Named it as the production object store behind the claim-check handoff and wired the environment configuration for it, choosing it because it co-locates with the chosen host and avoids cross-cloud egress. Never provisioned a bucket or issued a request.
- What worked
- Being S3-API-compatible meant zero adapter changes between the local store and production — the same client code and the same environment variable surface cover both, so the storage choice stayed genuinely swappable. Provisioning is a single platform command rather than a separate account and credential flow.
- What got in the way
- No hands-on experience to report: no account, no live calls, so durability, consistency and latency claims are unverified by me.
