n8n + Gmail: what you can automate

When the inbox is also the order desk

This is for office managers, owners and operations leads whose inbox doubles as the billing desk and the support line. If someone on your team spends the first hour of the day labeling mail, forwarding invoices to the bookkeeper and copying form details into replies, the case is simple. That hour is repeated work with clear rules, and clear rules are what a workflow handles well.

The hour is rarely the real cost, though. The real cost is the urgent message from a good customer that sat under forty newsletters until lunch, or the invoice nobody logged until the late fee arrived.

You work with two nodes. The Gmail Trigger watches a mailbox and starts a workflow when mail arrives. Its one event is Message Received, filtered by label, sender, read status or any Gmail search. The Gmail node does the work: send, reply, add or remove labels, mark as read, create drafts and manage threads. Its Send and Wait for Response operation pauses a workflow until someone answers, which is handy for approvals.

Five Gmail automations worth building

1. Sort incoming mail into labels with AI

The Gmail Trigger fires on Message Received. A Text Classifier node, powered by a chat model such as OpenAI, reads the subject and body and picks a category: billing, sales, support or noise. Then the Gmail node’s Add Label operation files the message. Your team opens a sorted inbox instead of a pile. Start with five clear categories plus a catch-all, not fifteen fuzzy ones.

2. Draft replies for review instead of auto-sending

For routine questions like opening hours, order status or where to find a price list, the workflow writes a reply with an AI model and saves it with the Gmail node’s Draft: Create operation, in the same thread. Someone reads it, fixes a line and clicks send. No model emails a customer unsupervised. This is the version I suggest most teams build first, and it’s the one my Gmail triage lesson walks through step by step.

3. Save invoice PDFs and log them to a sheet

Point a Gmail Trigger at mail with attachments. Its Search filter accepts normal Gmail operators such as has:attachment filename:pdf. The workflow uploads the PDF to Google Drive, pulls the vendor, amount and due date, and adds a row with Google Sheets. Your bookkeeper works from one list, and every row links back to the file. The flow below builds it.

4. Send templated follow-ups after a form or booking

An n8n Form Trigger, or your booking tool posting to a Webhook node, starts the flow. The Gmail node’s Send operation then emails the person by name, with what they asked for and the next step. A Wait node can hold a shorter second nudge for two days later. Leads hear back within a minute instead of whenever someone next checks the form inbox.

5. Forward urgent messages to Slack

Give the Gmail Trigger a Sender or Search filter, such as a key client’s domain or words like “cancel” or “not working”. Add an IF node to double-check, then use Slack’s Send Message to post the sender, subject and a short preview to the right channel. The aim isn’t more notifications. It’s making sure the few messages a week that really can’t wait reach someone who isn’t watching email.

Example flow: invoice PDF to a logged row

1. Gmail Trigger (Message Received). Poll every minute, with a search such as has:attachment filename:pdf -label:logged. Turn Simplify off and switch on the Download Attachments option.
2. IF. Continue only when an attachment arrived (the first one lands in a binary field named attachment_0).
3. Google Drive (Upload). Save the PDF to an Invoices folder and keep the file link.
4. Extract from File (Extract From PDF). Turn the PDF into text.
5. Information Extractor with an OpenAI Chat Model. Return the vendor, invoice number, amount and due date as fields.
6. Google Sheets (Append or Update Row). Match on invoice number, so an invoice sent twice updates its row instead of creating a duplicate.
7. Gmail (Message: Add Label). Apply “logged”. The trigger’s search excludes that label, so the message is never processed twice.

Seven nodes, and none of them sends anything to anyone, so it’s a safe first build. Once it runs cleanly for a week, the same trigger can feed the reply side. The AI email responder lesson builds that half, ending in a Gmail draft.

Gmail pitfalls to plan for

• OAuth setup and scopes. The Google credential needs a Google Cloud project with the Gmail API turned on and an OAuth consent screen. Gmail access is broad (read, modify, send), so connect a dedicated or shared mailbox where you can, not someone’s personal inbox. On Google Workspace, an admin may need to allow the app.
• The seven-day stop. If the consent screen is External and still in Testing, Google expires the tokens after seven days and every Gmail node fails until you reconnect. For Workspace, use an Internal app, or move an External app out of Testing.
• It polls, it doesn’t push. The trigger checks Gmail on the schedule you set and takes a capped batch each time, queuing the rest for the next check. A delay of a minute or so is normal, which is fine for triage.
• Quotas and sending limits. The Gmail API has per-user usage limits, and mail sent through the node counts toward your account’s normal daily sending limit, which is lower on a free Gmail account than on Workspace. Bulk campaigns belong in an email marketing tool.
• Threads and duplicates. Use Reply, not Send, to keep a response in the original thread. Always label what you’ve processed. Otherwise, changing a filter can pull in old mail and run it again.

Build it yourself, or get a hand?

Labeling, Slack alerts and a single follow-up email are good first projects. With a free lesson and an afternoon, most people get them working. I teach n8n on Udemy to about 15,000 students, and those are the builds they finish on their own.

Bring in help when the workflow touches several mailboxes, has to follow Workspace security policies, drafts replies to customers in your brand voice, or feeds invoice data into accounting. At that point you need error alerts, retries and testing on real mail, not just a happy path. You can see how I approach operations work on the operations automation page.