You now have three moves. Lessons 7 and 8 taught you how to handle a single response. This lesson adds a third strategy that often skips the problem entirely.
Reading
Exercise
Version picker
XP
The single-version trap
You've now seen how iteration works — and when to stop it. But there's a pattern that trips people up even when they're doing everything right: spending six rounds trying to perfect one response when two messages and three options would have been faster.
Round one: too formal. Round two: better, but too long. Round three: good length, but the opening is weak. Round four: the opening is fixed but now the tone has shifted again. Each correction improves one thing and nudges another.
The ceiling problem. Iterating a single version has diminishing returns. You're always working with what the AI committed to in round one. Sometimes you need to see three different starting points before you know which direction is actually right.
Options as a strategy
Asking for multiple versions isn't indecision — it's a deliberate technique. When you don't know exactly what you want, generating options and choosing the best one is almost always faster than iterating a single version toward an unclear target.
The prompt is simple. Instead of asking for the thing, ask for three versions of the thing:
Instead of: "Write a subject line for this email."
Try: "Give me five subject lines for this email — vary the tone across them. One direct, one curiosity-driven, one warm."
Now you're choosing, not correcting. Choosing is fast. You know what you want when you see it — you just needed something to react to.
See the difference
Same task — writing a bio for a new job application. Two approaches, side by side:
⚠ Iterating one version — 6 turns
U
Write me a professional bio.
AI
Sarah is a dedicated professional with extensive experience in her field...
↳ Too generic, too formal
U
Make it less formal.
AI
Hey! I'm Sarah — a passionate pro who loves what she does...
↳ Now too casual, wrong audience
U
Not that casual. Something in between.
AI
Sarah brings a thoughtful approach to her work, blending expertise with a human touch...
↳ Better, but now it's too vague again
3 turns in and still patching. The AI keeps overcorrecting because "in between" is not a real instruction.
✓ Asking for options — 2 turns
U
Write 3 versions of a bio for a UX designer moving into product. Vary the tone: one formal, one conversational, one punchy.
AI
Formal: Sarah Chen is a UX designer transitioning into... Conversational: I've spent 6 years making products easier... Punchy: UX → Product. I know both sides.
U
Use the conversational one but take the last line from the punchy version.
3 options in one response. One round of reaction. Done — and the result is genuinely better because you chose rather than corrected.
Pick and combine
The real power isn't in picking the best version — it's in combining parts across versions. The formal one has the right opening. The casual one has the right middle. The punchy one ends well. That's three instructions collapsed into one follow-up.
The combine instruction: "Use version 2 as the base, but take the opening line from version 1 and the closing from version 3." This is faster than explaining tone preferences in the abstract — you're pointing at actual text and saying what to keep.
Asking for options turns a vague goal into a concrete choice. And concrete choices lead to concrete instructions — which is exactly what the AI needs to do its best work.
Exercise
Rewrite your L7 prompt as an options request
Your improved prompt from Lesson 7 is on the left. On the right, rewrite it as an options-style request — ask for multiple versions that vary in some dimension. Then tell us what varies.
↩ Your Lesson 7 prompt
Loading your prompt...
Your options-style rewrite
What will vary across your versions?
Pick at least one — or type your own below.
Version picker
Three versions. One task.
Here are three AI-written bios for the same brief: a UX designer moving into product management. Pick your favourite — and steal one thing from another. Then watch the combine instruction write itself.
Formal
Sarah Chen is a product-minded UX designer with six years of experience building user-centred digital products. Having led design across three product teams, she is now transitioning into product management — bringing a deep understanding of user needs, cross-functional collaboration, and evidence-based decision making to the role.
Casual
I've spent six years designing products that actually make sense to use — and now I'm stepping into the role where the real decisions get made. I'm a UX designer moving into product management, and I'm bringing everything I've learned about users, teams, and shipping things that work.
Punchy
UX designer. Six years. Now: product. I know what users need because I've spent years watching them struggle with what product managers built. Time to be in the room where those decisions happen.
✦ Your combine instruction
This is what you'd send to the AI as your next message. Pointing at real text is always clearer than describing what you want in the abstract. That's the whole lesson.
🔁 Level 3 complete — The Iterating
+40
XP Earned
✦ 0 XP total
+30Lesson 7
+30Lesson 8
+40Lesson 9
100Level 3 total
Next up: Level 4 — Context is Everything
You've mastered iteration. Now you'll learn how the AI's memory actually works — what it remembers, what it forgets, and how to use context deliberately to get consistently better results.
You have all three moves now 🔁
Fail fast, iterate smart, ask for options. That's the full Level 3 toolkit.