01
- Branch cleanup and CI
- Preview environments and CI branches that automatically age out, with the right mix of ephemeral and long-lived branches so the limit stops being a blocker.
02
- Performance
- Compute sizing, connection pooling, suspend behavior and indexes reviewed against your workload, with latency and bill measured before and after.
03
- Migrations
- Moving onto Neon from a classic managed Postgres, or from Neon to another host, done with real rehearsal on a branch rather than a crossed-fingers deploy.
04
- pgvector and AI apps
- Embedding tables, index types, dimension and ANN parameters set against your retrieval quality target, inside the same Neon project as your application data.
05
- Backup and recovery
- Point-in-time retention set to a sensible window, external pg_dump copies for the paranoid, and a tested restore you have actually run.
06
- Managed Tech
- Someone watching compute hours, branch counts and egress every week, with alerts before overages rather than after, and a team on call when a deploy breaks a connection string.