Project 07 · Case study
AgentVC — AI Investor Interviewer
AI-powered pitch practice platform where founders rehearse against a simulated investor. Live demo shipped, insights from 50+ real investor conversations baked into the feedback logic.
Case study
The problem
As a first-time founder I had no idea how to approach investors. Every guide I found was abstract ("tell a story", "show traction") without the specific patterns that make a VC lean forward versus lean back. I wanted to build a training ground where founders could get that reaction in advance.
The approach
Next.js frontend, OpenAI API for the interviewer. I spent months cold-messaging VCs, angels, and founders who had raised to extract specific patterns: what questions they repeat, what framings they reject, what they say yes to. Those patterns became the system prompt and the evaluation rubric. The AI doesn't just ask generic questions; it probes the same way real investors do.
What worked
The live demo is shipping and usable. Feedback is specific enough that users tell me it surfaced issues their decks had that they hadn't noticed. Several have used it in real fundraising prep.
What I'd do differently
I overcomplicated the conversation model early. What users actually want is a 10-minute mock call that spits out three concrete things to fix, not a 40-minute guided coaching session. The product got better the more I cut.
More detail
A platform where first-time founders can practice pitching against an AI interviewer modeled on real VC behavior. Built after I spent months struggling to figure out what investors actually look for. Next.js + OpenAI under the hood.