Practice

How it runs.

What happens to an email here between the brief and the report.

  1. Receive the HTML, log it, list its faults.
  2. Rebuild on the house shell, from the library.
  3. Test: the QA gate, eight proofs, five clients, light and dark.
  4. Version: every change listed beside the proof before the sponsor sees it.
  5. Schedule: seed list, links tagged in one grammar, send time agreed.
  6. Deploy, then confirm on five clients.
  7. Report at day 7, as an email.

The sponsor rule

Every change goes to the sponsor as a numbered list before they see the render, colour included. Proof 1 swapped their green for an AA-safe tint without saying so, and they asked why their brand looked wrong.

The gate

Every email has a pre-send checklist of — items, counted by script, and a render matrix from email-hub’s local client emulation. Outlook is captioned as a structure check, not a Word-engine capture.

Machine and human

AI agents built much of this. The decisions, the checklist and every sign-off are mine. The agents assemble and check. I decide what ships.