[mirror_dude]

I mean I guess, but why not just use a lucene based system?

[philippemnoel]

Keeping it in Postgres avoids needing to denormalize the data, which is a big pain point. ParadeDB pg_search is based on Tantivy, a Lucene-inspired library, so in some ways it allows you to both benefit from a Lucene-based system and also keep it in Postgres

[cpursley]

Because that's another thing to set up, operate and pay for.