When should you use it?
- You start each day sorting mail
- You can describe your triage rules
- You do not want auto-replies
Not for: Shared team inboxes with complex routing; Mailboxes where archiving anything is risky
What do you need, and what do you get?
You need
- A Gmail account
- A Make account and an OpenAI key
- A written list of triage rules
You get
- Labels on new mail
- Low-priority items archived
- A daily summary email
How does it work, step by step?
- Write the rulesList which senders and topics count as urgent, ideas, follow-up or archive.
- Watch for unread mailUse Make's Gmail trigger so the scenario runs on arrival.
- Classify with the agent modulePass sender, subject and body to the AI agent with your rules.
- Act with limitsAllow labeling and archiving only; never sending.
- SummarizeEmail yourself a list of what was labeled and archived.
The prompt
Our own prompt, inspired by Wyndo and written to work in Make and ChatGPT. Values in lime are placeholders.
For each new unread email, read sender, subject and body. Apply these rules: [rules]. Add exactly one label from URGENT, IDEAS, FOLLOW-UP or ARCHIVE; archive only ARCHIVE items. Never reply or delete. At 7am, email me a summary listing each message, its label and a one-line reason.
Inspired by Wyndo: Email Triage AI Agent: How I Automated Inbox Zero Without Losing Control. The prompt above is written by playbooks.bot in our own words; nothing from the original is quoted.
What can go wrong?
- Folder-heavy sorting buries important mail; he prefers labels and a summary.
- Check the archive often at first in case the rules are too aggressive.
Our take
Wyndo upgraded a reactive email assistant to a proactive one in Make.com: a Gmail watch sends each unread message to an AI agent module that applies written triage rules, labels mail as urgent, newsletter idea or follow-up, archives low-value items and emails him a summary, never replies. Running cost is roughly twelve to fifteen dollars a month. Simple and sensible, with the right limits. The summary email is the trust mechanism, because you can see every action. Start with a narrow archive rule and widen it after a week of checking. Keep the rules in a document you can edit without opening the automation. Review a sample of archived mail weekly for false positives. Do not grant send permission to this flow. Separate work and personal inboxes if one set of rules cannot serve both. When labels proliferate, prune them so the summary stays readable. Re-test after Gmail or Make.com API changes.
Where the time goes back: Sorting happens before you open the inbox, so the first look is a short summary rather than a pile.
FAQ
Does it reply for him?
What does it cost?
Source and credit
Inspired by Wyndo. We describe the idea in our own words, link to the original, and wrote the prompt ourselves; nothing is quoted.
- Inspired by
- Email Triage AI Agent: How I Automated Inbox Zero Without Losing Control
- Creator
- Wyndo
- Published on
- aimaker.substack.com (Substack)
- Original date
- Jan 15, 2026
- License
- Not openly licensed (the poster keeps all rights), so we quote nothing and link to the original
- Originally for
- Make
- AI dry run
- Oct 23, 2026: passed. Dry-run tested by our AI test agent on sample input. PASS: Gates held on desk check; no live sends or purchases.
AI dry run on sample data; not run with live connected accounts. The playbooks.bot test agent filled the prompt with synthetic inputs, produced the output and checked it against the page's stated outputs and guardrails. - Verified
- Oct 23, 2026: source link resolves to the original, the credited creator published it, our excerpt or prompt matches it, and the license allows our use.
Are you the creator? Claim or correct this page.