Menu
Pitch Fabrice
Industry : Artificial Intelligence | Digital Media | Personal
Tech Stack : Next.js (TypeScript) | OpenAI Realtime API | OpenAI Agents SDK | MongoDB | AWS App Runner | Docker
Every week, a leading venture capital firm receives pitches from dozens of founders. Each one deserves a real conversation, a thoughtful set of questions, and a clear written summary the investment team can actually act on. For a team that is also running active portfolios, attending board meetings, and evaluating late-stage deals, doing that well for every single early-stage pitch is nearly impossible.
We built Pitch Fabrice to take that weight off the team’s shoulders entirely. It is a voice-first AI platform that sits at the very beginning of the funnel, conducting real-time structured interviews with founders, asking the right questions for their stage and business model, and generating a complete investor-ready debrief report after every session. The investment team wakes up to a structured brief, not a calendar full of screening calls.
The idea of an AI interviewing startup founders sounds straightforward until you try to build one that actually works. The challenge was not just automating questions, it was creating something founders would genuinely engage with.
A pre-seed founder and a Series A founder need entirely different conversations. The system had to intelligently adapt its question sets based on each founder’s stage, not run everyone through the same script.
Founders are smart people who can tell when they are talking to a machine going through motions. If the experience felt robotic or impersonal, they would disengage, and the data would be shallow and unreliable.
The whole value of automation disappears if someone still has to write up a summary after every call. The system needed to produce a clean, structured debrief covering fundraising details, traction, and stage-specific insights, without a human.
Founders are busy. Sessions get paused, calls drop, schedules change. The platform needed to hold full context and let any interview resume without losing a single thread.
Our development team designed and deployed FabriceAI, a robust AI-powered web application tailored to present Fabrice’s viewpoints through a seamless, engaging user interface.
We designed Pitch Fabrice around a supervisor AI that thinks of itself as a director of specialists. When a founder enters a session, the supervisor reads their context and routes them to the right specialist agent, one built specifically for their stage: Early, Growth, Confirmation, or Pitch Deck Analysis.
Each agent knows exactly what questions matter at that stage and how to probe deeper when an answer needs more clarity.
Before the first word is spoken, founders upload their pitch deck. The system extracts it using vision AI, reads it, and pre-fills what it already knows so the interview can skip the basics and spend time where it counts.
After the session, the platform writes the debrief automatically, a structured report covering company summary, fundraising instrument, traction data, and insights, ready for the investment team with no editing required. Every session is saved in full in MongoDB, so a founder who needs to pause and return picks up in mid-sentence.
Founders have a genuine voice conversation with an AI agent that listens well, asks follow-up questions naturally, and covers all the ground the investment team needs.
A supervisor AI routes each founder to the right specialist agent based on their startup stage, ensuring every interview is relevant, focused, and appropriately deep.
Founders upload their deck before the interview. The system reads it using vision AI, extracts the key facts, and enters the conversation already informed.
Every session ends with a complete, structured report covering company summary, fundraising classification, traction data, and stage-specific insights, generated without any manual input.
Full session state is stored in MongoDB with complete context replay. Founders can pause, step away, and return to exactly where they left off, with nothing lost.
The investment team now processes more deal flow with less effort, and every founder in the pipeline gets a consistent, high-quality conversation regardless of how busy the team is. What previously required hours of scheduled calls and manual note-taking now runs quietly in the background, producing structured data the team can act on the moment they open their inbox.
