The task
It’s 4:50 on Friday. Your manager asked for a weekly update by EOD. You have 12 things in flight, 4 you’d rather not surface, 1 decision that’s been quietly waiting on her since Tuesday, and a half-typed Slack thread in another tab. She’ll skim the email between her own 1:1s on Monday morning. You need an update she can read in 2 minutes that does three things: name the one decision she needs to make, flag one blocker before it becomes a Monday-morning fire, and stop short of becoming a task diary.
Where AI helps — and where it does not
AI is excellent at restructuring messy notes into the “decision / blockers / shipped / next” pattern, compressing 12 items to 4, and ensuring the ask appears in the first 5 lines instead of paragraph three. It is also good at converting passive-voice cover (“the security review was delayed”) into ownership-clear language (“I’m 3 days late on the security review; mitigation below”). What AI cannot do: decide which items deserve her attention this week and which belong in your 1:1 doc. That judgment stays with you — AI will surface everything you give it, including the things you should be solving on your own.
The named failure mode: the task diary. AI lists every meeting attended, every PR merged, every Slack thread participated in. Your manager skims, finds no decision to make, and tunes out. Force the prompt to demote anything that does not need her attention this week.
What to feed the AI
- Your raw notes from the week — Slack threads, Linear tickets, half-thoughts, calendar items, brain dumps
- One specific decision you need from her this week, with your recommended call
- One blocker or risk that needs her visibility, with your current mitigation
- Items shipped this week with the business impact (not the artifact count)
- Top 3 things you’ll work on next week, in priority order
- 2 items you’re explicitly NOT bringing to her — context for her, hidden from email
- Last week’s update for continuity (so phrasing rotates, not repeats)
- Her preferences — does she like a bottom-line-up-front opener, a TL;DR, a one-line subject?
Copy-ready prompt
Rewrite my messy weekly notes into a manager update email.
Notes (raw):
{paste Slack snippets, Linear titles, half-thoughts}
Decision I need this week + my recommendation: {decision, my recommended call, reasoning in one sentence}
Blocker / risk I need her visibility on + current mitigation: {blocker, what I'm doing about it now}
Shipped this week (business impact, not artifact count): {3-5 bullets}
Next week (priority order, max 3): {bullets}
Items I'm NOT bringing this week (context for the model, not in email): {list}
Last week's update (for phrasing variety only): {paste if available}
Her preferences: {BLUF / TL;DR / short subject / no greetings, etc.}
Structure of the email:
1) Subject line: 8-10 words, includes the headline word "decision" if there is one
2) One-line headline: status + the one ask
3) Decisions needed: bullet, each with my recommendation. If none, omit this section.
4) Blockers / risks: bullet, each with current mitigation. If none, omit.
5) Shipped: max 3 bullets, business-impact phrasing not artifact count.
6) Next week: max 3 bullets, priority order.
Rules:
- Total under 200 words.
- The one ask appears in the first 5 lines or in the subject.
- No filler ("As discussed", "Wanted to circle back", "Hope you had a great week").
- Every bullet starts with a verb, not a noun.
- Demote anything that does not need her attention this week — note it as "FYI / in 1:1 doc".Shorter variant — three-bullet Slack update
Compress these notes into a 3-bullet Slack DM to my manager.
1) The one decision needed + my rec, 2) the one blocker + mitigation, 3) one signal of momentum.
Max 60 words. Notes: {paste}Sample output
A useful headline: “On track for Friday launch — one open decision on analytics scope.”
A useful decision-needed line: “Decision needed: approve the slimmer v1 scope (recommend yes — saves 4 days, keeps the 3 metrics we agreed on, drops the dashboard polish that’s not on the critical path). Let me know by EOD Tuesday or I’ll proceed with the slim scope.”
A useful blocker line: “Blocker: security review pushed back on the OAuth scope change. Mitigation: I’ve pre-briefed the security lead and scheduled a 20-min sync for Monday 10am; if that doesn’t unblock, I’ll come to you.”
A useful shipped bullet: “Shipped: merchant onboarding rebuild rolled to 100% — first-week activation up 14% in early cohorts (vs control).” That beats “Merged PR #842.”
How to refine
- Front-load the ask: “Rewrite so the one decision needed appears in the subject line or within the first 5 lines. If I cannot find the ask in 10 seconds, neither can she.”
- Demote noise: “Demote anything that does not require her attention this week. Move it to a ‘FYI’ line or drop it entirely. Goal: 4 sections, max 200 words.”
- Lead with impact, not artifact: “Every shipped bullet must name the business or user impact, not the artifact (‘first-week activation +14%’ beats ‘merged the onboarding rebuild PR’).”
- Always include a recommendation: “Every decision needed must include my recommended call and one sentence of reasoning. No open-ended ‘what do you think?’”
- Soften the blocker, sharpen the ask: “For each blocker, name the current mitigation in the same line. If I can’t mitigate it myself, ask explicitly for the help I need.”
Common mistakes
- Listing every meeting attended — your manager doesn’t care that you sat in 7 meetings, she cares what changed because of them
- Burying the decision request inside paragraph 3 — she scrolls past it, comes Monday she has to ask
- No recommendation alongside the ask — managers experience open-ended “what do you think?” as work being kicked back to them
- Passive voice covering ownership — “the launch was delayed” instead of “I delayed the launch because…”
- Identical phrasing every week — same opener, same structure, same words; your manager starts skipping the read
- Padding a “decision needed” slot with a fake decision — dilutes the next real one
- Treating the update as the only channel — anything that needs discussion goes in the 1:1 doc, not email
- Letting AI hedge (“may potentially require”) — every hedge is a sentence that doesn’t earn its place
FAQ
- How is this different from a 1:1 doc?: Manager updates are “what she needs to know now and possibly act on.” 1:1 docs are “what we should talk about together.” If it can wait until Monday’s 1:1 and benefits from conversation, it goes in the doc, not the email.
- What if there are no decisions needed?: Skip that section. A 3-section email (blockers / shipped / next) reads fine. Padding a fake “decision needed” trains your manager to ignore that section, which costs you next time there’s a real one.
- What about peer FYIs (other team’s blockers, dependencies)?: Surface only if she would need to escalate or de-escalate. Otherwise it’s noise. A weekly note about “the platform team’s API delay” only belongs in the update if it changes her plan.
- Should I cc anyone?: Default to just her unless she has asked otherwise. Adding stakeholders changes the audience and forces you to write a different (longer, more defensive) email.
- How do I handle a week I underperformed?: Lead with what shipped (even if small), name the slip in blockers with the mitigation, and propose the recovery plan in “next week.” Avoid an apology paragraph — she doesn’t need it, and it makes the slip the headline.
Related
- AI exec summary
- AI project status update
- AI Prepares Your 1:1 Talking Points
- Rewrite a Tense Work Message With AI
- AI KPI Weekly Report
- AI Performance Review Draft
Tags: #AI writing #Career #Workflow #Email