Loading…
← All results// Case study 01 · Inbox Killer · live
The self-running back-office
Inbox Killer · live
Deployed in my own NL print-business ecosystem (2+ years in production). Same architecture I deploy for clients — not theory.
Live mailbox, 142 msgs/scan, human-in-the-loop on every send.
Before — Problem
- ✕ Manual inbox triage every morning — leads mixed with newsletters and noise
- ✕ Reply drafting started from scratch for every thread
- ✕ No consistent lanes for lead vs client vs invoice mail
- ✕ Risk of auto-send tools firing without owner review
After — Resolution
- ✓ Mailbox scanned on schedule — 142 messages classified per run in production
- ✓ Lead, client, invoice, and noise lanes with priority surfacing
- ✓ Draft replies ready for edit and approval before send
- ✓ Human approval gate on every outbound — architecture, not a toggle
Architecture
One page you can forward: what happens from a new email to an approved send. The approval gate is the architecture — not an optional safety setting you might forget to enable.
The Build
Five steps under the hood — the same flow running in FLEXGRAFIK production mail.
01
Read
Agent connects via secure OAuth and polls the mailbox on a defined schedule.
02
Classify
Each message is sorted into lead, client, invoice, or noise lanes with scoring.
03
Draft
AI drafts replies in your tone. You edit or discard — drafts never auto-send.
04
Approve
Outbound mail waits at a human gate. Nothing leaves without your explicit OK.
05
Send
Approved replies send and log to the audit trail. Overrides train the lanes.
Stack
Gmail / Microsoft 365 OAuthMake.comOpenAI APIWebhook routingHuman approval gate
Proof
Inbox classification lanes with approval gate
Lead · client · invoice · noise, with approval gate.
// Video
Inbox Killer walkthrough
Screen recording is in production. The classification lanes screenshot and flow diagram show how it works today.
Book Automation Map →Measurement
Live mailbox, 142 msgs/scan, human approval on every send.
Could this work for your business?
Start with a paid Automation Map. In 60–90 minutes we map your exact setup, score the ROI, and recommend whether this is the right first system — before you commit to anything bigger.
Book your Automation Map