PostgreSQL with pgvector for your AI app

Create a PostgreSQL database with pgvector for this app on Rowsafe Cloud and connect it with DATABASE_URL in .env.

Your agent picks a size, tells you the price and waits for your yes. Then it creates the server as you and, once it's ready (5 to 10 minutes), puts DATABASE_URL in .env. Through the Claude Code or Codex plugin, you create the login in the dashboard instead and paste it in: the password is made in your browser.

First time? Connect Rowsafe to your agent once.

rowsafe cloud create ai-db --extensions vector --wait
rowsafe env --on ai-db

rowsafe cloud create shows the price and asks first. rowsafe env makes the password on your computer, writes DATABASE_URL to .env and adds .env to .gitignore. Rowsafe never sees it. pgvector is on in every new database, so there's no CREATE EXTENSION to run.

CREATE TABLE documents (
  id bigserial PRIMARY KEY,
  content text NOT NULL,
  embedding vector(1536)  -- your embedding model's size
);
CREATE INDEX ON documents USING hnsw (embedding vector_cosine_ops);

import { Pool } from "pg";

const db = new Pool({ connectionString: process.env.DATABASE_URL });

export async function nearest(embedding: number[]) {
  const { rows } = await db.query(
    "SELECT id, content FROM documents ORDER BY embedding <=> $1::vector LIMIT 5",
    [JSON.stringify(embedding)],
  );
  return rows;
}

<=> is the cosine distance, closest first, in any language. In .env, change sslmode=require to sslmode=verify-full: the server's certificate is then checked too.

Rowsafe backs up every change, encrypted on the server with a key only you have, so you can restore to any second. Every week it restores a backup to prove it works. Restore a copy, bring back deleted rows or rewind the whole database, in one click. Before a risky change, save a Mark (rowsafe mark), or let your agent save one.

From $0.014 an hour, never more than $10 a month.