I _really_ wanted Postgress' FTS to work for my usecase, but it was simply too slow and didn't have enough flexibility. I can certainly see its use for simpler use cases (fuzzy searching a product/user record, etc). However, it's not a full replacement for a search engine.

I really wanted it to be. I wanted to live in a world where I didn't have to update Postgres and then send the update to re-index the record in solr/es.

I don't know if I would recommend using the FTS facility in any RDBMS.

Apache Lucene is still very much a thing and hard to beat with solutions like this.

https://benchmarks.mikemccandless.com/