Love seeing a whole new generation of companies, like Tembo, laser-focused not only on breaking PostgreSQL down to it's basic building blocks but also on all PG extensions that are out, many of which are 10+ years old, of mixed quality, and then going super deep on improving them, re-architecting them for modern workloads ("Twitter Scale as described in this blog post) with the idea there are enormous amounts of performance gains and/or efficiencies to be gained.