Start herebeginner15 min
Triage a Flooded Inbox with AI, and Draft Replies (You Always Hit Send)
Paste a long thread or a batch of subject lines into Claude or ChatGPT and ask it to summarize, tell you what actually needs a reply, and draft a response in your tone. A 40-message chain becomes three lines and a clear ask, and a reply you can edit and send in seconds. Nothing leaves your name until you hit send yourself.
Triage a Flooded Inbox with AI, and Draft Replies (You Always Hit Send)
Paste a long thread or a batch of subject lines into Claude or ChatGPT and ask it to summarize, tell you what actually needs a reply, and draft a response in your tone. A 40-message chain becomes three lines and a clear ask, and a reply you can edit and send in seconds. Nothing leaves your name until you hit send yourself.
The inbox tax is decisions
A flooded inbox doesn't cost you reading time, it costs you **deciding** time. Every unread message makes you stop and judge: urgent or noise, reply or ignore, now or later. Sixty of those tiny verdicts before lunch is exhausting, and the genuinely urgent one hides between a newsletter and a calendar invite. Letting AI do the first sort hands you the decisions already made, so you spend your energy on the messages that actually need you.
A chat can't hit send
Doing this in an AI chat has a quiet superpower: the AI **cannot send anything**. It hands you text, a summary or a draft, and you copy the finished reply back into your mail app and send it yourself. That single gap is the whole safety story. There's no "auto-send" to misfire, no message going out in your name that you didn't read. You get the speed of automation without ever losing control of what leaves your name.
Rank the pile first
Copy a batch of subject lines and senders, or forward a clump of messages into [Claude](/tool/claude/) or [ChatGPT](/tool/chatgpt/), and ask: ``` Here are today's unread emails (sender + subject). List the ones that actually need a reply or a decision from me, most urgent first. Flag anything time-sensitive. Ignore newsletters and notifications. [paste here] ``` Instead of scrolling sixty messages, you get a short **ranked list**. You open the furious client, not the newsletter.
Crush the monster threads
Long back-and-forth chains are where time vanishes. Paste the whole thread and ask: ``` Summarize this email thread in 3 lines, then tell me exactly what is being asked of me and any deadline. [paste thread] ``` A 40-message chain collapses into **three lines and a clear ask**. This single move saves the most time of anything here, because you stop re-reading the whole history just to find the one question buried at the bottom.
Draft it in your voice
Now turn the ask into a reply. Tell the AI the outcome and the tone, and give it a sample of your own writing: ``` Draft a reply: friendly but brief. Confirm Thursday 2pm and ask them to send the file beforehand. Match the style of this reply I wrote earlier: [paste an example]. ``` It writes the draft, you tweak a word, paste it into your mail app, and **you** hit send. That pasted style sample is the trick that makes the "robotic AI voice" largely disappear.
Built-in option, same idea
If your company gives you an AI assistant inside your mail app, like **Microsoft 365 Copilot** in Outlook or **Gemini** in Gmail, it does this same summarize-and-draft job right on the message, no copy-paste. The skill is identical: ask for the ranked list, the three-line summary, the drafted reply. And it still never sends until you click. Check your IT policy before pasting work mail into any outside chat.
Twice a day, 20 minutes
The pattern that actually sticks is two passes a day: once in the morning, once after lunch. Each pass: ask for the ranked list, summarize the threads it flags, draft the replies, send. Keep the prioritize and draft prompts saved as a **snippet** so they're one paste away. A 90-minute inbox slog becomes a focused 20 minutes, twice, and the rest of your day stays yours.
Try it on one ugly thread
Find the longest, most tangled thread in your inbox right now. Paste it into [Claude](/tool/claude/) or [ChatGPT](/tool/chatgpt/) with the summarize prompt above and read the three-line version. Then draft the reply, edit it, and send it yourself. That first "oh, that took 90 seconds" is the moment the habit sells itself.
Try this now
Your turn: open claude and set up the first step. Just do step one now — the rest takes minutes. Save this guide to pick up where you left off.
FAQ
Will the AI send emails on its own?
No, and that is the whole point of doing it this way. The AI hands you text, a summary or a draft, in the chat. You read it, edit it, and paste it into a reply you send yourself. There is no send button in a chat window, so every message that leaves your name was approved by you.
Is it safe to paste work email into an AI chat?
Follow your company policy first. For routine, non-confidential mail, summarizing and drafting in a reputable AI chat is low-risk. For sensitive or regulated content, use an assistant your IT department has sanctioned inside your mail tool, or strip identifying details before pasting. When unsure, ask, it is a one-line question to IT.
Can it write in my tone, not robotic "AI voice"?
Yes. Tell it the tone you want ("friendly but brief", "formal", "match the style of this reply I wrote") and paste an example. Edit the first few drafts and it adapts fast. Because you approve every send, anything that still sounds off never goes out.
How is this different from just reading my inbox?
Speed and decision-making. Instead of scrolling 60 messages deciding what matters, you get a ranked list of what needs you. Instead of re-reading a 40-reply thread, you get the ask in three lines. The reading you keep is the reading that counts.