The task
A small web-design agency for local businesses ran everything out of a single Outlook inbox, with no automation and no CRM. The owner had to categorize incoming mail (leads, invoices, support, spam) and answer it by hand. According to the source, this cost the owner hours every day.
What they did
The builder set up an n8n workflow that watches the Outlook inbox through the Microsoft Graph API. An AI step classifies each incoming message as a lead, a quote, an invoice or payment, a support request, spam, or an out-of-office reply. For messages that need some thought, the workflow writes a reply draft that uses the business name, the suburb, the requested service, and earlier messages in the thread, in local tone and spelling. The owner approves a draft with one click. The workflow also logs every action to a Google Sheet with its status and next step, so the owner can review what happened.
What the source reports
According to the source, the owner now spends roughly 30 minutes a day on the inbox, down from about 3.5 hours. The source says drafts reduce the replies that need more thought to a matter of minutes. The source attributes these results to the whole setup, including instant replies that this entry does not describe.
Where the approval sits
For messages that need judgment, nothing goes out until the owner approves the draft, which the source describes as a one-click step. The source also says that low-risk, high-confidence cases receive an instant reply without approval. This entry covers only the draft path, where a person approves each reply before it is sent.
Worth knowing
The source is the builder's own case study of one small agency on a personal blog, so the results are the builder's account. The same setup includes instant replies and scheduled reminders for leads and invoices. These run without a person approving each message, and this entry leaves them out. The source does not name the AI model, and it shows no calendar use or meeting booking.
Source: Personal blog · Aug 2025
Drafted with AI assistance and checked automatically. A person is responsible for the collection. How these entries are made
Prompt for your agent
Draft a customer reply that is careful with personal data
Draft a reply to a customer message that avoids personal data that is not needed, and get a list of points that you should check first.
More in this category
- Inbox and scheduling
A rules file guides Claude Cowork to draft email replies each morning
A professional writes a rules file that tells Claude Cowork how to sort the morning inbox and draft replies, then reads a one-page summary. The Gmail connector can only create drafts, so the author reads and sends every reply.
Read the entry - Inbox and scheduling
An n8n template labels Gmail threads and saves Gemini reply drafts
An n8n workflow template reads each new Gmail message with a Gemini agent, labels it, and saves a reply draft where one is needed. Each draft waits in Gmail, with the message marked unread, for a person to review.
Read the entry
Ninety minutes live on your tools. One automation running before we hang up.