Most weekly reports are dumped activity logs that nobody reads past the second paragraph — “I attended 12 meetings, reviewed 8 PRs, helped with onboarding…” — and they quietly train your manager to skip you. The fix is structure. Every prompt below forces the shape a busy manager actually scans: 3 wins each tied to a metric, 3 risks each with a named owner, 2 specific asks, and 1 commitment for next week. Paste them into ChatGPT, Claude, or Gemini; the table further down says which model handles a paste of a full week of raw notes best as of June 2026. Pair with executive summary prompts when the report rolls up to leadership.
TL;DR
- A bare “summarize my week” hands your raw notes back as paragraphs. The reader still has to hunt for the decision, the risk, and the ask — so they stop reading.
- The fix is structure-forcing prompts. The default shape that lands with managers is wins (with a metric) / risks (with an owner) / asks (specific) / next-week commitment. Reclaim, Asana, and Teamwork’s 2026 status-report guides all converge on the same skeleton: accomplishments, risks/blockers, upcoming work.
- All 13 prompts below run in any chat model. A full week of raw notes is small (~5K–15K tokens), so context window rarely bites — it only matters on the director-level rollup (#12) that ingests 6 manager reports at once.
- For that rollup, Claude Pro ($20/mo, $17 annual) and Gemini 3.1 Pro (Google AI Pro, $19.99/mo) both expose a 1M-token window in-app; ChatGPT Plus ($20/mo) caps the in-app window near 320 pages, full 1M only on the $200 Pro tier (all as of June 2026).
- The one rule that survives every format: a win without a number is not a win. “Made progress on checkout” tells the reader nothing.
Best for
- Engineering and PM weekly status
- Design and cross-functional updates
- Manager-to-manager and director-level rollups
- Founder weeklies to investors
- Edge cases: new-hire week 1, launch week, behind-schedule, boring week, and vacation week
Which AI to paste these into (June 2026)
Every prompt here runs in any chat model. One practical question decides which: will the model ingest your inputs in one paste? A single person’s week is small, but a director rolling up 6 manager reports, or a founder pasting a full metrics dashboard export, can run long.
| Tool | Plan / price (June 2026) | In-app context | Best for weekly reports |
|---|---|---|---|
| Claude Pro | $20/mo ($17 annual) | 1M tokens | Tightest wins/risks/asks structuring; multi-report rollups (#12) |
| Gemini 3.1 Pro (Google AI Pro) | $19.99/mo | 1M tokens | Reports built from Google Docs / Sheets metrics |
| ChatGPT Plus | $20/mo | ~320 pages in-app | General drafting; one person’s week at a time |
| ChatGPT Free | $0 | smaller, capped | Occasional reports; tighter message limits, ads on US Free |
| Claude Free | $0 | limited (Sonnet 4.6) | Occasional reports; no rollup-scale pastes |
Notes on the numbers: one person’s weekly inputs (standup logs, ticket titles, a few PR links) run roughly 5K–15K tokens, so every model above fits a single week comfortably. Context only constrains you on the director rollup (#12, six manager reports) or a founder weekly paired with a long dashboard export (#13) — there, Claude Pro and Gemini 3.1 Pro’s 1M in-app window is the practical edge. For raw structuring quality, Claude Opus 4.7 tends to produce the cleanest “cut the filler, keep the outcome” output.
Prompt anatomy
A weekly-report prompt that produces something worth sending carries five elements:
- Inputs: the raw material — standup logs, ticket titles, PR links, dashboard numbers. Paste it; do not summarize it yourself first.
- Audience: manager / skip-level / investor / cross-team. Each wants a different altitude.
- Output shape: wins / risks / asks / next-week — or a scenario-specific variant (launch week, behind schedule).
- Length cap: a word or page limit. Uncapped, the model pads.
- Anti-filler rule: an explicit instruction to drop the activity log and keep only outcomes. This is what separates a report your boss reads from one they archive.
1. WROUT format (Wins / Risks / Outlook / Updates / Tasks)
Draft my weekly report in WROUT format. Inputs: [paste raw notes / standup logs / Linear tickets]. Each section 3-5 bullets, each bullet ≤25 words. Mark each risk with severity (low / med / high) and a named owner.2. Manager-friendly 1-page report
Compress my work this week into a 1-page report my manager will actually read. Sections: 3 wins (each with a metric), 3 risks (each with mitigation), 2 specific asks, 1 commitment for next week. Cut all activity-log filler.3. Engineering weekly with metrics
Draft an engineering weekly report. Include: shipped (with PR / commit link), in flight (with % complete + ETA), blocked (with what unblocks it), and 1 chart-worthy metric (latency / error rate / on-call load / build time). ≤300 words.4. Founder weekly to investors
Draft a 250-word founder weekly to investors. Sections: 3 wins, 3 challenges, 2 specific asks (intros / advice / hires), key metric [MRR / ARR / users / retention]. Tone: honest, not promotional. Bad news first, addressed head-on.5. Cross-functional rollup
I lead 3 sub-teams: [names]. Draft a rollup weekly that names each team's top win, top risk, and any cross-team dependency. ≤200 words. Add 1 line of "decisions I need from my manager / leadership" at the bottom.6. Activity-log → outcomes converter
Below is my raw activity log for the week. Convert each entry from "what I did" to "what outcome it produced". Drop entries that produced no outcome. Group remaining entries by impact tier.
[paste log]7. “Boring week” honest report
I had a boring week — mostly maintenance and 2 small fixes. Draft an honest 150-word report that does not inflate. Explain why the quiet week is OK (planned, deserved, or a different team is the bottleneck) and what next week's priorities are.8. Behind-schedule weekly
My project is behind schedule. Draft a 200-word weekly report that owns the slip in the first sentence, names the cause specifically (not "unforeseen issues"), states the new ETA, and lists the 2 things I'm doing about it. Avoid blame language.9. New-hire week-1 weekly
I just joined the team this week. Draft a 150-word week-1 weekly report: what I learned, who I met, what I shipped (small things count), what I'm still confused about, what I'd like help with next week. Earnest tone.10. Project-launch week weekly
I just launched [project]. Draft a 200-word launch-week weekly: outcome metric vs target, what surprised us (good and bad), top 2 follow-ups, 2 lessons for the next launch. Avoid victory-lap tone — assume a skeptical reader.11. Vacation / no-show week
I was out 3 days this week. Draft a 100-word weekly that acknowledges the gap, names the 1 thing I shipped before leaving, what's continuing without me, and what next week will pick up.12. Director-level rollup with manager summaries
Below are 6 manager weeklies. Roll them into a 300-word director-level summary. Sections: top 3 wins across orgs, top 3 risks, top 3 asks. Pick the items that would land for an exec audience — anything below "important to a director" gets cut.
[paste]13. Weekly with attached metrics dashboard
My report is paired with a metrics dashboard ([URL]). Draft a 200-word weekly that interprets the dashboard — what moved, why it moved, what it means for next week. Do not re-state the numbers; explain them and flag the 1 number readers will misread without context.How to push results further
- Always paste the raw inputs, not your own summary. If you pre-summarize, the model just reformats your filler. Give it standup logs and ticket titles and let it find the outcomes.
- Run #6 first, then #2. Convert the activity log to outcomes (#6), then compress the survivors into the 1-page report (#2). Two passes beat one because the first pass kills the filler the second pass would otherwise keep.
- Attach a metric to every win, every week. A win the model can’t quantify usually isn’t one. Make it ask you for the number if you forgot it.
- Put the asks in a labeled section, top or bottom. Asks buried mid-paragraph get missed; that missed ask is next week’s “surprise” risk.
- Reuse one format, refresh the content. Cross-team consistency beats per-week creativity — but copy-pasting last week with cosmetic edits is the fastest way to get tuned out.
- For the metrics interpretation in #13, pair with data interpretation prompts when the dashboard needs real reading, not just restating.
Common mistakes
- Dumping the full activity log. Readers stop at paragraph 2. The log is your input, not your output.
- No metric or outcome attached to wins. “Made progress on X” tells the reader nothing.
- Hiding risks until they explode. Every “surprise” risk is a credibility hit. Surface it early in the “risks” section, with an owner.
- Asks buried in the middle paragraph. Put them in a labeled “Asks” section at the top or bottom.
- Same report copied week to week with cosmetic edits. Readers notice and tune out.
- Listing meetings attended as wins. Meetings are activity, not outcome.
FAQ
- Which AI writes the best weekly report? For a single person’s week, any of the three is fine — the inputs are small (~5K–15K tokens). On the director rollup (#12, six manager reports) or a founder weekly paired with a long dashboard export, Claude Pro and Gemini 3.1 Pro’s 1M-token in-app window is the practical edge over ChatGPT Plus’s ~320-page in-app cap (June 2026). For raw “cut the filler, keep the outcome” quality, Claude Opus 4.7 tends to produce the cleanest structure.
- Can I do this on a free plan? Yes. ChatGPT Free and Claude Free both run these prompts; the limits are message caps and smaller context, not capability. A single week’s report fits free tiers easily. The paid tiers matter only for multi-report rollups and heavy weekly use.
- How long should a weekly report be? Shorter than you think. A 1-page manager report (#2) or a 200-word founder weekly (#4) is the right length. If it scrolls, the wins are competing with filler and the reader skims past both.
- Should I include bad news? Yes, first. The behind-schedule (#8) and launch-week (#10) prompts open by owning the problem deliberately. A report that only ever contains wins reads as either dishonest or out of touch; a named risk with an owner reads as in control.
- Will my manager know I used AI? Not if you paste your real inputs and verify the output. The prompts structure your facts — they don’t invent them. The tell isn’t AI; it’s a generic report with no specific metrics, which a vague prompt and a lazy human produce just as easily.
- What about confidential metrics? Don’t paste unannounced revenue, headcount, or customer PII into a consumer AI tool without checking its retention/training policy. Redact the dollar amounts, or use a business/enterprise tier with a no-training guarantee.
Related
- Monthly business review prompts
- Meeting notes prompts
- Executive summary prompts
- Project planning prompts
- Action item extraction prompts
- AI KPI weekly report
External references: Reclaim’s 2026 weekly status report guide and Asana’s status report template for the standard accomplishments / risks / upcoming-work skeleton.