Inbox Triage & Reply-Drafting Agent
Not another set of templates — this is the live automation. A build-ready blueprint (n8n AND Make.com) that connects to your OWN Gmail or Outlook, labels and prioritises incoming mail the way you sort it, and drops a ready-to-review reply draft in your voice into each thread. You run it on your own accounts and keys; nothing sends without you clicking send. Built for sole traders, consultants, freelancers and agency owners drowning in email.
The Inbox Triage & Reply-Drafting Agent is a build-ready automation blueprint, A$129 one-off, for a system that runs inside your own Gmail or Outlook. Incoming mail gets labelled and prioritised according to rules you set, and a reply draft in your voice sits in each thread waiting for you to read it. It is aimed at sole traders, consultants, freelancers and agency owners whose new enquiries, supplier chasers and invoice questions all land in the same place. The document specifies the whole build for both n8n and Make.com, and it arrives on your order page as a Markdown download in about fifteen minutes.
Sorted the way you already sort it
At the centre of the document is a label and intent map — a table naming each type of mail you receive, how it should be classified, and which types jump the queue. It is written from your own answers about how you want mail sorted, so the categories are the ones you already think in rather than a generic urgent-or-later split. Alongside it sits a field-mapping table tracing the data from trigger through classifier and labeller to the draft step, so you are not guessing which variable feeds which module halfway through a build.
The reply prompt is personalised to your business and the voice you nominated, and it ships paste-ready. Four free AI revisions come with the blueprint, which matters more here than on most documents: a first pass at a label map reflects what you described, and how you actually sort mail usually turns out to be slightly different once you see it laid out in a table.
Built on your accounts, drafting into your threads
This is a specification rather than a hosted service. You build the scenario in your own n8n or Make.com account, on your own keys — the blueprint's data-safety note sets out where credentials go and confirms nothing is hosted by us. Because the full build is written out for both platforms, with steps, modules and triggers for each, the platform decision stays open until you are actually building.
The safety rule is written into the spec rather than left as advice: the agent classifies, labels and drafts, then stops. Every reply lands as a draft you open, edit and send yourself, and whatever you nominated as off-limits is carried through as an explicit exclusion in the rules. A "test it in 5 minutes" checklist gets you to a first live run without a full rollout, and a troubleshooting FAQ covers what tends to go wrong on the way there.
What you get
- Full build spec for BOTH n8n and Make.com (steps, modules, triggers)
- Label & intent map table — how each email type is sorted and prioritised
- Field-mapping table (trigger → classifier → labeller → draft step)
- Paste-ready brand-voice draft prompt, personalised to your business
- "AI drafts, you send" safety rule — nothing auto-sends
- "Test it in 5 minutes" checklist
- "Paste your OWN key here" data-safety note (your accounts, your keys, nothing hosted by us)
- Troubleshooting FAQ
- 4 free AI revisions included
Turnaround: ~15 minutes. Delivery: Automation Blueprint — Markdown download (n8n + Make specs).
Frequently asked questions
Who actually builds the scenario — me or you?
You do. What you receive is the build spec: steps, modules, triggers, field mappings and the prompts. You connect it to your own mailbox in your own n8n or Make.com account using your own API keys. Nothing is hosted by us and we never connect to your mailbox. You do need to be willing to work inside n8n or Make.com, though the spec names the modules, the order they go in and which field feeds which step, and the five-minute test checklist walks you to a first run. It assumes you can follow a technical build document, not that you have built automations before.
Is there a way to let it send replies on its own?
The blueprint is written the other way deliberately — the agent drafts, you send. Once a scenario is running in your own account you control what it does, but the safety rule, the exclusions you nominate and the whole draft-review structure are specified around a human reading every reply before it leaves.
What does running it cost after the A$129?
The A$129 is a one-off for the blueprint itself. Running it uses your own n8n or Make.com account and your own AI provider key, so whatever those providers charge for your volume is between you and them. We do not resell, bundle or mark up any of it, and we cannot quote it for you.
The first label map won't match how I sort — then what?
That is what the four included AI revisions are for. Send back which labels are wrong, which mail types were missed or which things should be prioritised differently, and the map, rules and draft prompt get rewritten. The revisions apply to the document, not to a build you have already wired up.
Included in Tradie AI Pack — save with a bundle.