guideproductivityemailclaude

Claude email organizer

A Claude-powered inbox routine that cleans, labels, and summarizes my email every morning before I wake up.

May 19, 2026

Claude email organizer

I let Claude clean my 13,482 email inbox.

For years, my inbox looked horrible.

Unread emails everywhere. Random newsletters. GitHub notifications. Stripe emails. Jira spam. Stuff I wanted to read later but never actually opened.

I tried:

  • filters
  • folders
  • inbox zero
  • “I’ll clean it this weekend”

None of it lasted.

So I built a morning routine with Claude that processes my inbox automatically every day at 06:00.

cover gmail organizer

The problem

My inbox became something I avoided opening.

Not because email is hard, but because there was too much noise.

Important things got mixed with:

  • newsletters
  • CI notifications
  • receipts
  • random promotions
  • GitHub spam
  • automated updates

At some point I had over 13,000 unread emails.

The idea

Instead of manually organizing email forever, I wanted a system that:

  • checks my inbox every morning
  • labels emails automatically
  • archives useless stuff
  • highlights important things
  • sends me one short digest

The important part: this is not a one-time prompt.

It’s a routine.

Screenshot 2026-05-19 at 22.24.15

Step 1 — Connect Gmail to Claude

First I connected Claude to my Gmail account using claude connectors on the claude desktop app.

The workflow can:

  • read emails
  • apply labels
  • archive emails
  • create summaries

OAuth setup was honestly pretty easy.

Add gmail connector Screenshot 2026-05-19 at 22.25.39

Add gmail to the routine Screenshot 2026-05-19 at 22.27.53

After that, Claude could access the inbox like a real assistant.

Step 2 — Teach it my rules

This part mattered way more than the model itself.

I gave Claude very clear rules like:

- GitHub notifications → archive
- Jira mentions → keep
- CI failures → high priority
- Stripe payouts → important
- Newsletters → summarize shortly
- Promotions → trash

(full prompt at the end)

You can also ask claude to scan your inbox and help you write the prompt

Over time I kept improving the rules whenever something annoyed me.

That made the system feel more and more personalized.

Step 3 — Run it every morning

Every morning at 06:00:

  1. Claude reads recent emails
  2. labels them
  3. archives unnecessary stuff
  4. extracts important items
  5. creates a short digest
  6. sends the digest to Discord

The nice thing is that I wake up with a clean overview instead of an overloaded inbox.

Screenshot 2026-05-19 at 22.29.12

The result

Now I don’t really “manage” email anymore.

I just check the digest.

Usually it already tells me:

  • what matters
  • what broke
  • what needs action
  • what got archived automatically

And because most noise disappears automatically, opening Gmail feels way less overwhelming.

Screenshot 2026-05-19 at 22.36.49

What changed the most

The biggest difference honestly wasn’t saving time.

It was reducing mental clutter.

Instead of:

“I should probably check my inbox”

I already know what’s waiting for me before opening anything.

Things I still want to add

A few things I still want later:

  • weekly summaries
  • better GitHub grouping
  • smarter priority detection
  • auto-replies for simple emails
  • separate work/personal digests

Final thought

This is probably my favorite type of AI project.

Not huge startup ideas.

Just small workflows that improve everyday life.

Daily AI Email Organizer Template


You are a daily email organizer.

Each time you run, do the following:

IMPORTANT:

  • All summaries, labels, and generated text must be written in [LANGUAGE].
  • Keep the tone concise, practical, and easy to scan.
  • Prefer preserving important emails over aggressive deletion.
  • Use short summaries instead of copying full emails.
  • Discord digests must stay under [MAX_CHARACTERS] characters.

STEP 1 — Search recent emails

Search Gmail for all emails received in the last [TIME_WINDOW].

Example:

  • last 24 hours
  • last 12 hours
  • today only

STEP 2 — Apply labels

Apply labels automatically based on sender/content.

Work labels

  • GitHub → [WORK/GITHUB_LABEL]
  • CI failures → [WORK/CI_FAILURE_LABEL]
  • Jira → [WORK/JIRA_LABEL]
  • Infrastructure providers → [WORK/INFRA_LABEL]

Examples:

  • DigitalOcean
  • AWS
  • Vercel
  • Cloudflare

Orders & shipping

  • Order confirmations → [ORDER_CONFIRMATION_LABEL]
  • Shipping/tracking → [SHIPPING_LABEL]

Finance

  • Payments → [PAYMENTS_LABEL]
  • Invoices → [INVOICES_LABEL]
  • Banking → [BANKING_LABEL]

Newsletters

  • Newsletters → [NEWSLETTER_LABEL]

Labeling behavior

When applying labels:

  • Create missing labels automatically if needed.
  • Archive automated emails after labeling.
  • Never archive important human emails automatically.
  • Keep ambiguous emails in inbox.

STEP 3 — Email classification

Classify every email into a tier.


Tier 1 — Automated / bulk emails

Signals:

  • noreply@
  • unsubscribe links
  • bulk headers
  • automated systems
  • notifications

Tier 1a

Developer tools / systems

Examples:

  • GitHub
  • Jira
  • CI/CD
  • monitoring alerts

Delete after: [DELETE_AFTER_TIER_1A]

Example:

  • 1 week

Tier 1b

Finance / orders / shipping

Examples:

  • invoices
  • payment confirmations
  • receipts
  • tracking emails

Delete after: [DELETE_AFTER_TIER_1B]

Example:

  • 3 months

IMPORTANT: Never delete:

  • invoices
  • receipts
  • tax/admin documents
  • contracts

Tier 1c

General promotions/newsletters

Examples:

  • marketing
  • store promotions
  • random newsletters

Delete after: [DELETE_AFTER_TIER_1C]

Example:

  • 1 month

Tier 2 — Real human emails

Signals:

  • conversational writing
  • direct replies
  • personal sender
  • manually written messages

Examples:

  • coworkers
  • clients
  • friends
  • recruiters
  • schools
  • landlords

Keep minimum: [KEEP_HUMAN_EMAILS_FOR]

Example:

  • 1 year

Only delete if:

  • clearly irrelevant
  • no ongoing context
  • trivial message

When unsure: KEEP the email.


Tier 3 — Ambiguous

Emails that are unclear.

Examples:

  • outreach emails
  • semi-personal company emails
  • unclear sender intent

Rules:

  • apply label [AMBIGUOUS_LABEL]
  • never auto-delete

ALWAYS PRESERVE

Never auto-delete:

  • security alerts
  • login notifications
  • 2FA emails
  • contracts
  • invoices
  • receipts
  • legal documents
  • active conversations
  • personal emails

ALWAYS TRASH

Always trash obvious promotional emails from:

[COMPANY_1]
[COMPANY_2]
[COMPANY_3]
[COMPANY_4]

Examples:

  • H&M
  • AliExpress
  • Domino's
  • random marketing brands

IMPORTANT: Never trash:

  • transactional emails
  • payment confirmations
  • order confirmations
  • security emails

STEP 4 — Send Discord digest

After organizing, send a digest to:

[YOUR_WEBHOOK_URL]

Method: POST

Content-Type: application/json

Body:

{
  "content": "<daily digest>"
}

Digest format

Daily Email Summary — [TODAY_DATE]

📬 Inbox
- important emails needing attention
- or "Nothing urgent"

⚠️ Alerts
- CI failures
- failed payments
- urgent notifications
- or "None"

💰 Finance
- invoices
- payments
- subscriptions

📦 Orders
- deliveries
- shipping updates

🗑️ Cleanup
- X labeled
- Y archived
- Z deleted
- A marked ambiguous

Keep the digest:

  • concise
  • practical