VC Intelligence Interface with Live AI Enrichment

A production-style SaaS interface that simulates how venture capital firms discover, analyze, and track startup companies — built as an intern take-home assignment.

The system follows a structured workflow:

Discover → View Profile → Enrich with Live Data → Derive Insights → Save to Lists

It uses a mock company dataset for fast discovery, real-time enrichment from public websites, and AI-powered structured extraction to surface actionable intelligence.

---

Live Demo

🔗 Deployed Application

---

Key Features

• Search & Filter — Browse companies by industry, stage, and location with sortable columns and pagination
• Company Profiles — View company metadata, business signals, and private notes
• Save to Lists — Organize companies into custom lists with JSON/CSV export
• Saved Searches — Persist and re-run search queries across sessions
• Live Enrichment — Trigger real-time AI analysis from the company's official public website
• Structured Output — Receive a summary, bullet points, keywords, and derived business signals
• Source Transparency — Every scraped URL is logged with HTTP status, success flag, and timestamp
• Local Caching — Enrichment results are cached for 10 minutes to reduce redundant API calls

---

Enrichment Pipeline

Frontend (React)
       │
       ▼
POST /api/enrich
       │
       ▼
Input Validation + SSRF Protection + Rate Limiting
       │
       ▼
Cache Check ─── HIT ──→ Return cached result (<100ms)
       │
      MISS
       │
       ▼
Public Page Fetch (/, /about, /docs, /blog, /careers)
       │
       ▼
HTML Sanitization (Cheerio)
       │
       ▼
AI Structured Extraction (Gemini 2.5 Flash)
       │
       ▼
Schema Validation (Zod)
       │
       ▼
Cache Store → JSON Response → UI Display

Key design decisions:

• All enrichment runs server-side — API keys are never exposed to the browser
• Only official, public website HTML is fetched — no access controls are bypassed
• A fallback crawl strategy attempts multiple paths (/, /about, /docs, /blog, /careers) to handle sites where the homepage alone may not contain enough content
• Pages with fewer than 300 readable characters are recorded but excluded from analysis

---

Tech Stack

Layer	Technology
Framework	Next.js 14 (App Router)
Language	TypeScript
Styling	Tailwind CSS
Icons	Lucide React
Backend	Next.js API Routes (Serverless)
HTML Parsing	Cheerio
AI Extraction	Google Gemini 2.5 Flash (@google/genai)
Validation	Zod
Persistence	localStorage
Deployment	Vercel

---

Security Considerations

• API keys are stored server-side only and never included in the client bundle
• Only public data is used — the scraper fetches standard HTML over HTTP/HTTPS
• No access control bypass — robots.txt-protected or login-gated content is not accessed
• SSRF protection blocks requests to private IPs and cloud metadata endpoints
• Cross-domain redirects are blocked to prevent data poisoning
• Rate limiting (5 req/min per IP) prevents abuse

---

Setup Instructions

1. Clone the repository

git clone https://github.com/LavKumarShakya/VC-Intelligence.git
cd VC-Intelligence

2. Install dependencies

npm install

3. Configure environment variables

Create a .env.local file in the project root:

GEMINI_API_KEY=your_gemini_api_key_here

The API key is read server-side only and is required for the enrichment endpoint.

4. Run locally

npm run dev

Access the application at http://localhost:3000

5. Build for production

npm run build

---

Screenshots

Company Discovery

Company Profile without Enrichment

Company Profile with Enrichment

Lists Management

Saved Searches

---

Repository Structure

VC-Intelligence/
├── src/
│   ├── app/
│   │   ├── api/enrich/       # Enrichment API route
│   │   ├── companies/        # Company list + profile pages
│   │   ├── lists/            # Lists management page
│   │   ├── saved/            # Saved searches page
│   │   └── layout.tsx        # Root layout with AppShell
│   ├── components/
│   │   ├── companies/        # CompanyTable, CompanyHeader, EnrichmentSection
│   │   ├── layout/           # AppShell, Sidebar, Topbar
│   │   └── ui/               # Button, Badge, Toast, Skeleton
│   ├── data/
│   │   └── mockCompanies.ts  # Mock company dataset
│   ├── lib/
│   │   ├── server/           # scraper, extractor, validator, security
│   │   ├── storage.ts        # localStorage utility
│   │   └── utils.ts          # Shared helpers
│   └── types/                # TypeScript interfaces
├── .env.local                # Environment variables (gitignored)
├── tailwind.config.ts
├── next.config.ts
└── package.json

---

Future Improvements

These are architectural stretch goals outside the scope of this assignment:

• Queue-based enrichment — Background processing with progress indicators via WebSocket
• Signal scoring — Weighted scoring model to rank companies by investment potential
• Vector similarity search — Semantic search using embeddings for "find companies like X"
• CRM integrations — Export enriched profiles to HubSpot, Salesforce, or Notion
• Persistent storage — Replace localStorage with PostgreSQL for multi-device access
• Authentication — User accounts with role-based permissions

---

This project focuses on shipping a usable product workflow with safe real-time enrichment under an 8-hour constraint.