AI Communication Assistant

A learning + portfolio AI SaaS built with Python, Streamlit, LangChain, Google Gemini, RAG, and LangGraph.

This repository evolved from a simple email generator (v0.1.0) into a multi-mode communication assistant with auth, history, usage tracking, retrieval, and workflow orchestration.

---

Features

• SaaS-style dashboard UI (sidebar, modes, composer, result card)
• Write Email with tone + length and structured subject/body
• Reply to Email
• Ask AI (general Q&A)
• Demo authentication + user profile (Supabase email/password)
• Supabase Postgres history and usage/token estimates
• RAG over files in knowledge/
• LangGraph workflow: understand → retrieve? → generate → review
• Input validation, rate limiting, logging, GitHub Actions CI

---

Quick start (Windows PowerShell)

cd "D:\AI Projects\email-assistant"
python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Create .env (never commit this file):

GOOGLE_API_KEY=your_gemini_api_key_here
SUPABASE_URL=https://YOUR_PROJECT_REF.supabase.co
SUPABASE_ANON_KEY=your_supabase_anon_key_here

Run supabase/schema.sql in your Supabase SQL Editor, then start the app:

streamlit run app.py

Sign up or sign in with your email and password.

---

Architecture

Streamlit UI
  -> Auth / Profile
  -> LangGraph workflow
       -> RAG retrieve (Ask AI)
       -> Write | Reply | Ask services
       -> Review node
  -> Supabase Auth + Postgres (profiles, history, usage)

See PROJECT_SUMMARY.md for the phase roadmap and file roles.

---

Streamlit Cloud

• Main file: app.py
• Secrets: GOOGLE_API_KEY, SUPABASE_URL, SUPABASE_ANON_KEY
• Run supabase/schema.sql in Supabase before first deploy

---

Security

Never commit:

• .env
• API keys
• passwords / tokens

.gitignore already excludes .env, .venv/, and __pycache__/.

---

License

Educational / learning purposes.