AI Embeddings, pgvector & AI Coding Playground

This playground runs AI models, Python, and PostgreSQL entirely in your browser. No data leaves your machine.

Requirements:

Local AI (WebLLM): Chrome/Edge with WebGPU support. Downloads a ~1-2GB model on first use. Needs a GPU with 4GB+ VRAM.
API mode: Any modern browser. Bring your own API key (OpenAI, Anthropic, or OpenRouter). Keys are stored only in your browser's localStorage.
Embeddings: Run locally via Transformers.js (~30MB model download). No API key needed.
Storage: PostgreSQL 16 + pgvector runs in-browser via PGlite WASM.

AI-assisted coding editor + vector embeddings + PostgreSQL with pgvector. Everything runs in your browser.

AI Provider:

WebLLM

Local, no API key

OpenAI

GPT-4o-mini

Anthropic

Claude Haiku

OpenRouter

Many models

HuggingFace

Mistral 7B

Waiting for setup...

AI Assistant Ask it to write embedding code, SQL queries, or data analysis

Ask me to generate embeddings, store them in pgvector, and run similarity searches. Try: "Create a product catalog with embeddings and find similar items"

Python + SQL Editor

Results

Results will appear here after running code or SQL.

Direct SQL

Quick Actions