Reflections on what VC is becoming and what it shouldn’t forget.
“Data-driven”. A buzzword that’s been everywhere for the past decade. The kind of phrase that makes pitch decks look smarter, job posts seem more attractive, and strategies feel more rigorous. It lights up the brains of consultants and founders alike. But what does it actually mean, especially in VC? And why are we still talking about it in 2025?
Lately, I’ve been thinking a lot about this. Partly because I’ve seen how often “data-driven” is thrown around without clarity or depth. Partly because I’ve read some excellent reflections from people like Lawrence and Andre, both challenging and defending the idea of a “data-driven VC.” And partly because, like most of us, I now rely on AI tools every single day. So I thought: maybe it’s time to add my voice to the conversation not with definitive answers, but with questions, doubts, and a little personal perspective.
We all use AI now but do we understand what it’s doing to VC? Let’s start with the obvious. From summarizing calls to drafting investment memos, from scraping startup databases to analyzing pitch decks. We’re all augmenting ourselves with some kind of machine intelligence. I personally don’t remember the last time I wrote a cold email without an LLM polishing it. I’m probably 5x faster today than I was two years ago, thanks to these tools. But does that make me a data-driven VC? I’m not so sure.
Here’s the thing: speed is not the same as depth. Just because I can evaluate a deck in half the time doesn’t mean I’m asking better questions. Just because I can pull trend data faster doesn’t mean I understand what’s really going on in the market. Just because I can run a model to predict founder success doesn’t mean I’ve become better at identifying outliers. Sometimes, I worry that “data-driven” is turning into a shortcut to avoid using our own judgment. A way to feel safe. A way to defer judgment. Our job isn’t to feel safe, it is to be bold. It’s to make high-risk, high-conviction bets in moments of uncertainty and data rarely helps with that.
So what should Data-Driven VC mean? It shouldn’t mean automating everything. It shouldn’t mean outsourcing human instinct. And it definitely shouldn’t mean pretending that AI knows something we don’t just because it returns a chart. Being data-driven should mean knowing how to ask better questions. It should mean using data to inform (not replace) judgment. It should mean integrating new tools while staying grounded in the most important one: your ability to understand people.
I get it. Data feels objective. Safe. Scalable. But the best venture decisions I’ve seen were anything but. They were irrational. Contrarian. Founder-led, not model-approved. VC isn’t private equity. We’re not buying cash flows. We’re investing in potential founders and potential doesn’t show up in spreadsheets.
There’s a difference between using AI to accelerate your process and using it to replace your thinking. Today, there’s a tool for every part of the VC workflow: Market research? Automated. Founders scoring? Trained on historical outcomes. Due diligence? Half-done before you even meet the team. Internal deal memos? Pre-drafted. That’s not necessarily a bad thing. It means we can spend less time on repetitive tasks and more time on what actually matters. Here’s the trap: the more efficient we get, the more we risk losing the messy human parts of our job. The conversations, the doubt, the chemistry. That’s where insight lives and that’s where edge comes from.
This is still a people business. At the end of the day, I believe this job is (at its core) a human one. We invest in founders, not features. In ambition, not just traction. The best startups often start out as weird, flawed, or seemingly irrational (you know the names too, I don't need to write them down). Their pitch decks don’t always make sense. Their models are based on assumptions that will be obsolete in three months. They often pivot. They usually evolve. Their drive (their will to build something big) that doesn’t come from a spreadsheet. In fact, the best founders I’ve met would probably fail most automated screening systems. So yes, use the tools. Leverage the speed. Respect the data. But remember: the job isn’t to predict the future with perfect precision. It’s to see something others don’t and back it, early. That’s not an AI skill. That’s a human one.
You’ve heard the metaphor before: investing in a startup is like getting married. You’re committing to a 10+ year relationship. Well then, would you let AI or some data choose your life partner? Let it recommend a catering or a wedding location, sure. But the person? That’s your call. VC is no different. Use AI to scout faster, take better notes and build stronger workflows. But when it’s time to commit, when you’re looking across the table at a founder and asking yourself, “Can they do it?” that decision has to come from you. From experience. From instinct. From something you can’t reduce to numbers.
So where do we go from here? Let’s be honest. The tools are only going to get better. We’ll have models that predict failure rates. We’ll have internal GPTs trained on every deal we’ve ever seen. We’ll have investor copilots whispering insights in our ears 24/7. But let’s also be clear about one thing: The best decisions in VC will still be made by people willing to go against the grain, trust their gut, and believe in something not yet proven. So yes, long live data-driven VC but let’s also celebrate people-driven VC because that’s where the magic is. No matter how good the tools get, that magic still belongs to us.