Your first AI conversation
Use short speech loops to learn from replies
Revise one named weakness at a time, then check the whole draft again.
1Learn the idea
Read
The conversation loop
See it
Role + task + context + format = clearer output
- Draft from bounded source material.
- Check against the quality bar.
- Name one failure and request a specific revision.
- Compare versions.
- Verify facts and approve, revise, or stop.
Mira’s welcome is accurate but too dense to read aloud. She asks: “Keep every confirmed fact. Use sentences under 18 words and add signposting for three sections.” She then reads the revision aloud and checks that no fact disappeared.
Read
Misconception
“More rounds always improve the answer.” Revisions can introduce new errors or erase good material. Keep versions and stop when the draft meets the predefined bar.
Read
Guided practice
Revise this feedback into an actionable turn: “I don’t like it.” Name the audience effect, the element to preserve, and the change requested.
Read
Independent transfer
Run a two-turn revision on a low-risk draft. Keep a tiny log: failure, requested change, observed result, decision. If the second version is worse, restore the first.
Go deeper
In the wild
See how this idea shows up as a product and a company — then come back to the lesson. Skills transfer across vendors.
Related lessons
Check your understanding
Page assessment
Answer from memory. Completion is saved from this evidence, not from opening the next page.
Why must Mira recheck facts after a style-only revision?
All responses are required.