PivotVault — AI-Powered Startup Failure Intelligence Platform

Team CodeRegime (KH024)

Founders use PivotVault to learn from failed startups, validate ideas, scan risks, and make smarter decisions before building or raising money. Over 90% of startups fail by repeating known failure modes (unit economics collapse, premature scaling, lack of PMF). PivotVault provides defensive intelligence powered by 419+ verified post-mortems, pgvector semantic search, and Google Gemini RAG.

---

Repository Structure

KH024-CodeRegime/
├── README.md                           # Project documentation & execution guide
├── LICENSE                             # MIT Open Source License
│
├── src/                                # Project Source Code
│   ├── frontend/                       # Vite + React 19 + Tailwind CSS platform
│   ├── backend/                        # Node.js + Express + Prisma + BullMQ API
│   ├── ml-services/                    # Python FastAPI IdeaScore ML scoring engine
│   └── pivotvault-landing/             # Standalone landing showcase
│
├── docs/                               # Documentation & Architectural Artifacts
│   ├── project-documentation.pdf       # Formal project presentation & report (3 pages)
│   ├── architecture.png                # Full-stack system architecture diagram
│   └── other-diagrams/                 # Pipeline & Entity-Relationship diagrams
│       ├── rag-agent-pipeline.png      # RAG forensic retrieval workflow
│       └── database-schema.png         # PostgreSQL + pgvector relational schema
│
├── screenshots/                        # High-Resolution UI Demonstrations
│   ├── screenshot-1.png                # Startup Failure Archive (419+ Startups)
│   └── screenshot-2.png                # AI Risk Scanner & Autopsy Matchmaker
│
├── data/                               # Canonical Forensic Datasets
│   ├── README.md                       # Data dictionary, taxonomy & sources
│   └── seed.json                       # 419+ verified startup post-mortems
│
├── requirements.txt                    # Python environment dependencies
├── package.json                        # Monorepo workspace scripts
├── vercel.json                         # Vercel deployment configuration
├── render.yaml                         # Render backend web service configuration
└── .gitignore                          # Git exclusions

---

Quickstart Guide

1. Root Workspace (Frontend & Backend)

Install all dependencies across the monorepo:

npm run install:all

Start the frontend development server:

npm run dev
# Or: cd src/frontend && npm run dev

Build the production bundle:

npm run build
# Or: cd src/frontend && npm run build

---

2. Backend Service (Node.js + PostgreSQL + pgvector)

cd src/backend
npm install
cp .env.example .env

# Generate Prisma Client & Push Database Schema
npm run prisma:generate
npm run prisma:push

# Seed Canonical Failure Records & Run Direct Ingestion
npm run prisma:seed
npm run ingest:direct

# Start API Server
npm run dev

---

3. ML Scoring Service (Python FastAPI)

cd src/ml-services
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

uvicorn main:app --host 0.0.0.0 --port 8001 --reload

---

Core Platform Features

• Startup Failure Archive: 419+ verified post-mortems indexed with failure scores, capital burned, and root causes across 15 industries.
• AI Risk Scanner: Automated model audit cross-referencing startup hypotheses with vector embeddings to compute an objective risk score (0-100).
• Historical Autopsy Matchmaker: RAG-retrieved top 3 nearest failed ancestors via cosine distance over 768-dim embeddings.
• Hall of Ghosts: Conversational founder personas reconstructed from public testimony to interrogate historical decision points.
• Pitch Deck Autopsy: Slide-by-slide diagnostic identifying premature scaling, missing unit economics, and unverified TAM assumptions.
• Macro Failure Heatmap Matrix: Failure mode correlations and capital loss patterns across 7 major tech sectors.

---

Verification & Documentation Assets

• Complete technical documentation is compiled in docs/project-documentation.pdf.
• System architecture diagram: docs/architecture.png.
• RAG pipeline workflow: docs/other-diagrams/rag-agent-pipeline.png.
• Database schema: docs/other-diagrams/database-schema.png.
• UI Screenshots: screenshots/screenshot-1.png and screenshots/screenshot-2.png.
• Dataset documentation & data dictionary: data/README.md.