This VC fund raised $75m and runs on AI agents — no analysts needed

Marina Davidova and Nick Davidov are a couple of serial founders. This, perhaps, explains why they run their venture capital firm like a startup: constantly experimenting, iterating, and making improvements. One time, their first principles thinking took them as far as eliminating all their analysts and replacing them with an army of founder LPs equipped with AI.

They started experimenting with large language models (LLMs) a few years ago, which made them take a step back and reassess the traditional structure of VC firms. 

Now they have what Davidov dubs the “infinite analyst”.

“When we can do some of the work with AI, we need to constantly re-evaluate which parts are becoming infinitely scalable,” he tells me. 

“When you scale a venture capital firm, you scale the number of analysts so you can process more deals,” he continues. “But what if you have an infinite number of analysts?”

Davidov co-founded Davidovs Venture Collective (DVC) alongside his wife in 2021 with about 50 LPs. Until now, they have been the only general partners, but alongside now launching a $75m fund for Series A and B AI startups, they’ve also brought on board Mel Guymon and Charles Ferguson to join them.

DVC’s seed fund says it has already invested more than $20m across 120 startups, including Perplexity AI, AI chipmaker Etched, and former OpenAI CTO Mira Murati’s Thinking Machine Labs.

DVC eliminated analysts over a year ago — one now acts as product manager — and has instead “hired” 170 founders-turned-funder LPs who make use of a suite of AI agents that automate traditionally manual VC workflows, assisting with deal memos and due diligence. 

The VC firm uses a mix of external platforms like Lindy, Cursor, and Lovable, alongside internally developed tools, to create a personalised stack of agents for everything from screening and call prep to job matching.

Indeed, other VCs are joining the AI and LLM bandwagon to improve sourcing and workflows. Flavia Levi, a Deeptech VC at Join Capital, commented: “We use LLMs to speed up the administrative side of investing, such as polishing language, extracting key insights from reports, and drafting investment memos… We don’t use LLMs to do due diligence. That still comes from engaging with experts, going to conferences, and reading research.”

Alex Patow, Principal Analytics Engineer at Inflection added: “We've built Kepler, an LLM-powered research platform with agents that help us collect and synthesize market intelligence in minutes instead of days… LLMs allow us to process unstructured data and automate a lot of what passed as knowledge work but was really just grunt work.”

DVC’s limited partners (LPs) come from the likes of OpenAI, Google, Meta, Microsoft, Tesla, SpaceX, and other high-performing AI companies and assist in sourcing new deals as well as helping the portfolio companies with hiring, sales, and product.

In line with the couple’s serial founder thinking, the seed fund only invests in repeat founders. It is evergreen and invests in 10 companies each quarter, and the majority of the community contributes to this because it is built for smaller cheques.

The new fund for startups further along on their journey is backed by LPs including TechCrunch founder Michael Arrington, Perplexity co-founder and CTO Denis Yarats, and Semrush founder Oleg Shchegolev.

It will invest both in companies that DVC backed at seed level and companies they missed out on the first time, including first-time founders they couldn’t get to at seed.

“The good thing about VC is, it’s actually good to be wrong,” Davidov says, “because if you’re wrong, you can only lose 1x. But if you’re right — and lucky — then you’ll get 100 or 200x so you need to try more and be humble. I can be wrong a lot of times. I personally passed on Series B of Uber many years ago.”

The new fund still has $20m left to sign – though there’s been a verbal commitment from an institution, Davidov says – but it has written five cheques with the first $55m in the fund to Perplexity’s Series B, online bookkeeping software Kick (also backed by OpenAI and General Catalyst), HR software Human, AI video startup Higgsfield and Bioptic.

Beyond just providing capital, the DVC’s limited partners contribute to the deal flow and provide a range of expertise. Plus, he says, he gets to work with his wife.

“These are the coolest people to be around,” he says. “Every time I head hunt somebody else for the community, like, for example, one of the lead researchers from OpenAI just joined, the former VP of research at Meta is joining now. And every time I, you know, get my hands on them I’m like ‘Oh yes, that’s a really cool gem in my collection of smart people, smart and good people.’”

“When you work with nerds, with engineers that you know share the passion,” he adds, smiling like a giddy kid. “You wake up and it’s like, whoa.”

Community is clearly a big part of DVC. 

There’s a Halloween party coming up with its LPs — an opportunity for some of the founders to share horror stories from the startup world — and every April it hosts its ‘Alice in Wonderland’-themed gathering, Mad Hats. Last year, there were about 600 people, and the night ended with a small fine from the San Francisco fire marshal. So next year, a bigger venue is probably in order.