Handle your inbox with AI

You can't hand it your live inbox, so you copy the text in. Grab subject lines plus the first few lines of each email and paste them as one block — that's usually enough for it to sort.

1. Select your unread emails and copy subject + preview text into one document
2. Paste into Claude and ask it to bucket each into Reply Now, Later, Delegate, or Ignore
3. Ask for a one-line summary of what each Reply Now email wants
4. Have it draft replies for the urgent ones in your tone
5. Work down the list, sending or editing as you go

The trick is order: acknowledge before you explain. Most bad replies jump straight to defending yourself, which reads as 'you're wrong.' AI is good at flipping that order for you.

1. Paste the customer's email and your first-draft reply
2. Ask Claude to lead with acknowledgment, then the fix, then the explanation
3. Tell it to cut every word that sounds defensive or blame-shifting
4. Ask for the shortest version that still feels human
5. Read it out loud — if it sounds like you'd say it, send it

You don't need to read 300 emails — you need to know which 15 matter. Feed the pile in and let AI surface the ones that actually moved while you were gone.

1. Copy subject lines and previews from your unread pile into one block
2. Ask Claude to flag anything with a deadline, decision, or your name in it
3. Have it group the rest by sender or topic so threads collapse
4. Ask for a two-line summary of any long thread you got added to
5. Handle the flagged ones first; archive the noise in bulk

Payment chasers work when they're short, specific, and assume good faith. AI will keep you from either groveling or threatening — both of which make people slower to pay.

1. Give Claude the client name, amount, invoice date, and how many days late
2. Ask for a firm but warm reminder that states the number and a due date
3. Ask for a second, slightly harder version for the follow-up if they ghost
4. Have it keep both under five sentences
5. Swap in each client's details and send

Long threads bury the one thing you need. Paste the whole thing and ask for the decision, the ask, and who's waiting on whom.

1. Copy the full thread including everyone's replies
2. Paste it and ask Claude to summarize what's been decided and what's still open
3. Ask specifically: what is being asked of me, and by when?
4. Have it draft your reply that closes the open loop
5. Skim the original once to make sure it didn't miss context

Left alone, it defaults to safe corporate mush because that's the average of everything it's read. You fix it by giving it your actual voice to copy and banning the filler it reaches for.

• Paste two or three emails you've actually written and say 'match this voice'
• Ban specific phrases: 'I hope this finds you well', 'just circling back', 'per my last email'
• Tell it your reply should sound like a text to a smart colleague
• Give it a length cap — short forces it to drop the padding
• Ask it to keep contractions and cut adverbs
• Read the draft and delete one sentence — there's almost always one that adds nothing

Drafting is safe — you read before it goes out, so a weird sentence dies at your desk. Auto-sending hands a stranger your reputation on every reply, and it will eventually send something tone-deaf or factually wrong.

For anything with a real person or money on the other end, keep yourself in the loop. Draft, glance, send. The two seconds of review is cheap insurance against the email you'd have to grovel to take back.

Inbox zero isn't an empty inbox — it's that every email has been decided on: replied, delegated, archived, or scheduled. Nothing sits unread staring at you.

It's worth it not for the tidiness but for the mental load it kills. AI makes it reachable because the slow part was always deciding and drafting, and that's exactly what it takes off your plate.

Don't burn the bridge with a flat no. Paste the recruiter's email and ask Claude to draft a reply that says what's off, what you'd actually want, and leaves the door open. Recruiters remember people who are clear and easy — you'll get the next role that fits.

Ask it to keep the reply warm and specific rather than 'not interested at this time,' which tells them nothing and gets you filed under dead lead.

That 'thoughts?' means 'read this and tell me what to do.' Paste the whole thread and ask Claude to pull out what's being decided, the options on the table, and a recommendation with a one-line reason.

Then send your boss three tight sentences: here's the situation, here are the options, here's what I'd do. You look sharp because you did the thinking they didn't want to.

Paste their request and ask Claude for a reply that says yes-to-the-relationship, no-to-the-freebie: acknowledge the ask, name that it's outside scope, and offer it as a quick paid add-on or a next-project item.

The move is being friendly and firm in the same breath. Clients respect it more than a pushover, and AI is good at hitting that exact register when you tell it to.

Templates win for the exact same email you send fifty times. AI wins the second anything is one-off or needs to react to what the other person actually wrote — which is most email that matters.

Built-in AI is convenient because it sits right in your inbox, but it's usually shallow — quick summaries and stiff drafts. Claude gives you real triage, tone control, and reasoning, at the cost of copy-pasting text over.

For a fast one-liner, use what's built in. For the pile-up, the sensitive reply, or the decision, take it to Claude.

A little, if you never look at the output. The skill fades when you paste and send blind. It stays sharp when you read every draft and fix the one thing that's off — you're still making the judgment call, just faster.

Use it as a first draft, not a final one. The people who get worse are the ones who stopped reading. The people who get better use it to try three tones and notice which one actually lands.

Trust it to sort the obvious stuff — newsletters, receipts, clear asks. It's genuinely good at flagging deadlines and direct requests. But it can miss the quiet-but-critical one: the vague email that actually meant 'this is urgent.'

Let it do the heavy sorting, then eyeball the 'ignore' pile for thirty seconds before you archive. It's a great filter, not a safety net you can fully close your eyes on.