Internal Assistant

AI-powered document chatbot. Import documents from SharePoint and ask questions — Internal Assistant retrieves relevant content and generates answers using RAG.

Architecture

• Backend: Node.js + Hono + Vercel AI SDK + Drizzle ORM
• Frontend: React (Vite) + shadcn/ui + Vercel AI SDK
• Vector DB: Neon Postgres + pgvector (halfvec with HNSW index)
• Documents: Microsoft SharePoint via Graph API
• Auth: MSAL (Microsoft Entra ID, delegated)
• AI Provider: OpenAI-compatible API (configurable)

Prerequisites

• Node.js 18+
• A Neon Postgres database with the pgvector extension enabled
• An OpenAI-compatible API key
• A Microsoft Entra ID app registration (see setup docs)

Quick Start

1. Provision Neon

Create a project at neon.tech, enable the vector extension, and copy the pooled connection string into backend/.env as DATABASE_URL.

2. Backend

cd backend
cp .env.example .env
# Edit .env with your API key and Azure credentials

npm install
npm run db:migrate   # apply Drizzle migrations (creates resources + embeddings tables)
npm run dev

3. Frontend

cd frontend
cp .env.example .env
# Edit .env with your Azure client ID and tenant ID

npm install
npm run dev

Open http://localhost:5173 — sign in with your Microsoft account and import documents from SharePoint.

Run everything at once

Once dependencies are installed and .env files are configured, start the whole stack — the 9router AI proxy (:20128), backend (:8000), frontend (:5173), and a Claude Code pane — in a tmux session with a single command from the repo root:

./dev.sh

The session uses the project .tmux.conf (labeled pane borders, mouse support, violet status bar). Re-running the script attaches to the existing session.

• Detach (leave everything running): Ctrl-b d
• Switch panes: click, Option+arrow, or Ctrl-b h/j/k/l
• Stop everything: tmux kill-session -t internal-assistant

Requires tmux (brew install tmux).

Configuration

Backend (backend/.env)

Variable	Description	Default
OPENAI_API_BASE	AI API base URL	https://openrouter.ai/api/v1
OPENAI_API_KEY	API key	—
CHAT_MODEL	Chat model name	deepseek/deepseek-v4-flash:free
EMBEDDING_MODEL	Embedding model	nvidia/llama-nemotron-embed-vl-1b-v2:free
DATABASE_URL	Neon Postgres URL	—
AZURE_CLIENT_ID	Azure app client ID	—
AZURE_TENANT_ID	Azure tenant ID	—
AZURE_CLIENT_SECRET	Azure client secret	—

Frontend (frontend/.env)

Variable	Description
VITE_AZURE_CLIENT_ID	Azure app client ID
VITE_AZURE_TENANT_ID	Azure tenant ID

Supported File Types

PDF, TXT, Markdown, DOCX, CSV, XLSX