BXUP Track progress & test yourself — free
AI at Office · Playbook · Foundation · ~4 min read
AI AT OFFICE · PLAYBOOK · FOUNDATION

Generate Status Updates and Weekly Reports from Work Logs

Turn a scrappy work log — Slack scraps, ticket IDs, half-sentences in a notes app — into a crisp weekly status update: done, blocked, next, risks surfaced, written at the altitude your boss actually reads instead of the altitude you lived through.

Tier: Foundation · Hands-on time: ~25 minutes · Tools: ChatGPT, Claude, Google Docs

The hard part of a status update was never the writing. It is deciding what your reader needs, compressing a week of noise into six lines, and saying "blocked" out loud. This recipe makes the AI do the compression while you keep the judgment.

When to use — and when not to

  • Weekly or fortnightly status updatesyou keep any kind of running log (notes doc, task tracker, sent-mail folder) and owe a recurring summary to a manager or stakeholder group.
  • Project checkpoint reportsa steering-committee or client update where the same facts must be re-pitched at a higher altitude than your team update.
  • Handover or return-from-leave briefscompressing two weeks of activity into one page for whoever picks up the thread.

And when not to:

  • You have no log. The model cannot recover a week you did not record. Two minutes a day of bullet-scraps is the real prerequisite; start that habit today, use this recipe next week.
  • The update carries bad news that needs a conversation. A slipped deadline or a personnel issue should reach your manager by voice first, in writing second. Do not let a polished paragraph be the first they hear of it.
  • The update is three lines long. If a week genuinely reduces to "shipped X, on track," type it yourself.

The recipe

  1. Dump the raw log, unedited (3 minutes). Collect the week's raw material into one text block: notes-app scraps, ticket titles you closed, calendar entries, key sent emails. Do not clean it up — cleaning is the model's job. Do strip anything confidential you would not paste into a hosted tool (salary figures, personal data, client secrets).

  2. Generate the first draft with a strict frame. Paste the log with this prompt:

    You are helping me write my weekly status update. Below is my raw,
    messy work log for the week. Turn it into a status update with
    EXACTLY this structure:
    
    1. Headline: one sentence — the single most important thing my
       manager should know this week.
    2. Done: 3-6 bullets, each starting with a verb, each stating an
       outcome (not an activity). Merge related items.
    3. Blocked / Needs decision: bullets naming WHO I am waiting on
       and WHAT I need from them. If nothing, write "None".
    4. Next week: 2-4 bullets of committed priorities.
    5. Risks: anything in the log that could slip or surprise. If
       nothing, write "None flagged".
    
    Rules:
    - Use ONLY what is in the log. Never invent progress, numbers, or
      dates. If an item is ambiguous, mark it [CHECK] instead of guessing.
    - Audience: my manager, who has 60 seconds. No jargon-decoding
      needed for ticket IDs — drop them or translate them.
    - Tone: plain, confident, no filler ("various", "ongoing", "touched
      base") and no self-praise.
    
    MY LOG:
    [PASTE WORK LOG]
    

    A good draft looks like this:

    Headline: Vendor migration is a week ahead of plan; the only open
    risk is the unsigned data-processing agreement.
    
    Done
    - Migrated 3 of 4 vendor data feeds to the new platform (finance
      feed remains).
    - Closed the Q3 access review — 47 accounts certified, 2 revoked.
    - Drafted the DPA and sent it to Legal for review.
    
    Blocked / Needs decision
    - Waiting on Legal (Priya) to return DPA comments — needed by
      Thursday to hold the cutover date.
    
    Next week
    - Migrate the finance feed and run parallel reconciliation.
    - [CHECK] Confirm cutover date of 14th with the vendor.
    
    Risks: None flagged beyond the DPA timing above.
    
  3. Resolve every [CHECK] and audit the claims (5 minutes). Each [CHECK] is a fact the model could not confirm from your log — confirm or delete it. Then verify every number, date, and name against your own records. The model compresses honestly only if you audit honestly.

  4. Re-pitch for a second audience if needed. Same conversation:

    Now rewrite this same update for [AUDIENCE, e.g. the client steering
    committee]. Keep every fact identical. Raise the altitude: lead with
    business impact, cut internal ticket detail, keep it under 120 words.
    
  5. Paste into your template and make one human edit. Move the result into Google Docs or your team's format, then rewrite at least the headline in your own voice. The headline is the one line your manager reliably reads — it should sound like you.

Make it yours

  • ChatGPT or Claude with a reusable setup: save the step-2 prompt as a Project or a custom instruction ("You are my status-update formatter…") so each week is just pasting the log. After three weeks, paste the previous updates too and ask it to keep continuity: "last week said X was next — report against it."
  • Google Docs (Gemini) where your log already lives: keep a running "Work log" doc, add entries daily, then use "Help me write" with the step-2 rules to draft in place. Weakest at strict structure — paste the exact rules rather than a loose ask.
  • Scale up to a team roll-up: collect each member's update, then prompt "Merge these five updates into one team report: shared headline, cross-team blockers first, credit by name for major items." Manager judgment still decides what the shared headline is.

Quality check

  • Every fact traces to your logno invented progress, no rounded-up percentages, no dates the model inferred.
  • "Done" bullets state outcomes, not activity ("cut invoice errors 12%" beats "worked on invoice process").
  • The blocked section names a person and an aska blocker without an owner is a complaint, not an update.
  • Last week's "next" items are accounted fordone, moved, or explicitly dropped. Readers notice silent disappearances.
  • 60-second test: a reader who sees only the headline and bullet leads still gets the true story of the week.
Warning

A work log is full of names, deal details, and internal metrics — use your organisation's approved AI tenant, and strip client identifiers and personal data before pasting into a personal account.

Pro Tip

Keep every generated update in one running doc and, at review time, ask the model to summarise the quarter from it — the same recipe, one altitude higher, and your self-assessment writes itself from evidence.

This is the free reading view. Open BXUP to track your progress, run the hands-on labs, and prove it with original, exam-pattern assessments.
© Anvi Innovate · BXUP — Be Exponentially Upskilled · BXUP is an independent learning platform, not affiliated with or endorsed by AWS, Microsoft, Google, NVIDIA, IBM, Salesforce, Databricks, Oracle or Anthropic. All trademarks belong to their respective owners. · bxup.in