How PGSync Pro works¶

PGSync writes the SQL that stitches them into one document — automatically.

SELECT jsonb_build_object(
    'title',       book.title,
    'description', book.description,
    'genre',       book.genre,
    'author',      jsonb_agg(author.name)
)
FROM book
JOIN book_authors ON book_authors.book_isbn = book.isbn
JOIN author       ON author.id = book_authors.author_id
GROUP BY book.isbn;

One denormalized search document, kept fresh by change data capture.

{
    "title":       "Nineteen Eighty-Four",
    "description": "A clerk secretly rebels against a Party
                   that watches every citizen…",
    "genre":       "dystopian",
    "author":      [ "George Orwell" ]
}

The text that carries meaning becomes a vector.

Edit the price and the cached vector is reused — zero cost. Edit the description and exactly one embedding regenerates. That's the change-guard.

Users search by meaning — not keywords.