# AI-Assistant
This repository contains the first sprint implementation of the AI Assistant, an intelligent planner and productivity tool. The system integrates an agentic AI architecture with vector search, persistent memory, and external APIs to assist users with scheduling, email, and task management.
Sprint 1 establishes the foundational infrastructure, including backend and frontend scaffolding, authentication, database setup, and a minimal AI pipeline.
Backend (FastAPI)
- Project structure initialized with FastAPI
- API routing and middleware configured
Database (PostgreSQL + pgvector)
- Postgres configured with pgvector extension
- SQLAlchemy models for user and embedding data implemented
Authentication (Supabase)
- User sign-up and login with Google session management
- FastAPI middleware for token verification
AI Core (LangChain + GPT-4)
- LLM integration via LangChain
- Initial memory persistence using pgvector
- Basic conversational endpoint with context retention
Frontend (React)
- Project scaffolded with React
- API client established for backend communication
- Login page and authenticated dashboard shell created
Deployment
- Frontend deployed to Netlify
- Backend deployed to Render with PostgreSQL
- GitHub Actions workflow for linting and build verification
- Backend: Python, FastAPI, SQLAlchemy, PostgreSQL, pgvector (through Supabase)
- Frontend: React
- AI Layer: OpenAI GPT-4, LangChain
- Authentication: Google OAuth2
- Integrations: Google Calendar and Gmail APIs (OAuth2)
- Deployment: Netlify (frontend), Render (backend)