Module 3 of 7

Module 3: Email, Writing, and Replies

Why this matters

Email and short-form writing eat more of the average workday than meetings do. Most of it is structurally repetitive: replying, confirming, declining, clarifying, summarizing. AI is exceptionally good at this exact shape of work. If you only ever apply AI to one part of your job, make it this one, it pays back fast.

Reminder from Module 2: before pasting any real email content, check that it’s safe (no client data, no PII, no internal-only info) or redact first.

The bullet-to-draft pattern

A high-leverage email move is to type bullets and let AI write the prose.

Draft a [tone] email to [recipient/role] covering:
- [bullet 1]
- [bullet 2]
- [bullet 3]
Length: [short / medium / long]. Avoid: [what to leave out].

You stay in control of the content (the bullets are yours, that’s the part only you know). AI handles the prose (transitions, phrasing, opening, closing). This is the 70/30 rule applied to email.

Tone shifts

You can rewrite any draft into a different tone, the level of formality, warmth, or directness:

Use tone shifts to handle the part of email writing that drains you, finding the right voice for an awkward situation. You write the content once; AI gives you three versions; you pick.

Inbox triage

For a clogged inbox, AI helps you decide what needs you and what doesn’t. (This is the assistant analogy from Module 1, applied, your AI is filtering and surfacing, you’re still the decider.)

Below is the body of an email. Tell me:
1. What is the sender actually asking for?
2. Does this need a reply? (Yes / no / FYI only)
3. If yes, what's the shortest acceptable reply?
4. Anything I should be careful of?

[paste email]

Two warnings:

Reply patterns

Three patterns cover the bulk of professional replies:

Acknowledge + commit + timeline:

“Got it, I’ll [thing] by [date]. Let me know if anything changes.”

Decline gracefully:

“Thanks for thinking of me. I won’t be able to take this on right now, [brief reason]. [Suggest alternative if appropriate.]”

Clarify before committing:

“Quick clarification before I dive in: [question 1]? [question 2]? Once I’ve got that, I’ll [next step].”

You can ask AI to draft any of these from a one-line context.

Try It At Work: Draft a Real Reply

Time: 10–15 min

You’ll need: Your actual inbox, open right now. One email you’ve been putting off. (Confirm it’s safe to paste, or redact first.)

Do this:

  1. Open the email. Type 3 bullets covering what you want to say.
  2. Paste the bullet-to-draft prompt. Let AI write the reply.
  3. Read the draft. Score yourself: how close to “I’d actually send this”, 70%? 90%? 50%?
  4. If under 70%: your bullets were too thin or you skipped tone. Try again with more context and a tone instruction.
  5. Edit the last 30% and send (or save as draft).

Done when: You have a reply you’d actually send sitting in your sent folder or drafts.

Try It At Work: Three Tones, One Decision

Time: 5 min

You’ll need: A draft you wrote this week, any email or message.

Do this:

  1. Paste the draft into your model.
  2. Ask: “Give me three versions of this, formal, direct, and warm. Same content, same length.”
  3. Read all three. Notice which one matches the relationship and the situation.

Done when: You can name why one version fits and the others don’t. That instinct is the skill.

Key takeaways

Quick Check

1. In the bullet-to-draft pattern, what does the human provide and what does AI provide?

2. You're writing a hard "no" reply. Which tone instruction is best?

3. What's the right way to use AI for inbox triage?

4. The "draft is under 70% of what you'd send" diagnostic usually means