Reviewable Revision Packet
Topic 4 homework: show that an AI-assisted revision was permitted, protected, reviewed, disclosed, and understood.
Your own Draft 1
Start from the Draft 1 you wrote in class. AI asks and points; it never writes or rewrites your sentences. Share only your own draft with a tool: no names, nothing private.
One consolidated evidence packet
Complete Parts 1–6 in order; Part 4 is the only feedback log. Submit one PDF as Lastname-Topic04-RevisionPacket.pdf through eCampus.
/grill-me in Claude Code (plan mode), or the coach card on your own (the six questions on page 2 of your Tuesday sheet). Same rubric, same credit.
Lastname-Topic04-RevisionPacket.pdf, on eCampus. Submit your completed Parts 1–6, but do not submit the later “Completed Example” section. If PDF conversion is inaccessible, request the instructor-approved accessible equivalent.
Local draft ready.
Part 1 — Permission and privacy
What the assignment allowed AI to do (coach only), and what you shared with a tool (your own draft only; no names, nothing private).
Privacy: Claude Code sends your draft to Anthropic through your account, so use only your own, non-private draft. The coach-card route, on your own, earns equal credit.
Part 2 — Plan and Draft 1
Your lane and problem, what you saw, the coach’s questions, the five-box outline, and Draft 1 as written in class (a photo or a typed copy; anything added later is marked).
Part 3 — Final pitch
250–400 words, every sentence yours, addressed to the person who can say yes.
Part 4 — Feedback log
Three questions from /grill-me or the coach card: one acted on (quote your new sentence), one adapted, one set aside. Each has your reason. This is the only log; do not duplicate it.
| Decision | The question | What you did | Your reason |
|---|---|---|---|
| ACT ONthe question, and the sentence you wrote | |||
| ADAPTthe question, and what you did instead | |||
| SET ASIDEthe question, and why it does not fit your reader or your facts |
Part 5 — Verification, disclosure, defense
Each fact or number, with where you saw it or how you checked it. Disclosure by stage, in the frame “I used ___ at ___ to ___; afterwards I ___.” A two-sentence defense of your strongest reason, written without AI.
Part 6 — Before / After / Why
One passage from Draft 1, the final version of it, and why it reads better for this reader.
Submission check
Quick vocabulary
Authorized source
Material the assignment and the content owner permit you to use for this task.
Approved tool
A product the instructor or institution has explicitly named for this assignment; approval does not authorize every document.
Agency
Who performs an action, makes a decision, or remains responsible in the sentence.
Claim strength
How certain, broad, or causal a statement is relative to its evidence.
Grading Rubric
| Criterion | Points |
|---|---|
| Writing for the reader: purpose, organization, flow, specificity, voice | 30 |
| Process traceable from Draft 1 (Parts 2 and 6) | 20 |
| Feedback decisions (Part 4) | 20 |
| Verification, disclosure, defense (Part 5) | 20 |
| Permission and privacy (Part 1) | 10 |
| Total | 100 |
Completed Example
Use this to understand the evidence standard. Do not copy it as your own submission.
Sam is the lost-and-found writer from class (Tuesday’s Pitch B). Sam saw two things: four “lost AirPods” posts in three weeks, and twice the missing AirPods already in the front desk’s box. The example uses no other facts about what Sam saw.
1 — Permission and privacy
What the assignment allowed AI to do: Coach only: AI could ask, critique and point. It never wrote or rewrote my sentences.
What I shared with a tool: On Tuesday, a chatbot got my outline notes. On Thursday, Claude Code got draft1.md. At the end, a spelling check got the final pitch, with my name left off. All three were my own words, with no names and nothing private. Claude Code sends my draft to Anthropic through my account, so I checked the file before I ran /grill-me.
2 — Plan and Draft 1
Lane and problem: LEFT OUT. Commuters can’t see what is already in the front desk’s lost-and-found box.
What I saw: Four “lost AirPods” posts in our commuter group chat in three weeks. Twice, the missing AirPods were already in the front desk’s box. That is everything I saw; the rest of my pitch is reasons and a request.
The coach’s questions (Tuesday, from the coach card on the screen): “Who can actually say yes?” The front desk: the box is theirs. “What’s the smallest version that could start this month?” A list, not an app.
The five-box outline (notes):
- Problem: 4 “lost AirPods” posts in our chat in 3 weeks.
- Fix: a one-screen found-items list, each afternoon.
- Two reasons: twice the item was already in the box; about 2 minutes a day.
- The strongest “no”: a list could help someone claim AirPods that aren’t theirs. Answer: keep it general; owners describe the item.
- Ask: the front desk; the list each afternoon; by when? Weakest box. My move: name a start day.
Draft 1, as written in class (typed copy; nothing added later):
To the front desk: in three weeks, four “lost AirPods” posts in our commuter group chat—twice the item was already in your box. Could you post a one-screen found-items list each afternoon? Two minutes of typing.
Sam’s Draft 1 is Tuesday’s short Pitch B. Yours is the 200–300-word pitch you wrote in class.
3 — Final pitch
To the front desk team,
I’m a commuter student, and I have a small idea for your lost-and-found box.
In the last three weeks, I counted four “lost AirPods” posts in our commuter group chat, and twice I saw the missing AirPods already sitting in your box. Both times, someone was asking a group chat for something you were already keeping for them. I can only speak for the AirPods I saw, and you would know better than I do whether other things wait in the box the same way. Either way, right now nobody in the chat can see what’s in your box.
A short daily list would connect the two. Each afternoon, you would type one line for each item in the box, such as “AirPods case, found Tuesday,” on a single screen. I’d guess that takes about two minutes, though I haven’t timed it. Anyone who loses something could check the list before posting, and anyone whose item is on it would know to come to your desk.
The strongest reason to say no is that a shared list might help someone claim AirPods that aren’t theirs. The list can stay general, though, and you can ask whoever comes for an item to describe it before you hand it over.
Here is my request: could you reply to this email each afternoon with that one-screen list, starting Monday, for two weeks? I’ll share it in our commuter group chat, so the only work on your side is the typing. I’ll also keep counting the “lost” posts, and after the two weeks I’ll tell you what changed. Then you can decide whether to keep the list going.
Thanks for reading.
Sam
283 words, including the greeting and the name.
4 — Feedback log
From one round of /grill-me questions on draft1.md, three logged: one per decision.
| Decision | The question | What I did | My reason |
|---|---|---|---|
| ACT ON | “Where does your evidence come from? Did you count those posts yourself, and how do you know the AirPods were already in the box?” | New sentence: “In the last three weeks, I counted four ‘lost AirPods’ posts in our commuter group chat, and twice I saw the missing AirPods already sitting in your box.” | The desk doesn’t know me. Saying I counted the posts and saw the box myself tells them these are facts I can stand behind, not rumors from a chat. |
| ADAPT | “Is your ask one request or two? Are you asking the desk to make a list, to get it to students, or both?” | I kept one request for the desk, the list sent as a reply each afternoon, and took the second job myself: I share it in our group chat. | Right idea: a list nobody sees fixes nothing. Wrong fix for my reader: asking the desk to post in our chat as well would add a second job to a two-minute favor. |
| SET ASIDE | “Who is your reader? Can the person at the front desk say yes to this, or does someone else decide?” | Kept the front desk as my reader. | It doesn’t fit my facts or my reader. I know of no one above the desk and won’t invent a title; the box is theirs and the job is about two minutes of their typing, so they can say yes. Writing over their heads would make a small favor sound like a complaint. |
5 — Verification, disclosure, defense
| Fact or number in my pitch | Where I saw it or how I checked it |
|---|---|
| Four “lost AirPods” posts in our commuter group chat in three weeks | I counted them myself, in the chat. |
| Twice, the missing AirPods were already in the box | I saw them in the box myself, both times. |
| About two minutes a day | My estimate, not a measurement, so the pitch says “I’d guess” and “I haven’t timed it.” |
Everything else in the pitch is a reason or a request, not a fact.
Disclosure by stage:
- I used a chatbot at the outline stage, on Tuesday, to improve my outline; afterwards I rejected the outline it wrote, asked it for three questions instead, and answered them in my own notes.
- I used
/grill-mein Claude Code, in plan mode, at Draft 1 on Thursday to ask where a reader gets lost; afterwards I picked three of its questions (Part 4) and rewrote the pitch in my editor, in my own words. - I used a spelling check at the final pitch to flag misspelled words; afterwards I fixed each one myself.
Defense (written without AI): My strongest reason is that twice, the AirPods someone was asking our chat about were already in the desk’s box, and the person asking couldn’t see that from the chat. I saw both cases myself, so I can explain them without notes or AI.
6 — Before / After / Why
Before: “Could you post a one-screen found-items list each afternoon? Two minutes of typing.”
After: “Here is my request: could you reply to this email each afternoon with that one-screen list, starting Monday, for two weeks? I’ll share it in our commuter group chat, so the only work on your side is the typing.”
Why: Draft 1 asked for a list but never said where it goes or when it starts, and “Two minutes of typing” was a fragment the desk had to decode. The new ask is one request with a start and an end, and it shows this reader exactly how small their part is.