Turning AI Ideas into Buildable Prompts

Discussion prompt

Before we open Turning AI Ideas into Buildable Prompts: without looking back, what was the main idea of Concepts, Tools & the Mapping Worksheet, and what could you do by the end of it that you could not do before?

Hint: One sentence for the idea, one for the skill. If the second one is blank, that is the part to revisit.

Comparison matrix

From Recap: no garbage in, garbage out: refill the What comes back column from what you know. The rest of the table is as it appeared.

Match the pairs

From The four moves, one more time — match each one to what it actually does. The descriptions have been shuffled.

• c1. 1 · Map
• c2. 2 · Check
• c3. 3 · Pick
• c4. 4 · Build

• b1. Fill the six worksheet fields for the idea.
• b2. Any field you can't answer is a gap to close first.
• b3. Let Reuse choose: chat, Artifact, Skill, or API.
• b4. Ask precisely, judge the result by your Success test.

Why: 1 · Map, 2 · Check, 3 · Pick, 4 · Build are easy to tell apart while they are sitting next to their descriptions and much harder afterwards, which is what this checks.

Discussion prompt

You know the six fields. Today each one becomes a pass/fail question you ask an idea. An idea that passes all six is ready to build; every fail is your next to-do.

That is stated as though it always holds. Do one of two things: produce a case where it fails, or say precisely what rules such a case out. "It just does" is not on the menu.

Hint: Hunt at the extremes first — zero, one, negative, empty, equal. If every extreme survives, the reason they survive is the proof.

Discussion prompt

Read the picture before the words. What is this showing, and what is the one thing it is built to make obvious? Commit to an answer, then read on.

Hint: Name the parts, then say what changes between them — and if nothing changes, say what is being held still.

Discussion prompt

Explain Every blank is a guess you outsourced by analogy to something with no AI Foundations in it at all — a queue, a recipe, a map, a bank balance, whatever fits. Then say where your analogy breaks.

Hint: An analogy that never breaks is not an analogy, it is the same idea wearing a hat. Find the seam — that is the part that is actually new.

Put in order

Put the moves of Audit a fuzzy idea, live into the order they have to happen.

1. Goal: "help with my listings" - help HOW? Rewrite them? Grade them? Translate them?
2. Inputs: which listing? The current text? The product facts? Photos?
3. Outputs: no shape at all - a rewrite? a score? a list of fixes?
4. Constraints and Success test: nothing yet - nothing to constrain until the goal exists.
5. Reuse: this one we CAN answer - every new product, so repeated.

Why: These are the moves of the worked example in the order it makes them, and each one is set up by the one before it. Not one checkable job. FAIL - and it's the most important fail, because every other field depends on it.

Discussion prompt

Work backwards. The example finished here:

Reuse: this one we CAN answer - every new product, so repeated.

What was it asked to do, and what must it have been given? Reconstruct the problem from its answer.

Hint: Every quantity in the result had to enter somewhere. Account for each one.

Predict first

A student writes this, and it looks reasonable:

Goal: "A tool that rewrites my listings and tracks my weekly sales and plans seasonal inventory."

It is wrong. Say what breaks — and say it before you turn the page.

Correct: Three jobs need three different inputs and three different output shapes - the worksheet jams immediately.

Split at every "and": three one-sentence goals, three small tools.

Why: Three jobs need three different inputs and three different output shapes - the worksheet jams immediately.

Comparison matrix

From Your turn: audit your first idea: every row here is a choice with a cost. Fill the Verdict column, then say which row you would actually pick and what you give up for it.

Comparison matrix

From A prompt is the worksheet, re-worn: refill the Comes from column from what you know. The rest of the table is as it appeared.

Discussion prompt

Explain A chat is a negotiation. A prompt is a contract. to a student a year behind you. No notation, no jargon they have not met — and it still has to be true.

Hint: If your explanation needs a symbol they have never seen, you are describing the notation rather than the idea.

Put in order

Put the moves of The model conversion: meeting notes -> prompt into the order they have to happen.

1. Goal becomes the Role: "You are my meeting-notes assistant. Your job: turn my raw notes into a clean action-item list."
2. Inputs and Outputs become the INPUT and OUTPUT lines - announced before any notes appear.
3. Constraints become RULES, and the never-invent rule gets a partner: what to do INSTEAD.

Why: These are the moves of the worked example in the order it makes them, and each one is set up by the one before it. The one-sentence goal IS the role sentence - that's why we insisted on one sentence.

Comparison matrix

From The model conversion: meeting notes -> prompt: every row here is a choice with a cost. Fill the Answer column, then say which row you would actually pick and what you give up for it.

Predict first

Now the tightened listing idea from Part 1 - the worksheet passed all six, so the conversion is mechanical:

Commit before you compute: what does Same conversion, your domain: the listing rewriter come out to? A rough magnitude and the right form is enough — the point is to have something concrete to be wrong about.

Correct: Read it once more and name each part: role, input, output, rules, missing-info instruction.

Why: A prediction you can defend turns the computation into a check rather than a leap of faith — and an answer that contradicts it is caught on the spot. If you can point to all five, the prompt is structurally complete - anything still wrong is a wording tweak, not a rebuild.

Comparison matrix

From Same conversion, your domain: the listing rewriter: refill the Answer column from what you know. The rest of the table is as it appeared.

Predict first

A student writes this, and it looks reasonable:

Keep it short: "You are a copywriter. Write a great Amazon listing for this hoodie."

It is wrong. Say what breaks — and say it before you turn the page.

Correct: It was told to produce a finished listing at all costs, so it fills every gap with something plausible.

Same ask, plus two lines: "Use only the facts I give you. If a fact is missing, ask me at the top of your reply."

Why: It was told to produce a finished listing at all costs, so it fills every gap with something plausible. Plausible is not true.

Discussion prompt

The rule this trap just fixed:

You gave it a legal move OTHER than guessing - so it takes it. The gap surfaces to you instead of into the copy.

Now break it on purpose. Build a case that violates it and follow the consequences until something visibly fails. Where does the failure first show up — and would you have noticed it if you had not been looking?

Hint: The dangerous rules are the ones whose violation still produces an answer. If yours fails loudly, try to find one that fails quietly.

Discussion prompt

Explain The working distinction by analogy to something with no AI Foundations in it at all — a queue, a recipe, a map, a bank balance, whatever fits. Then say where your analogy breaks.

Hint: An analogy that never breaks is not an analogy, it is the same idea wearing a hat. Find the seam — that is the part that is actually new.

Discussion prompt

Read the picture before the words. What is this showing, and what is the one thing it is built to make obvious? Commit to an answer, then read on.

Hint: Name the parts, then say what changes between them — and if nothing changes, say what is being held still.

Put in order

Put the moves of Classify four ideas from your own shelf into the order they have to happen.

1. Product-listing rewriter -> DO again, every new product -> Skill
2. Weekly sales summary from pasted Seller Central numbers -> DO again, every Monday -> Skill
3. Seasonal inventory checklist -> OPEN again, twice a year -> Artifact
4. Vendor-facing product sheet generator -> both, in sequence -> Artifact template first, Skill if vendors multiply

Why: These are the moves of the worked example in the order it makes them, and each one is set up by the one before it. The process repeats with fresh facts each run; the old outputs don't get reopened.

Discussion prompt

Work backwards. The example finished here:

Vendor-facing product sheet generator -> both, in sequence -> Artifact template first, Skill if vendors multiply

What was it asked to do, and what must it have been given? Reconstruct the problem from its answer.

Hint: Every quantity in the result had to enter somewhere. Account for each one.

Predict first

A student writes this, and it looks reasonable:

The prompt is written - immediately save it as a Skill. It's going to be a weekly job, after all.

It is wrong. Say what breaks — and say it before you turn the page.

Correct: You're now debugging two things at once: the wording of the instructions AND the packaging around them.

The prompt is written - paste it into a plain chat first.

Why: You're now debugging two things at once: the wording of the instructions AND the packaging around them.

Sort into buckets

Some of these hold up and some are the exact mistakes this lesson is built to prevent. Sort them.

Holds up

The north star from Session 1: AI gives back the quality of what you put in. Same rule, your business this time:; Every idea you will ever have goes through the same four moves, in order:; You know the six fields. Today each one becomes a pass/fail question you ask an idea. An idea that passes all six is ready to build; every fail is your next to-do.

Breaks

Goal: "A tool that rewrites my listings and tracks my weekly sales and plans seasonal inventory."; Keep it short: "You are a copywriter. Write a great Amazon listing for this hoodie."

sound

These are stated as this lesson states them — each one survives the edge cases Turning AI Ideas into Buildable Prompts puts it through.

flawed

Each of these is lifted from a trap in this deck: reasonable-sounding, and wrong in a way that only shows up once you rely on it.

Put in order

These are the steps of The full recipe: idea -> tested tool, scrambled. Put them back in order before the next slide shows you.

1. Audit the idea with the six questions - every blank is a to-do, and every "and" in the Goal is two ideas.
2. Convert the finished worksheet into a five-part prompt: role, input, output format, rules, missing-info instruction.
3. Keep the Success test out of the prompt - it's your grading pen, not Claude's instructions.
4. Test in a plain chat on real inputs, 2-3 runs, tweaking wording between runs.
5. Grade every run pass/fail against the Success test - not by vibes.
6. Promote what passes: do-it-again -> Skill; open-it-again -> Artifact; must-run-by-itself -> API, later.

Why: This is the order the recipe itself gives. Recalling the sequence without the slide in front of you is the difference between recognising the method and being able to run it — most of what goes wrong in practice is a step done out of turn.

Discussion prompt

The full recipe: idea -> tested tool works on the cases you have just seen. Push it to the edge: what is the most degenerate input it still handles — empty, zero, one item, everything equal — and what is the first case where it stops being true? Name the case, not just "it breaks".

Hint: Try the smallest legal input, then the largest, then the one where two things collide. Methods are specified at their edges; the middle takes care of itself.

Eliminate the wrong options

A worksheet reads - Goal: "Summarize my weekly sales." Inputs: "the numbers I copy from Seller Central each Monday." Outputs: "a useful summary." Constraints: "never invent numbers." Reuse: "every week." Which field fails the audit?

3 of these 4 are wrong. Strike them one at a time, and say what rules each one out before you strike the next. The survivor is the answer.

• A. Goal - it should describe the whole workflow, not one task
• B. Inputs - copying from Seller Central is too manual to count
• C. Outputs - "a useful summary" has no shape a stranger could sketch
• D. Reuse - "every week" is too frequent for a hand-run tool

Survives elimination: C

Why: "A useful summary" is exactly the kind of blank Claude fills by taste - one week you get paragraphs, the next a table, and you can't grade either. The fix is a sketchable shape: e.g. "total revenue, top 3 products by units, and any product down 20% or more from last week." Every other field passes: one-sentence goal, a nameable paste-able input, a real never-rule, and a clear reuse answer.

Check your understanding

A worksheet reads - Goal: "Summarize my weekly sales." Inputs: "the numbers I copy from Seller Central each Monday." Outputs: "a useful summary." Constraints: "never invent numbers." Reuse: "every week." Which field fails the audit?

• A. Goal - it should describe the whole workflow, not one task
• B. Inputs - copying from Seller Central is too manual to count
• C. Outputs - "a useful summary" has no shape a stranger could sketch (correct)
• D. Reuse - "every week" is too frequent for a hand-run tool

Answer: C

Why: "A useful summary" is exactly the kind of blank Claude fills by taste - one week you get paragraphs, the next a table, and you can't grade either. The fix is a sketchable shape: e.g. "total revenue, top 3 products by units, and any product down 20% or more from last week." Every other field passes: one-sentence goal, a nameable paste-able input, a real never-rule, and a clear reuse answer.

Why A tempts people

The opposite of the real rule - a Goal should be ONE job in one sentence. "Summarize my weekly sales" passes; widening it to a workflow would ADD an "and".

Why B tempts people

Manual is fine - the audit only asks whether you can name exactly what you'll paste and where it comes from, and this answer does both.

Why D tempts people

There's no such thing as too frequent - frequent-by-hand is precisely the pattern that points at a Skill. Reuse is the field that's answered BEST here.

Predict first

Your listing prompt keeps producing "100% organic cotton" for a hoodie whose fabric you never provided. Which change targets the actual failure?

Answer it in your own words, now, with nothing to choose from. The options are on the next slide — and picking the right one off a list is an easier skill than producing it.

Correct: Add: "Use only the facts I give you. If the fabric is missing, ask me instead of guessing."

Why: The failure is a missing rule plus a missing instead-instruction: nothing told Claude that inventing facts is out of bounds, or gave it a legal move (ask) when a fact is absent. A never-rule with an escape hatch turns "don't lie" into a mechanism. Role, output length, and re-rolling all leave the actual gap open.

Check your understanding

Your listing prompt keeps producing "100% organic cotton" for a hoodie whose fabric you never provided. Which change targets the actual failure?

• A. Add: "Use only the facts I give you. If the fabric is missing, ask me instead of guessing." (correct)
• B. Upgrade the role to "world-class award-winning copywriter"
• C. Expand the output format from 5 bullets to 8 for more detail
• D. Re-run the same prompt a few times until a run comes out accurate

Answer: A

Why: The failure is a missing rule plus a missing instead-instruction: nothing told Claude that inventing facts is out of bounds, or gave it a legal move (ask) when a fact is absent. A never-rule with an escape hatch turns "don't lie" into a mechanism. Role, output length, and re-rolling all leave the actual gap open.

Why B tempts people

A fancier role changes tone and polish, not what the model is allowed to invent - a world-class copywriter still fills gaps with plausible fabric claims if nothing forbids it.

Why C tempts people

Output format controls SHAPE, not truth. Eight bullets instead of five just gives the invention more room.

Why D tempts people

Re-running is the 10-tweak loop mapping exists to kill - and each run still has the same hole, so any accurate run is luck you can't repeat next product.

Predict first

The weekly sales summary - same format every Monday, fresh numbers pasted each time, run by hand in the Claude app. Once the prompt passes its success test, where should it live?

Answer it in your own words, now, with nothing to choose from. The options are on the next slide — and picking the right one off a list is an easier skill than producing it.

Correct: A Skill - saved instructions you run on new numbers each week

Why: This is DO-again, not OPEN-again: the process repeats weekly with fresh input, and the old summaries aren't the ongoing value - the repeatable process is. Saved instructions run on new data each time is the definition of a Skill.

Check your understanding

The weekly sales summary - same format every Monday, fresh numbers pasted each time, run by hand in the Claude app. Once the prompt passes its success test, where should it live?

• A. A fresh chat each Monday, retyping the instructions
• B. A Skill - saved instructions you run on new numbers each week (correct)
• C. An Artifact - one saved summary document you reopen and overwrite weekly
• D. The API - weekly is frequent enough to automate

Answer: B

Why: This is DO-again, not OPEN-again: the process repeats weekly with fresh input, and the old summaries aren't the ongoing value - the repeatable process is. Saved instructions run on new data each time is the definition of a Skill.

Why A tempts people

Session 1's trap exactly - retyped instructions drift week to week and waste time. Repeated work belongs in something saved.

Why C tempts people

An Artifact holds value in the document itself. Overwriting one document weekly means the value was never the document - it was the process, which is Skill territory.

Why D tempts people

Frequency alone doesn't demand the API - you're still happily in the app, pasting by hand. The API earns its place only when the job must fire WITHOUT you, from other software.

Eliminate the wrong options

The seasonal inventory checklist: built once, reopened twice a year before each buying season, ticked off and edited as you learn what to watch for. Where should it live?

3 of these 4 are wrong. Strike them one at a time, and say what rules each one out before you strike the next. The survivor is the answer.

• A. A Skill - anything reused belongs in a Skill
• B. A fresh chat each season - it's rare enough to rebuild
• C. An Artifact - a kept document you reopen and edit each season
• D. The API - inventory data means it needs automation

Survives elimination: C

Why: This is OPEN-again: the checklist document itself is the value - your accumulated edits and lessons live in it, and twice a year you return to that same object. Nothing repeats often enough to be a habit, and nothing runs without you. Kept, reopened, edited = Artifact.

Check your understanding

The seasonal inventory checklist: built once, reopened twice a year before each buying season, ticked off and edited as you learn what to watch for. Where should it live?

• A. A Skill - anything reused belongs in a Skill
• B. A fresh chat each season - it's rare enough to rebuild
• C. An Artifact - a kept document you reopen and edit each season (correct)
• D. The API - inventory data means it needs automation

Answer: C

Why: This is OPEN-again: the checklist document itself is the value - your accumulated edits and lessons live in it, and twice a year you return to that same object. Nothing repeats often enough to be a habit, and nothing runs without you. Kept, reopened, edited = Artifact.

Why A tempts people

"Reused" isn't the test - reused HOW is. A Skill re-RUNS a process on new input; here you re-OPEN one evolving document. Saving it as a Skill would regenerate a fresh checklist each season and throw away your edits.

Why B tempts people

Rebuilding each season loses exactly the thing that makes the checklist valuable: the edits and hard-won additions from past seasons.

Why D tempts people

Touching inventory data doesn't require automation - you open this by hand twice a year inside the app. Nothing here fires on its own.

Discussion prompt

You're leaving with a finished worksheet and a written prompt. This week, put the prompt on the bench:

That is stated as though it always holds. Do one of two things: produce a case where it fails, or say precisely what rules such a case out. "It just does" is not on the menu.

Hint: Hunt at the extremes first — zero, one, negative, empty, equal. If every extreme survives, the reason they survive is the proof.

Draw it

One page, no notation unless you need it: draw how these connect — Audit Your Ideas · Worksheet -> Prompt · Artifact or Skill?. Put an arrow wherever one of them is what makes another possible, and label the arrow with why.