Week 3: Building the Framework

Quick recap of last week: I went on-site for the first time, built out a research spreadsheet of early-stage startups across Bright Ventures’ three core verticals, and continued refining my investor survey. This week, I dived into my biggest project.

This week I was on-site at Bright Ventures working on what I’d say is the core deliverable of my project so far: a side-by-side comparison framework of traditional due diligence versus AI-native due diligence. The goal is to map out how the questions VCs ask have shifted now that so many startups are built on AI models, APIs, and proprietary datasets.

To build it, I organized the framework into five categories: Product & Technology, Team & Talent, Market & Competitive Moat, Business Model & Unit Economics, and Risk. For each one, I researched what investors asked before AI-native companies dominated deal flow, what they ask now, and why the frameworks has changed. I did my research and through that, I built my framework based off articles I found.

For my sources, I pulled from public writings by firms like a16z, Sequoia, and First Round Capital. A lot of their work touches on questions like whether a startup has a real AI moat, whether the product is AI-native (The AI is adding value to the startup) or an AI wrapper (the AI is built on top of an already existing LLM, essentially making it so that they don’t own any proprietary data). This turned out to be the important angles for filling out the framework.

The biggest takeaway so far is that the core goals of due diligence haven’t changed; Investors still want to know if a company can build something scalable in a big market with the right team, but the way you answer those questions looks completely different when the product is built on AI. This gap is exactly what my research question is trying to capture.

Next week I’ll finish up the framework and write a short narrative summary of the most important shifts. More updates soon!