Intern Project
Benita Besa
Duke University • Junior
Founder Market Fit Analyzer
The FMF (Founder-Market Fit) Analyzer is an AI-powered due diligence tool built for Visible Ventures to evaluate early-stage founders at the top of the deal funnel. Given a founder’s name, company name, and optional website URLs, the tool autonomously gathers public signals from across the web, Hacker News (via Algolia’s API), Google search results (via SerpAPI), GitHub profiles, interviews, podcasts, and essays, then uses GPT-4o to score them across 10 Founder-Market Fit criteria and generate a structured investment memo in seconds.
The scoring engine evaluates founders across 10 criteria including Domain Immersion Depth, Quality of Customer Insight Evidence, Pre-Founding Public Artifacts, Technical and Operational Specificity, Community Credibility in Problem Space, and Adaptability Signal. Each criterion is scored 1–10 with a confidence level (low/medium/high) that feeds into a confidence-weighted overall FMF score. The tool also produces a Visible Ventures Thesis Alignment score, explicitly reasoning about whether the founder’s company sits at a generational fracture in demographics, behavior, or broken systems, directly mapped to Visible’s thesis: “We invest where generational fractures create outsized market opportunities.” The final Visible Fit Score blends 65% FMF and 35% thesis alignment into a single number.
The full-stack application is built on a Next.js 14 frontend (TypeScript, Tailwind CSS, App Router) deployed on Vercel, and a FastAPI backend (Python, Pydantic, uvicorn) deployed on Render, with CORS middleware locking cross-origin communication between the two services. The research pipeline uses httpx to run a multi-source evidence pass on every request, querying Hacker News via Algolia’s search API, running two targeted SerpAPI Google searches scoped to interviews/podcasts and Reddit/GitHub respectively, and deduplicating all results by URL before passing them downstream.
Founder Market Fit Analyzer — UI Images
Inputs
- Founder name (required)
- Company name (required)
- Company website (optional)
- Founder website / LinkedIn (optional)
Outputs
- Visible Fit Score (1–10): weighted blend of 65% FMF + 35% thesis alignment
- Overall FMF Score (1–10): confidence-weighted average across 10 criteria
- Thesis Alignment Score (1–10): with reasoning, confidence level, and a “fracture & opportunity” narrative
- 3 Dimension Scores: Domain Fit, Execution Credibility, Character Signals
- 10 Criterion Scores: each with score, confidence badge, evidence used, and reasoning
- Top 3 Strengths + Top 2 Risks
- Data Quality Rating: limited / moderate / strong
About Benita Besa
Benita Besa is a junior at Duke University studying Computer Science and Economics (Class of 2027). At Visible Ventures, she built an AI-powered due diligence tool that translates the firm’s investment thesis into something a model can reason about, bringing rigor and speed to one of the most subjective parts of early-stage investing: evaluating founders.
“Building the FMF Analyzer gave me exposure to LLM-powered research pipelines, and how to translate a firm’s investment thesis into something a model can actually reason about. Working at the intersection of AI and venture capital showed me that the most interesting AI applications in venture aren’t about replacing analyst judgment but about getting to conviction faster.”
— Benita Besa