Venture Capital in the age of Artificial Intelligence
The world of venture capital from conception has evolved alongside technology, but we currently see a seismic shift in how investors discover and evaluate startups.
From Andrej Karpathy's recent Y Combinator talk, it became clear that the arrival of Software 3.0, meaning software driven by large language models and AI agents, is rewriting the investor playbook.
How venture capital has evolved
He identifies four groups of venture capitalists shortly described as below:
Old School
This group still makes up about half the market. Their approach is built on relationships, intuition, and closed networks. They attend conferences, rely on referrals, and often believe their access to a unique pool of founders gives them an edge. Data barely factors into their decision-making.
Software 1.0
These investors have started to use data, but it's mostly structured and filtered by human-designed rules. For example, they might look for companies in a hot sector that are growing their headcount quickly. It's systematic but still hinges on the idea that the perfect ‘rule’ will unearth the next winner.
Software 2.0
Here, the process moves beyond simple rules. These firms use machine learning models to analyze a wide range of company features and performance metrics, such as future funding rounds or unicorn status. While this approach is more sophisticated, it often misses the mark on qualitative factors, like founder vision or market nuance, that seasoned analysts intuitively understand.
Software 3.0
The newest wave is all about the use of agentic tech powered by large language models. These agents can answer complex, qualitative questions: What's the real size of this company's addressable market? How defensible is its competitive moat? Is the founder truly equipped to scale? Far from replacing human judgment, these tools add a powerful new dimension of insight, one that's rigorously tested and always human-verified.
Why this matters
The sheer explosion in startup activity in part thanks to artificial intelligence, has made traditional evaluation methods unworkable. With tens of millions of new ventures launching each year, even the most diligent investors are drowning. Manual review is not only slow and expensive, but rife with bias.
A staggering 85% of VCs confess to relying on gut feeling, and junior analysts frequently overlook critical cues that seasoned partners instantly spot.
Kuanta is built for this new era. By automating key analyst functions, we enable investors to evaluate thousands of startups with the depth and nuance of an entire team, in a matter of minutes. Our proprietary models dig through unstructured data, uncover hidden gems and red flags, and deliver transparent, research-backed assessments in a fraction of the time.
With a database of over 3.7 million startups, and a reliability engine that blows traditional tools out of the water, we deliver unrivaled speed and undeniable trust. Whether you want to empower your analysts or automate the first pass: we bring the power of Software 3.0 to your investment process.
Kuanta evaluates startups across 19 industry verticals, each with its own scorecard, its own warning signs, and its own definition of what good looks like. We scout across 3.7 million validated profiles using semantic matching that finds companies by what they do, not by what label they carry. We built it this way because the old way stopped working. Curious how we can revolutionize your deal flow?
Book a demo with us today.
