Bolt is really easy to use. It's written in go, so you don't need to install any dependencies to get it work - include Bolt to your source and you're ready. Bolt supports transactions. Actually, you can't do anything without a transaction. Good for Bolt - it makes transactions very easy and transparent. Though Bolt is a key-value storage, it supports nested collections (a.k.a. batches). It's a killer feature when you need to use namespaces. Bolt is very good on workloads.

portability and simplicity to setup in dev process

My favourite feature is how small you can keep your database, which is great for embedded linux systems where storage is a premium. The fact that it is written in pure Go, and the stack I wrote was also pure Go was great, so I didn't have to struggle with cross-compilation of something written in C-with-Go-binding.