Triage unread email with an approval step
I don't sort emails anymore. Task: The author describes a workflow where Gmail triggers a Make.com scenario upon receiving unread emails. A Make.com AI Agent analyzes sender, subject, and content, applying custom triage rules to categorize, label, and archive messages. The agent sends a summary email to the user detailing actions taken, such as flagging urgent items or archiving newsletters. The source text cuts off before detailing the specific system prompt or tool configuration, so the exact implementation steps are incomplete.; tools: Make. Human role: The original post does not say how a person stayed involved.; limitations: Prerequisites include a Make.com paid account (starting ~$9/month), an OpenAI API key (~$5-10 credits), and a Gmail account. The source notes that the full system prompt and configuration details are behind a paywall, limiting reproducibility. There is no mention of how the agent handles ambiguous emails or if human approval is required before archiving, which poses a risk of misfiling important correspondence..
Tools
Published on Agent Work Atlas:
Overview
I don't sort emails anymore.
Workflow
The author describes a workflow where Gmail triggers a Make.com scenario upon receiving unread emails. A Make.com AI Agent analyzes sender, subject, and content, applying custom triage rules to categorize, label, and archive messages. The agent sends a summary email to the user detailing actions taken, such as flagging urgent items or archiving newsletters. The source text cuts off before detailing the specific system prompt or tool configuration, so the exact implementation steps are incomplete.
Reported result
The author reports receiving a summary email containing 3 urgent labels, 2 newsletter ideas, 1 follow-up required, and 4 archived emails, stating they did not open Gmail manually. However, the source does not provide measured metrics on accuracy, time saved, or error rates. It is unclear if the agent successfully handled all edge cases or if the summary reflects a single instance or an average.
Human role
The original post does not say how a person stayed involved.
Limitations
Prerequisites include a Make.com paid account (starting ~$9/month), an OpenAI API key (~$5-10 credits), and a Gmail account. The source notes that the full system prompt and configuration details are behind a paywall, limiting reproducibility. There is no mention of how the agent handles ambiguous emails or if human approval is required before archiving, which poses a risk of misfiling important correspondence.
Try it
Act as an automation engineer. I need to build a proactive email triage agent using Make.com and OpenAI. Inputs: [Gmail Account], [Make.com Scenario ID], [OpenAI API Key], [Triage Rules List]. Steps: 1. Configure a Gmail 'Watch New Emails' trigger in Make.com. 2. Set up an OpenAI module to analyze email metadata (sender, subject, body) against my [Triage Rules List]. 3. Define actions: Label urgent items, archive low-priority newsletters, and flag follow-ups. 4. Create a final step to send a summary email to [My Email Address] listing all actions taken. 5. Dry Run: Process 5 sample emails manually first. Show me the proposed labels and actions for each. Wait for my confirmation before enabling the live trigger. Output: A JSON configuration for the Make.com scenario and a sample summary email format. Ensure no emails are archived without my explicit approval in the dry run phase.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.