of the big areas these so-called AI copilots (in some cases referred to as vertical AI companies) are focusing on is financial services. Rex Salisbury, the founder and general partner at Cambrian Ventures and a former fintech partner at Andreessen Horowitz, told me that while he hasn’t invested in any of these companies yet, he’s seeing a lot of examples. “There are AI copilots for everything: For medical billing, for compliance at banks, for customer support at banks…for accountants, for tax firms, for law firms. [For] any kind of paper-based process—of which there are a lot in financial services—there are AI copilots,” he says, adding that the most “crowded” area he’s seeing is with AI financial analysts.
Where Salisbury is interested, though, is in companies led by people who have been working in the industry for a long time and understand who their customer is. “They have that relationship with the end buyer, who they've probably sold to before. Because if you want to build, say, an AI customer support widget, and you want to sell that into banks, you probably need to know FIS, Jack Henry, Fiserv,” he says. “Those are the kinds of people where it's like, 'Okay, I know you can build this, because the toolset is out there and you're sophisticated. And...you already know all [these] complex puzzle pieces of what the go to market looks like.' And that's the kind of deal that I like to underwrite as a pre-seed and seed investor.”
Some VCs, though, think there may be some challenges for this host of new AI copilots. “I think they all are underestimating how complicated that is in reality, especially in an enterprise setting,” Andreas Goeldi, a partner focused on AI at Europe-based firm b2venture, said. “We are now seeing the first startups really thinking much more deeply about that: What does it really mean to deploy a large language model inside a bank? What about security and audit trails and all this other fun stuff?” Goeldi believes that European companies have a particular advantage: “We think that Europe is actually in a very good position to deal with that stuff because many companies here—and also university research—are dealing with that kind of problem,” he says.
Specifically, Goeldi is looking into companies like Kodex AI, which describes itself as the AI copilot for the financial industry; CYQIQ, focusing on consultants; and Frontnow, which assists with customer experience for e-commerce. “There is certainly a place for generic copilots like Microsoft's in general office productivity, but the vertical use cases need to be much more tied to the specific needs of an industry,” Goeldi believes.
However, as with the countless AI startups that have popped up over the last year, the key will be to figure out which of those AI helpers will actually be winners. It’s “very hard for me as a pre-seed investor to decide which of the 10 pre-product, pre-revenue teams—raising too much money at too high of a price—is, like, a good place for me to invest,” Salisbury added.
Perhaps these VCs will end up using some kind of copilot to help them figure it out.
See you tomorrow,
Anne Sraders
Twitter: @AnneSraders
Email: anne.sraders@fortune.com
Submit a deal for the Term Sheet newsletter here.
Joe Abrams curated the deals section of today’s newsletter.
