MATCHPoint

Modern, AI‑assisted platform for evaluating startups with consistent signals, risk checks, and investor‑grade insights.

---

✨ Highlights

• Dynamic response formatting: chat answers render as tables, lists, comparisons, or narratives automatically
• Natural‑language intent: understands “top 10 above score 20”, “compare X vs Y”, “tell me about Z”
• Robust SSE streaming: ordered chunks with deduping for clean, flicker‑free chat
• Investor‑grade guardrails: never fabricates numbers; clearly states when data isn’t available
• Clean UI: About, Pricing, FAQs, and a global footer integrated

---

🏗️ Architecture

frontend/ (Vite + React + Tailwind)
  ├─ components/
  │   ├─ layout/ Navbar, Footer
  │   └─ investor/chat/ (Chat UI: SSE stream, renderer for tables/lists)
  ├─ pages/ About, Pricing, FAQs, Investor views
  └─ services/ chatService.js (SSE reader with id-based dedupe)

backend/ (Django + DRF)
  ├─ core/services/ai_service.py  ← intent detection + deterministic formatters
  ├─ core/views/chat_views.py     ← SSE, sequence id, deterministic first
  └─ config/settings/*            ← env profiles

Key flows:

• Deterministic pre‑answer: backend inspects the question and returns a structured answer when possible (no LLM needed).
• SSE streaming: when using LLMs (if keys configured), messages stream with id: headers; client dedupes and preserves order.
• Frontend rendering: pipe‑delimited tables auto‑render to a styled table; bullets/headings auto‑format for readability.

---

🚀 Quick Start

Prerequisites

• Node.js 18+
• Python 3.11+

Dev uses SQLite by default. No extra DB setup required.

1) Backend (Django API)

cd backend
python -m venv .venv
. .venv/Scripts/activate  # Windows PowerShell: .venv\Scripts\Activate.ps1
pip install -r requirements.txt

# Configure environment (optional; defaults are sensible)
copy .env.example .env  # or cp .env.example .env

python manage.py migrate
python manage.py runserver

API runs at: http://127.0.0.1:8000/

2) Frontend (Vite + React)

cd frontend
npm install
npm run dev

App runs at: http://127.0.0.1:5173/

If your API runs on another host/port, set VITE_API_URL in frontend/.env.local: VITE_API_URL=http://127.0.0.1:8000/api/v1

---

🧠 Chat: Dynamic Response Formatting

Implemented in:

• ai_service.py
• chat_views.py
• Frontend renderer: src/components/investor/chat/*

Detected intents (examples):

• Table: “show the company table”, “table with columns”
• List: “list startups with stage and score”
• Rank: “top 10 companies score more than 20”
• Compare: “compare Unicorn Tech vs MySQLTestCo”, “compare X, Y”
• Company profile/metric: “tell me about Beta”, “what is Alpha’s MRR?”
• Summary: “quick overview of visible startups”

Formatters:

• Rank table: Rank | Company | Score | Stage | Country | Rating
• Comparison table: Metric | <Company A> | <Company B> …
• Clean lists and summaries with missing data shown as “—”

Streaming:

• Server emits id: for every SSE chunk
• Client dedupes and orders chunks; tables and lists render cleanly

Guardrails:

• If a metric is missing, the system explicitly says it’s not in platform records
• Unrelated requests trigger a professional refusal

---

🧪 Tests

cd backend
python manage.py test core -v 2

---

🧭 UI Pages

• /about – Mission, value props, and contact CTA
• /pricing – Plans for founders, investors, and accelerators
• /faqs – Collapsible Q&A with platform expectations

All pages share the global Navbar and new Footer.

---

🔧 Configuration

Backend .env (optional):

• OPENAI_API_KEY – enable LLM answers (fallbacks if not set)
• GEMINI_API_KEY – optional Gemini fallback
• DJANGO_SETTINGS_MODULE – choose environment profile

Frontend .env.local (optional):

• VITE_API_URL – base URL for API (e.g., http://127.0.0.1:8000/api/v1)

---

🖼️ Screenshots

Add screenshots to docs/screenshots/ and they will appear here automatically:

Tips:

• Use 1400×800 (or similar) for consistent aspect ratio.
• Dark mode screenshots look best with the current theme.

---

🙋 Contributor Quick FAQ

Q: How do I run both apps together?
A: Start Django (python manage.py runserver) on :8000 and Vite (npm run dev) on :5173. Set VITE_API_URL to your API base.

Q: Where is chat formatting logic?
A: Backend: backend/core/services/ai_service.py (intent + formatters).
Frontend: src/components/investor/chat/ (SSE reader + renderer).

Q: Why do I sometimes see “Information not available…”?
A: Guardrails avoid fabrication. Add the metric to platform records or relax the display if appropriate.

Q: How do I add a new deterministic format?
A: Add detection in _detect_intent, the formatter function, and a simple renderer pattern in ChatMessage if needed.

Q: Tests?
A: python manage.py test core -v 2. Add unit tests in backend/core/tests.py.

---

🗺️ Developer Guide

Where to extend:

• Intent detection: ai_service._detect_intent
• Structured outputs: _format_rank_table, _format_compare_selected, _company_profile
• Stream behavior: chat_views.event_stream (SSE, ids)
• Frontend rendering: ChatMessage.jsx (table, list, headings)

Tips:

• When adding new formats, return deterministic text the renderer can recognize (e.g., pipe tables or bullet lists).
• Keep sections concise; the UI styles headings and bullets automatically.

---

📄 License

Private – Internal use only.

---

_{Built with ❤️ for clarity in startup investing.}