Inside the Rise of AI in Venture Capital Operations

January 9, 2026

By The Artemis Fund
The Artemis Fund believes technology can create prosperity for all. With offices in New York, Texas, Massachusetts, and Nevada, Artemis leads seed rounds for companies creating resilient families, individuals, and businesses across the US.
Saanika Gupta is a Venture Fellow at The Artemis Fund, where she focuses on enabling AI across fund operations and internal workflows. In this interview, she shares a pragmatic view of where AI is already creating leverage in venture capital, where human judgment still matters most, and how small, non-technical teams can adopt AI without getting lost in the hype.

Get The Artemis Fund’s stories in your inbox.

Q: What's your background? How and why did you start working in the intersection of AI x Venture?
A: I worked in startups for a few years and spent a year in corporate finance. They were two completely different environments that taught me how to operate in fast-paced, ambiguous situations and in more structured spaces that still demand creativity. During that time, I stayed close to venture through the ESG fund at Northeastern and by leading Girls into VC.
I’ve seen that corporate finance and venture have both run on legacy systems and traditional processes for a long time, and only recently have they started to seriously modernize. Seeing the whitespace and the opportunity to rethink how work gets done is what pulled me back into building and pushed me toward the intersection of AI and venture.
Q: What have you learned about how small, fast-moving teams can start using AI — especially when they don’t have in-house technical talent?
A: I’ve learned that non-technical teams can move surprisingly far with AI, but they need a basic technical understanding to evaluate the tools they’re adopting. The trade-offs matter: time, cost, and how much the team actually wants to learn.
Even if you outsource or use an external tool, you still need enough technical context to understand how things fit together. The fastest way to build that is by creating a small workflow yourself or talking to people who’ve shipped similar systems. It doesn’t require an engineering background, but it does require hands-on exposure so you can separate what’s real from what’s overcomplicated.
AI can also help you decide where to use AI. It’s good at pointing out non-obvious opportunities if you give it the right context. That’s why I think the mindset shift is to treat AI like a teammate, not a tool. If it gives you something off, don’t throw it away. Give it feedback, add context, and guide it the same way you’d guide a peer. And because hallucinations are real, grounding it in high-quality knowledge sources is important so it has something accurate to reason from.

Q: How can non-technical teams evaluate and adopt AI tools without getting lost in the noise? What’s your framework for separating what’s genuinely useful from what’s just hype?
A: Non-technical teams shouldn’t start with AI, they should start with their actual work. The fastest way to cut through the noise is to map the core workflows you already run, the tools your team relies on, and the specific outcomes you need to hit. My framework is to start with one real bottleneck. Look at tools that integrate into your existing stack so you aren’t creating parallel workflows, and define the outcome you want improved.
AI is there to solve problems, so before you look at any features, you need to understand the problem you’re trying to solve. Otherwise it’s easy to get pulled into things that look impressive but don’t actually improve anything. And once you understand the bottleneck you want to address, AI becomes a great way to augment your creativity in solving that problem.
Q: Many funds are experimenting with AI for research and analysis. What area of VC ops are low hanging fruit for AI tools? Are there any that are harder to tackle than you might think?
A: The easiest wins are the workflows that follow predictable patterns and rely on structured or semi-structured data. Sourcing and research augmentation is a good example because the inputs are public and the output doesn’t need to be perfect on the first pass. It saves time and increases surface area, but it still needs human judgment.
The hardest areas are the ones that rely heavily on human judgment: real diligence calls, pass vs. pursue decisions, portfolio support that depends on trust, and pulling together messy internal data. All of these share the same issue: an absence of clean ground truth. Venture decisions depend on ambiguous inputs, inconsistent signals, and context that lives in people’s heads, not in structured systems.
The way to handle this isn’t to replace judgment, but to create pockets of high-quality, trustworthy reference data. Standardize data flows, improve CRM hygiene, capture partner reasoning in more structured ways, and use AI to surface patterns instead of final answers. When inputs improve and decision boundaries are clearer, AI becomes a more reliable complement to human judgment across venture decision-making.
Q: AI innovation is moving incredibly fast — what are the most underappreciated opportunities you see right now for founders building in this space?
A: The most underappreciated opportunities are in categories where AI adoption is already happening, but it’s not publicly visible. A lot of attention goes to what’s easy to demo, but the stronger signals often sit in day-to-day workflows where teams are already getting real value from agents. A clear area is operational work with reliable ground truth. DevOps, security, logistics, and finance operations all produce clean, machine-verifiable outcomes, which makes them well suited for agentic systems. These teams care about accuracy and speed, and they already have the data structure to support both.
The mid market is another space that doesn’t get enough attention. Mid-sized companies adopt automation quickly because they feel resource constraints more directly and their workflows tend to be more repetitive. You can see meaningful traction in categories like spend management, claims processing, and reconciliation. There’s also a large opportunity in workflows that sit on top of legacy systems. Many institutions can’t replace their core systems, but they still need ways to modernize the work around them. Agents that integrate on top of existing infrastructure create real value without forcing a full system overhaul, and that’s a practical path for many sectors.
Q: What are your favorite AI tools right now? 
A: My favorite AI tools right now are Claude, OpenAI’s deep research and agentic mode, Tavily, and Miro. I also like Runway for its short-clip video generation and Canva AI for similar lightweight creative work.
Q: How do you stay on top of emerging trends, use cases, and tools in the space?
A: I stay on top of emerging trends by talking to peers and mentors who’ve used AI across different industries. Those conversations give me real-life perspectives that I don’t get from articles or Substack posts, and they’re the fastest way for me to see what people are actually using in practice, and where the real whitespaces are.
We believe once in a generation companies will be built by unexpected founders. If that sounds like you, pitch us here!

Newsletter Signup