Nick Davidov, Co-founder and Managing Partner at DVC – Interview Series

Nick Davidov, Co-founder and Managing Partner at DVC, has invested in 140+ early-stage AI startups, including Perplexity AI, Etched, and Mira Murati’s Thinking Machines Lab.

DVC is a San Francisco-based venture capital firm focused on backing early-stage artificial intelligence startups and building a community-driven ecosystem around AI innovation. Founded by Marina and Nick Davidov, the firm combines an active network of founders, engineers and researchers with proprietary AI-enabled workflows to support companies from pre-seed through Series A/B growth stages, investing $100K–$300K at the earliest stage and $1M–$3M in follow-on rounds. Its model emphasizes hands-on support, deep community engagement and leveraging AI tools alongside experienced limited partners to source deals, conduct due diligence and help portfolio founders grow — all aimed at creating meaningful impact in the AI startup landscape.

DVC has become known for rethinking venture capital from first principles. What experiences or frustrations in traditional VC inspired you to build a model that runs entirely on AI agents and community intelligence?

Primarily, problems with scaling a venture capital business. Right now, the only way to grow the business is by increasing assets under management. At some point, your motivation shifts — you’re no longer driven by values but by management fees. And then you start prioritizing deployment over returns, which is not healthy. So we’re trying to find an alternative way to scale an investment firm horizontally, not vertically. And also, we’re trying to democratize access to VC and bring in diversity this way.

​You’ve said DVC “fired its analysts and hired its LPs.” Can you walk us through how that system actually works — how do LPs contribute to sourcing and due diligence, and what tools empower them?

I’ll start with an anecdote here. When new LPs join us, we ask them to fill out a questionnaire, including the question: “Are you willing to actively help startups?” One LP — a test engineer — was curious to see what questions the questionnaire would show next if he selected “No,” but nothing happened. He was more or less our 100th applicant, who made us discover that a positive answer was set as mandatory in this question — no one could answer “No.” Amazingly, one hundred applicants before him had clicked “Yes” and never noticed.

We built an internal “social ladder” for our LPs. In DVC, every LP can obtain the title of a Deal Advisor and then grow into a Super Advisor, if he or she actively supports our portfolio companies with hiring, sales, product, and connections, contributing to startup growth — and getting a slice of carried interest in return. Also, if an LP brings us a deal, he or she becomes a Deal Captain. We made it very easy for them to contribute at every stage, and automated most of the processes.

How do DVC’s proprietary AI agents automate core VC functions such as deal sourcing, due diligence, and portfolio monitoring? Could you describe one or two workflows that have seen the biggest transformation?

At DVC, AI is the backbone of a complete rethink of how a venture firm operates. It helps us flip the traditional due diligence process on its head. Instead of reaching out to founders first, our AI agents build a comprehensive deal memo and conduct most of the preliminary due diligence before initial contact. This is possible because much of the necessary information (like company decks, market data, competitor analysis, and valuation principles) is available before the call so there’s no need to waste founders’ and investors’ time. What used to take a full day for an expensive analyst can now be done in minutes for under 30 cents.

When a startup deck hits the system — usually via an LP intro — AI parses the materials, enriches them with external sources, sorts the data and spits out a deal memo. So we can focus on what’s hard to automate — founder motivation, soft skills, culture and team dynamics.

​What are the most unconventional or surprising “signals” your AI tracks when evaluating early-stage startups? How do these differ from what human analysts traditionally look for?

Among the unusual and hard to notice by human analysts signals I’d flag recent layoffs of key team members at the startup. In total, our system tracks approximately 120 signals. Part of them helps us understand the startup traction. The second group of signals is about the startup team. The third category is focused on potential investor competition. The goal here is to predict how quickly a founder can raise a round, as the founder’s network is a significant factor.

​With over 170 LPs coming from companies like OpenAI, Meta, and Tesla, how do you coordinate this community and ensure quality input rather than noise?

When there were only a couple of dozen LPs, it was possible to manage them manually. But as their number was growing, matching the right expert to the right startup at the right time became a bottleneck. So we developed AI agents acting as “super connectors,” remembering all the details about LPs’ expertise and networks to suggest relevant introductions for portfolio companies. This makes the community aspect incredibly scalable.

And to ensure quality, we always reach out to the founders to get their feedback. For us it’s not enough, for example, to simply make an introduction to a potential company; we need to verify that it actually worked. Over time, we’ve gathered enough data to see what’s truly effective.

The fund’s thesis emphasizes core AI infrastructure and vertically focused model stacks. What segments of AI do you think are most undervalued right now — and why?

There are no undervalued segments, only underdeveloped ones — and the underdeveloped segments tend to be cheaper. Healthcare, industrial applications, robotics, and construction are lagging. There are a lot of applications, but their adoption is horrible.

​DVC has already invested in names like Perplexity AI and Etched. How does your AI system help you identify these breakout companies before they hit mainstream attention?

Most of our investors are founders and engineers who are primarily early users and adopters of AI technologies. And when they become obsessed with something, they bring it into our community. Our AI stack handles the preparation. As partners, our job is to listen to our intuition. That’s exactly how we ended up investing in Perplexity.

As AI reshapes venture capital, what aspects of investor intuition, judgment, or human connection do you believe can never be replaced by algorithms?

AI agents already outperform us at deal memos, research, and prep work. What they can’t replace is the human-to-human connection between a founder and an investor. That relationship often lasts longer than an average marriage. The founder’s path is complex, lonely at times, and high-pressure. Having a human partner is crucial — that’s what an investor is really for. And no algorithm can substitute this.

​How do you balance the advantages of automation with the need for transparency and trust among founders and LPs?

Actually, it’s the other way around — automation provides transparency, which in turn builds trust. With automation, you can clearly see which inputs were used and how decisions were made. It makes the decision-making process as transparent and analyzable as possible.

​Looking ahead, do you envision a future where most venture firms operate without analysts — and if so, what does that mean for the next generation of VCs entering the industry?

The traditional role of the analyst is inherently inefficient. The structure itself — having an analyst gather and analyze data — is prone to human error, both in data collection and analysis. Using AI simply makes the process much more efficient, so the old role becomes unnecessary. However, new roles emerge. So when we fired all of our analysts, we hired products and engineers who handle these tasks, and the human judgment doesn’t disappear — it just shifts into different functions. The role migrates to where it adds the most value.

Thank you for the great interview, readers who are interested in learning more about this VC should visit DVC.