Write to Think: What Amazon's New AI Memo Habit Gets Wrong
"How can I tell what I think till I see what I say?" That line isn't just clever. It's a creative operating system for anyone who makes a living with words.
For years, Amazon baked that idea into its meetings. No slides. Just six-page narratives, read in silence, so people had to think clearly before they spoke. The memo wasn't documentation. It was a thinking tool.
Now there's a shift. The company's internal chatbot, Cedric, can spin up "six-page narratives in seconds." Efficient? Sure. But it skips the part that matters: the hard, productive struggle of finding out what you actually mean.
Writing is thinking (and skipping it is expensive)
As writers, we don't pay the bills with keystrokes. We pay them with clarity. The memo discipline worked because the act of drafting forces trade-offs, priorities, and logic to surface. By the time you speak, you've wrestled with the work.
Outsourcing that wrestle to a chatbot swaps productive effort for surface polish. It feels faster. It is also emptier. You get a document that says something, but not necessarily what you mean.
AI is enabling and disabling
LLMs widen your reach, compress research, and help you stress-test arguments. They also create a trap: if you let them do the first pass, you weaken the habit that builds judgment. We've seen this story before. GPS reduced our need to form mental maps; the brain adapted by offloading the load.
If you care about durable craft, protect the part of the process that hurts in a useful way. That first messy draft is the gym. Skip it and your thinking atrophies over time.
Related research: London taxi drivers developed measurable brain changes through demanding navigation practice. Tools change us as much as they help us. See Navigation-related structural change in the hippocampi of taxi drivers in PNAS for a classic example. Read the study.
Use AI without outsourcing your brain
Here's a simple rule: human-first draft, AI-second pass. Keep authorship of the argument. Use the model to expose blind spots and tighten language-after you've done the thinking.
- Zero-draft by hand: bullet your thesis, stakes, constraints, and counterpoints. Don't open the chatbot yet.
- Write an ugly first narrative in full sentences. Aim for meaning, not elegance.
- Only then, ask AI to critique: "Where is this vague, contradictory, or unsupported? What would a smart skeptic say?"
- Synthesize its best hits back into your draft. Keep anything that sharpens intent. Delete anything that dilutes it.
- Final pass: your voice, your structure, your call-to-action. If it doesn't sound like you, it isn't done.
A 30-minute "memo sprint" for writers
- 5 minutes - Intent: What's the one claim? Who needs it? Why now?
- 10 minutes - Outline: Problem, evidence, counterargument, decision or takeaway.
- 10 minutes - Narrative: Full sentences, no polishing. Get to six tight paragraphs, not six pages.
- 5 minutes - AI critique: Ask for edge cases, missing data, and tighter phrasing. Merge, don't merge blindly.
This preserves the part that builds clarity and still lets you move fast.
Team guardrails that keep thinking intact
- No AI-first memos. The person responsible for the decision writes the first draft.
- Silent-read rule stands. Everyone reads before anyone speaks.
- Label AI assistance. Note where models were used and what changed because of them.
- Separate idea generation from decision language. Brainstorm with AI; decide in your own words.
- Archive memos as institutional memory. Your drafts become training data for the team's judgment.
For students, pros, and teams: the assignment is the experience
In classrooms, the essay isn't the product. The thinking is. Same for client decks, book proposals, and product briefs. If a machine writes the draft, you've dodged the exercise that was supposed to change you.
That's the real cost. Not a moral failure-an opportunity cost. You robbed future-you of better judgment.
Where AI actually helps
- Compression: Summarize long research into notes you can interrogate.
- Counterarguments: Ask for the strongest case against your point.
- Clarity passes: Shorten sentences, surface jargon, fix rhythm-after you set the meaning.
- Scenario mapping: "If this is true, what follows in 3, 6, 12 months?"
If you want structured ways to keep authorship while using tools well, see AI for Writers for approaches that treat models as amplifiers, not replacements.
The thesis in one line
The thing that makes AI attractive-it spares you the struggle-is the same thing that makes it weak as a thinking partner. Keep the struggle. That's where clear writing (and better decisions) come from.
Your membership also unlocks:
