The problem
Westpac runs 45 shared email boxes across teams — compliance, operations, customer service. Each box receives hundreds of messages a day. When a specific email needs to be retrieved for an audit, a dispute, or a compliance review, the question is always the same: where is it, and how fast can you find it? Without a structured archive, the answer is “somewhere, eventually.” That’s not good enough for a bank.
What we built
AutoArchive connects to all 45 shared mailboxes and runs a nightly backup job, pulling every email down and writing it to a secure, air-gapped local disk. The key design decision: each file is named using the email’s unique message ID. No database, no index to maintain, no search UI to build — the filesystem is the search. A staff member looking for a specific email pastes the message ID into a file search and has it in under a second.
The tool handles the full mailbox set in a single run, logs what it backed up and what it skipped, and alerts on failure. It’s designed to be auditable — a compliance team can verify the backup without understanding the code.
The automation angle
The interesting constraint was reliability over cleverness. A nightly job that touches 45 live mailboxes at a bank cannot fail silently. Every step writes a structured log entry. If a mailbox is unreachable, the job continues with the rest and flags the failure — it doesn’t abort. The result is a system that keeps running even when individual mailboxes have issues, and gives ops a clear picture of exactly what was and wasn’t archived each night.
How it’s used
- Compliance teams retrieve specific emails by message ID for audits and regulatory requests.
- Operations staff recover emails that were deleted or moved from shared boxes.
- IT verifies nightly backup completeness from the structured run log.
What it taught us
That the right data structure eliminates the need for a search product. Naming files by message ID means retrieval is a filesystem operation — fast, reliable, and requires no software beyond a file browser. The constraint that felt like a limitation (no database) turned out to be the feature: nothing to maintain, nothing to corrupt, nothing to explain to an auditor.