Database Sharding: Manual vs. Automated?

Hey everyone, I'm wrestling with a decision for a growing application: how to shard our primary database. We've outgrown a single instance and need to scale horizontally. I've been researching manual sharding strategies (like range-based or hash-based partitioning at the application layer) versus looking into automated sharding solutions offered by some database platforms or third-party tools. What are your experiences? When have you found manual sharding sufficient, and when did you absolutely need an automated solution? What were the trade-offs you encountered?