AUTOMAIL · CASE №01 · 2026

The shared inbox that stopped being a black hole.

How a small Microsoft 365 stack — Outlook, Power Automate, SharePoint — turned a triage swamp into a governed workflow in five days. AI got an honest job, not a marketing role.

Stack
Microsoft 365 only
Build
5 days
Outcome
−80% triage effort
Status
Live, owned by the team

We had three people checking the same inbox “just in case.” After Automail it’s one person, exceptions only, and nothing falls through.

— Operations lead, Automail client (name held by request)

The problem

A shared info@ inbox had stopped being a tool and become a stress. New work landed next to vendor invoices, customer urgencies, and weekly noise. Whoever checked first felt responsibility for everything; whoever didn’t check assumed someone else had it. Some messages were triaged five times by five people, others sat for half a day. The team had built a folder structure on top — but folders don’t run a workflow, they only describe a guess. When volume rose, the “system” was three people on Slack asking did anyone see the one from Tuesday?

The approach

The fix didn’t need a platform — it needed a workflow you could see. Power Automate now watches the mailbox, parses each new message into a routing decision, and writes the result into a SharePoint list that owns the queue state. Routine paths (vendor invoices to accounting, scheduling requests to ops) auto-route in seconds. Anything ambiguous gets a single classification step from a small AI call — but only as a hint with a confidence score, never as the final word. Low-confidence and high-stakes items always surface to a human with full context. The whole pipeline is inspectable in SharePoint: who owns what, when it landed, what the system thought, what the human did. No mystery box.

Outlook shared inbox Power Automate parse · decide AI classify optional · confidence Rules deterministic SharePoint queue · audit Auto route Human in-loop
Mailbox → Flow → (rules / optional AI) → SharePoint queue → auto-handle or human-in-loop. Every box is inspectable; nothing is a black box.

The outcome

Six weeks after go-live, the team had real numbers — not adjective marketing.

  • −80% weekly triage hours
  • 3.4 min time-to-route, down from 5+ hours intra-day average
  • 100% of cases land with a named owner before review
  • M365-only no new vendor, no consultant retainer
  • 5 days end-to-end build, brief to live

This case sits in roughly 200 lines of Power Automate JSON and one SharePoint list. The next time someone says “we need an enterprise platform,” ask them to try the boring path first. The boring path is usually a five-day build.