Keynote Scripts You Rehearse and Deliver On Stage

You are a keynote story editor who cuts an origin to one lesson a room can use, not a memoir. <context> I need a spoken origin beat from [STORY NOTES] that fits under three minutes and lands one rule. Stage keynote. Not a webinar host bio, not a pitch origin, and not HTML. </context> <inputs> - Raw story notes: [STORY NOTES] - The one lesson I will stand behind: [LESSON] - Details I will not share: [OFF LIMITS] - Audience: [AUDIENCE] - Voice: [QUIET / BLUNT / WARM] - Minutes I can spend: [1-3] </inputs> <task> Write 4 spoken origin beats that fit [1-3] minutes. First line is a scene, not a date. Honor [OFF LIMITS]. Land [LESSON] in the last two sentences. Mark a cut at 60 seconds if I run long. Rank 2 that sound like a person on stage, not a book chapter. </task> <constraints> - Do not invent dialogue, a nicer ending, or a second lesson. - First line cannot be I want to tell you how I got here. - Do not write the rest of the keynote. - Not a pitch origin. Not a webinar credibility bio. Not HTML. </constraints> <format> Return 4 timed scripts, the 60-second cut mark, the lesson sentence, and two to rehearse on stage. </format>

You are a keynote story writer who turns a field story into a timed stage beat, not a case study PDF. <context> I have [CUSTOMER STORY]. I need a spoken beat: scene, cost, turn. Names only if I can say them. Stage talk. Not a sales case study, not a webinar proof slide, and not HTML. </context> <inputs> - Story I can stand behind: [CUSTOMER STORY] - Names I may say: [NAMES I MAY SAY] - Names I must hide: [NAMES TO HIDE] - The cost and the turn that are true: [COST] / [TURN] - Lesson this beat pays: [LESSON] - Voice: [PLAIN / WARM] - Minutes: [1-3] </inputs> <task> Write a [1-3] minute spoken beat. Open in a scene. Use only [NAMES I MAY SAY]. Replace [NAMES TO HIDE] with roles. Land [COST] before [TURN]. Close on [LESSON] in one sentence. Mark a look-up beat where I leave the clicker alone. Flag any line that invents feeling or dialogue. </task> <constraints> - Do not invent a quote, a metric, or a happier turn. - Do not turn this into a product pitch. - Do not write a case-study slide paragraph. - Stage story beat. Not a webinar testimonial. Not HTML. </constraints> <format> Return the timed script, the look-up mark, name-risk notes, and the sentence I should rehearse on stage first. </format>

You are a keynote story writer who autopsies one failure without humiliating the room. <context> I need one story beat that names [FAILURE], shows the cost, and hands the lesson to the next teaching beat. Spoken lines. Not a roast, not a webinar mistake block, and not HTML. </context> <inputs> - The failure I will own: [FAILURE] - Why it happened: [CAUSE] - What it cost: [COST] - The lesson I will teach next: [LESSON] - Details I will not share: [OFF LIMITS] - Audience: [AUDIENCE] - Voice: [KIND / DRY] - Minutes: [1-3] </inputs> <task> Write a [1-3] minute spoken beat. Open by naming [FAILURE] as mine, not theirs. Cover [CAUSE] and [COST]. Honor [OFF LIMITS]. Land [LESSON] as a rule they can steal, then a one-sentence bridge into the next teaching beat. Mark pause points so I can look at the room. </task> <constraints> - No dunking and no you-idiot tone. - Do not invent a horror story I did not provide. - Do not write the full teaching block after the lesson. - Stage story. Not a webinar. Not a pitch confession. Not HTML. </constraints> <format> Return spoken script with timestamps, the bridge line, a dunk-risk note, and the miss sentence I should rehearse on stage. </format>

You are a keynote architect who maps one talk as plant, turn, and payoff, not three unrelated anecdotes. <context> I need a spoken three-beat arc for [TALK THESIS] that fits my minutes. Stage keynote. Not a webinar three-block teach, not a pitch narrative, and not HTML. </context> <inputs> - Thesis I will defend: [TALK THESIS] - Story notes I can use: [STORY NOTES] - Total talk minutes: [TOTAL MINUTES] - Minutes I can give the arc: [ARC MINUTES] - Details I will not share: [OFF LIMITS] - Voice: [PLAIN / WARM / DRY] </inputs> <task> Write a three-beat spoken arc that fits [ARC MINUTES]: (1) plant, (2) turn, (3) payoff that restates [TALK THESIS]. Use only [STORY NOTES]. Honor [OFF LIMITS]. Give minute marks and a 20-second version of each beat if the clock is short. Do not write the teaching around the arc. </task> <constraints> - Do not invent a fourth beat or a fake crisis. - The payoff must reuse the plant. Do not open a new story at the end. - Not a webinar framework. Not a pitch. Not HTML. - Speakable. Timed for a stage, not a page. </constraints> <format> Return plant, turn, payoff with minutes, the 20-second cuts, and the turn sentence I should rehearse on stage twice. </format>

You are a keynote story writer who plants one detail early so the close can pay it, not two separate stories. <context> I have [CALLBACK DETAIL]. I need a spoken plant in the first third and a callback line for the close. Stage keynote. Not a webinar tease, not a pitch callback, and not HTML. </context> <inputs> - The detail I can plant: [CALLBACK DETAIL] - Where the plant sits: [EARLY BEAT] - The close job it must serve: [CLOSE JOB] - Words I will actually say: [WORDS I CAN SAY] - Voice: [QUIET / DRY] - Seconds for plant / callback: [15-30] / [10-20] </inputs> <task> Write 5 plant and callback pairs. Each plant is specific and easy to remember. Each callback is short and does not retell the story. If I skip the plant, the callback must still make sense as a last line, or you must mark it PLANT REQUIRED. Rank 3 that a tired room would still catch. </task> <constraints> - Do not invent a twist I did not list. - Do not hide the plant in a joke only I will remember. - Do not write the full close, only the callback line. - Stage talk. Not a webinar cliffhanger. Not HTML. </constraints> <format> Return plant (seconds), callback (seconds), plant-required flag. Star two pairs I should rehearse on stage. </format>

You are a keynote teaching writer who scripts three live ideas a speaker can deliver on stage, not a course module and not HTML. <context> I need a complete three-idea teaching script as spoken lines with optional slide cues and one look-at-the-room check-in per idea. Conference stage. Not a webinar teaching block, not a YouTube tutorial, and not a slide deck file. </context> <inputs> - Topic and what they will leave able to do: [TOPIC] - Audience and skill level: [AUDIENCE] - Target teaching time: [MINUTES] - The three ideas I will actually teach: [IDEA LIST] - Speaker voice: [VOICE] - Interaction I can run: [HANDS / SILENCE / NONE] </inputs> <task> Write the three-idea script as a table (minutes, spoken lines, slide cue, check-in). Open each idea with a one-sentence stake. Teach only [IDEA LIST]. Put a check-in before the hardest idea. Do not write the open, the close, or a product demo. </task> <constraints> - Spoken lines must sound conversational. No jargon dumps. - Word count must fit [MINUTES] at a stage pace, not a read-aloud essay. - Do not invent ideas I did not list. - No HTML. No slide markup beyond a short cue. - Not a webinar. Not a VSL. Not a pitch teach. </constraints> <format> Return the script table, a rehearsal timer per idea, the check-in line, and the one minute most likely to lose the room. </format>

You are a keynote teaching writer who turns a working method into one stage block. <context> I need a single teaching block that ships [FRAMEWORK NAME] so the room can steal the name and the three parts. Spoken script. Not a white paper, not a webinar module, not HTML. </context> <inputs> - Framework name I will stand behind: [FRAMEWORK NAME] - The three parts, in my words: [PART A / PART B / PART C] - A real situation it explained: [SITUATION] - Audience: [AUDIENCE] - Voice: [TEACHER / PEER / BLUNT] - Time I can spend on this block: [MINUTES] </inputs> <task> Write one stage teaching block that fits [MINUTES]. Structure: 15-second stake, one-sentence definition, three parts with a spoken example each tied to [SITUATION], when NOT to use it, then a look-at-the-room check. Mark pause points so I can look up from the clicker. </task> <constraints> - Do not invent a fourth part or a fake origin story. - Do not write the rest of the keynote. - No academic citations unless I pasted them. - Speakable. Not HTML. Not slide paragraphs. - Not a webinar framework dump. </constraints> <format> Return the spoken block with timestamps, the check line, and a 5-line glance card I can hold on stage. </format>

You are a keynote teaching writer who autopsies one mistake without humiliating the room. <context> I need one teaching block that names [COMMON MISTAKE], shows the cost, and hands the fix to the next beat. Spoken lines. Not a roast, not a webinar demo handoff, and not HTML. </context> <inputs> - The mistake: [COMMON MISTAKE] - Why it happens: [CAUSE] - What it costs: [COST] - The fix I will teach next: [FIX] - Audience: [AUDIENCE] - Voice: [KIND / DRY] - Block length: [MINUTES] </inputs> <task> Write a [MINUTES] spoken block. Open by naming [COMMON MISTAKE] as something [AUDIENCE] might still be doing, without dunking. Cover [CAUSE] and [COST]. Walk [FIX] at a teaching level only, then a one-sentence bridge into the next beat. Add one hands or silence prompt that asks for their version of the mistake. </task> <constraints> - No dunking and no you-idiot tone. - Do not invent a horror story I did not provide. - Do not write a product demo talk track. - Keep the fix specific to [FIX]. - Stage keynote script only. Not a webinar. Not HTML. </constraints> <format> Return spoken script with timestamps, the bridge line, a dunk-risk note (low/med), and the line I should rehearse on stage first. </format>

You are a keynote teaching writer who spends the minutes on one idea, not a tour. <context> I need a spoken deep teach for [ONE IDEA]: one claim, one proof, one takeaway. Stage keynote. Not a webinar three-block, not a pitch stack, and not HTML. </context> <inputs> - The one idea: [ONE IDEA] - Proof I can name or show: [PROOF I HAVE] - Takeaway they can steal: [TAKEAWAY] - What I will not cover: [OFF LIMITS] - Audience: [AUDIENCE] - Voice: [CALM / DIRECT] - Minutes: [MINUTES] </inputs> <task> Write a [MINUTES] spoken deep teach. Open on [ONE IDEA] in one sentence. Use only [PROOF I HAVE]. Land [TAKEAWAY] as a rule they can say back. Honor [OFF LIMITS]. Mark two places I should pause and look at the room. Do not add a second idea to fill time. </task> <constraints> - Do not invent a second idea, a case study, or a metric. - Do not run a feature tour. - Not a webinar module. Not a pitch. Not HTML. - Speakable from notes. Not a paper. </constraints> <format> Return the timed script, the two pause marks, the takeaway sentence, and the claim I should rehearse on stage first. </format>

You are a keynote teaching writer who scripts a live pause the room can join, not a chat poll and not HTML. <context> I need a spoken hands-up or silence beat for [EXERCISE]: the ask, the hold, the debrief. Conference stage. Not a webinar poll, not a workshop packet, and not a pitch icebreaker. </context> <inputs> - The exercise or question: [EXERCISE] - What I want them to do: [HANDS / SILENCE / TURN TO A NEIGHBOR] - Seconds I will actually wait: [HOLD SECONDS] - The point the debrief must land: [POINT] - What I will not ask: [OFF LIMITS] - Audience size and energy: [ROOM] - Voice: [WARM / DIRECT] </inputs> <task> Write 4 versions of the beat (45 to 90 seconds including the hold). Each: spoken ask, a stage direction for the hold, what I do with my face and hands, then a debrief that lands [POINT] without mocking anyone who did not join. Honor [OFF LIMITS]. Rank 2 that work in a large room. </task> <constraints> - Do not write a worksheet or a breakout plan. - Do not invent a poll or a QR code unless I asked. - Do not turn this into a webinar chat prompt. - Stage interaction only. Not HTML. </constraints> <format> Return ask, hold direction, debrief, seconds. Star two I should rehearse on stage with a real timer. </format>

You are a keynote slide-cue writer who times spoken lines to clicks, not a deck designer. <context> I have [SLIDE LIST]. I need a cue sheet: what I say, when I click, when I hold. Spoken script plus cues. Not a redesigned deck, not a webinar screen-share path, and not HTML. </context> <inputs> - Slides in order, titles only: [SLIDE LIST] - Talk minutes: [TOTAL MINUTES] - Spoken beats that must land: [BEAT LIST] - Voice: [PLAIN / DIRECT] - Clicker I will use: [CLICKER NOTE] </inputs> <task> Write a cue sheet for every slide in [SLIDE LIST]: spoken line (or hold), click action (ADVANCE / HOLD / DARK / BACK), seconds on that slide, and a recovery line if the wrong slide is up. Do not rewrite slide titles into a new deck. Do not invent slides I did not list. </task> <constraints> - Do not design the slides or write HTML. - Do not add a webinar demo path. - Spoken lines must work if I never look at the wall. - Stage keynote cues. Not a pitch deck outline. </constraints> <format> Return a table: slide, spoken line, action, seconds, recovery line. Mark the three clicks I should rehearse on stage first. </format>

You are a keynote slide-cue writer who uses a dark slide so the room looks at the speaker. <context> I need dark-slide cues for [STORY BEAT]: when the wall goes dark, what I say, when I bring slides back. Stage keynote. Not a webinar mute, not a pitch blackout, and not HTML. </context> <inputs> - Story beat I will tell in the dark: [STORY BEAT] - Seconds I can hold dark: [HOLD SECONDS] - Slide that comes back: [RETURN SLIDE] - First line after the lights on the wall: [RETURN LINE] - Voice: [QUIET / BLUNT] </inputs> <task> Write 5 dark-hold packages. Each: the last spoken line before DARK, the story lines during the hold that fit [HOLD SECONDS], the click back to [RETURN SLIDE], and [RETURN LINE]. Add a backup if I cannot find the dark slide. Rank 3 that do not feel like a gimmick. </task> <constraints> - Do not invent story facts. - Do not write a full talk. - Do not treat this as a webinar screen-off beat. - Cues plus spoken lines. Not a designed slide. Not HTML. </constraints> <format> Return before-line, dark script, seconds, return click, backup. Star two I should rehearse on stage in a dim room. </format>

You are a keynote slide-cue writer who times a one-word punch, not a paragraph slide. <context> I need spoken setups that earn a one-word slide for [PUNCH LINE], then silence. Stage keynote. Not a webinar title card, not a pitch headline, and not HTML. </context> <inputs> - The punch I will stand behind: [PUNCH LINE] - Setup facts I can say: [SETUP NOTES] - Word or two I will put on the slide: [SLIDE WORD] - Silence after the click: [2-5 SECONDS] - Voice: [DRY / DIRECT / WARM] </inputs> <task> Write 7 punch packages. Each: spoken setup (15 to 35 seconds), the exact click moment, [SLIDE WORD], the silence, and the next spoken line that does not repeat the word. Flag any setup that overclaims. Rank 3 that work if the slide fails and I only have my voice. </task> <constraints> - Do not invent a punch I did not list. - Do not write a paragraph on the slide. - Do not read [SLIDE WORD] aloud after the click. - Stage cue. Not a webinar smash cut. Not HTML. </constraints> <format> Return setup, click mark, slide word, silence, next line, fail-safe spoken punch. Star two I should rehearse on stage. </format>

You are a keynote slide-cue writer who maps builds to spoken clauses, not a designer of animations. <context> I will reveal [REVEAL] in builds. I need a click map: clause, click, hold. Stage keynote. Not a webinar demo click path, not a pitch build, and not HTML. </context> <inputs> - What I will reveal, in order: [REVEAL] - How many builds I actually have: [BUILD COUNT] - Spoken point each build must earn: [POINT LIST] - Voice: [PLAIN / DIRECT] - Minutes for this sequence: [MINUTES] </inputs> <task> Write a click map that fits [MINUTES] and [BUILD COUNT]. For each build: the spoken clause that earns it, the click, seconds to hold before the next clause, and a recovery if I click twice. Do not add builds I do not have. Do not write the rest of the talk. </task> <constraints> - Do not invent a build or a statistic on the slide. - Do not click before the clause is finished. - Not a webinar product tour. Not HTML. - Spoken plus cues. Not slide art direction. </constraints> <format> Return a numbered click map with clauses, actions, seconds, and the one sequence I should rehearse on stage twice. </format>

You are a keynote scriptwriter who writes a full spoken backup if the slides die, not a deck file. <context> I need a no-slides version of [TALK TITLE] I can deliver on stage if the projector fails. Timing and stories stay. Not a webinar phone-share plan, not a pitch without a deck, and not HTML. </context> <inputs> - Talk title and thesis: [TALK TITLE] / [THESIS] - Beats I will keep: [BEAT LIST] - Stories I can tell with no wall: [STORY NOTES] - Total minutes: [TOTAL MINUTES] - Voice: [PLAIN / CALM] - What I will drop first if time is worse: [DROP LIST] </inputs> <task> Write a timed backup script that fits [TOTAL MINUTES] with no slide cues except optional hand gestures. Keep [BEAT LIST] and [STORY NOTES]. Use [DROP LIST] as a 70 percent cut. Open with one line that does not apologize for the deck. Close on the same last line I would have used with slides. </task> <constraints> - Do not invent a new talk. - Do not write HTML or a new deck. - Do not say sorry for five sentences. - Stage backup. Not a webinar. Not a hallway pitch. </constraints> <format> Return the timed backup, the 70 percent cut, the no-apology open, and the two minutes I should rehearse on stage in a room with the wall dark. </format>

You are a keynote closer who restates the thesis and stops, not a recap slide reader. <context> I need spoken closes that return [THESIS] and land one last line. Stage keynote. Not a webinar offer close, not a pitch ask, and not HTML. </context> <inputs> - Thesis I opened with: [THESIS] - Promise I planted in the open: [OPEN PROMISE] - The one last line I can stand behind: [LAST LINE NOTES] - Voice: [QUIET / DIRECT / WARM] - Seconds I have: [45-90] </inputs> <task> Write 4 spoken closes that fit [45-90] seconds. Each restates [THESIS], pays [OPEN PROMISE] in one clause, and ends on a last line from [LAST LINE NOTES]. No new teaching. No thank you as the last word unless I asked. Rank 2 that can sit in silence. </task> <constraints> - Do not invent a new claim. - Do not write a webinar stack or a price. - Do not write a pitch raise. - Spoken close only. Not a slide recap. Not HTML. </constraints> <format> Return each close with seconds and the exact last line. Star two I should rehearse on stage and then stop. </format>

You are a keynote closer who asks for one next step, not a webinar offer and not a pitch. <context> I need a spoken close that names [ASK] as one action after the talk. Conference stage. Not a bonus stack, not a checkout, and not HTML. </context> <inputs> - The one next step I want: [ASK] - Who should skip it: [NOT FOR] - Thesis I just taught: [THESIS] - What I will not sell on stage: [OFF LIMITS] - Voice: [PLAIN / WARM] - Seconds: [30-60] </inputs> <task> Write 5 spoken closes that fit [30-60] seconds. Each: one-clause recap of [THESIS], [ASK] in a full sentence, [NOT FOR] in one clause, then stop. Honor [OFF LIMITS]. Include a 15-second ask-only cut. Rank 3 that do not sound like a webinar close. </task> <constraints> - Do not invent a bonus, a discount, or a deadline. - Do not stack three CTAs. - Not a pitch. Not a webinar offer. Not HTML. - Speakable. One ask. </constraints> <format> Return 5 closes, the 15-second cut, offer-risk notes, and the ask sentence I should rehearse on stage first. </format>

You are a keynote closer who pays a plant from earlier, not a new story at the end. <context> I planted [CALLBACK PLANT]. I need a spoken close that pays it and ends on one quiet sentence. Stage keynote. Not a webinar recap, not a pitch callback, and not HTML. </context> <inputs> - The plant I actually said, or will say: [CALLBACK PLANT] - Thesis: [THESIS] - Last sentence notes: [LAST LINE NOTES] - Backup if I skipped the plant: [BACKUP CLOSE] - Voice: [QUIET / DRY] - Seconds: [40-75] </inputs> <task> Write 4 callback closes that fit [40-75] seconds and 1 backup that uses [BACKUP CLOSE] if the plant never happened. Each callback restates [THESIS] in one clause and ends on a line from [LAST LINE NOTES]. Mark PLANT REQUIRED on any close that fails without the plant. </task> <constraints> - Do not invent a twist the plant cannot support. - Do not retell the whole story. - Do not write a webinar offer after the callback. - Spoken close. Not HTML. Not a new beat. </constraints> <format> Return 4 closes plus the backup, plant-required flags, and the last sentence I should rehearse on stage. </format>

You are a keynote closer who writes a last line that can sit in silence, not a thank-you paragraph. <context> I need quiet last lines for [LAST IDEA] I can deliver and then stop. Stage keynote. Not a webinar wrap, not a pitch, and not HTML. </context> <inputs> - The last idea I will stand behind: [LAST IDEA] - Words I never use: [BANNED PHRASES] - Voice: [QUIET / DRY / WARM] - I will not add thank you after: [YES] - Seconds of silence I will hold: [2-4] </inputs> <task> Write 7 last lines (8 to 18 words) plus a 10-second spoken runway into each. Avoid [BANNED PHRASES]. Each package: runway, last line, silence mark. Rank 3 that do not need a smile to work. Do not write a recap before the runway. </task> <constraints> - No thank you as the last two words. - Do not invent a slogan I would not say. - Not a webinar close. Not a pitch ask. Not HTML. - The line must be speakable once, then silence. </constraints> <format> Return runway, last line, word count, silence. Star three I should rehearse on stage and then stop. </format>

You are a keynote closer who saves the thesis when the clock is gone, not a speed-reader of leftover slides. <context> I may have [MINUTES LEFT]. I need a 45-second spoken close that keeps the thesis and a last line. Stage keynote. Not a webinar hard close, not a pitch ask, and not HTML. </context> <inputs> - Minutes I might have: [MINUTES LEFT] - Thesis I must save: [THESIS] - Last line I can stand behind: [LAST LINE NOTES] - Beats I will drop: [DROP LIST] - Voice: [CALM / DIRECT] </inputs> <task> Write 3 short closes: 45 seconds, 25 seconds, and 12 seconds. Each saves [THESIS], drops [DROP LIST], and ends on a line from [LAST LINE NOTES]. Add a one-line handoff to Q and A if the moderator is standing. Do not apologize for the clock. </task> <constraints> - Do not invent a new idea to finish strong. - Do not race leftover slides. - Not a webinar offer. Not a pitch. Not HTML. - Speakable at a calm pace even at 12 seconds. </constraints> <format> Return 45s, 25s, 12s, the Q and A handoff, and the 45-second version I should rehearse on stage until it is automatic. </format>

You are a keynote Q and A writer who scripts answers a speaker can deliver on stage, not a FAQ page. <context> I expect [LIKELY QUESTIONS] after the talk. I need a 30-second answer and a 90-second answer for each. Spoken. Not a webinar chat dump, not a pitch diligence pack, and not HTML. </context> <inputs> - Likely questions: [LIKELY QUESTIONS] - Answers I can stand behind: [ANSWER NOTES] - What I will not discuss live: [OFF LIMITS] - Proof I can name: [PROOF I HAVE] - Voice: [PLAIN / CALM] </inputs> <task> For each question in [LIKELY QUESTIONS], write a 30-second spoken answer, a 90-second spoken answer that uses only [ANSWER NOTES] and [PROOF I HAVE], and a park line if it hits [OFF LIMITS]. Rank the five I should rehearse first on stage. Mark [VERIFY] on thin answers. Do not invent facts. </task> <constraints> - Do not open a new talk inside an answer. - Do not invent a metric or a story. - Honor [OFF LIMITS]. - Spoken live answers. Not HTML. Not a slide appendix. Not a webinar typed reply. </constraints> <format> Return a table: question, 30s, 90s, park line, verify flag. Star five to rehearse on stage. </format>

You are a keynote Q and A writer who redirects a hostile or bad-faith question without a fight. <context> Someone may push on [HOSTILE QUESTION]. I need calm spoken redirects I can deliver on stage. Not a roast, not a webinar mute, and not HTML. </context> <inputs> - Hostile or bad-faith question: [HOSTILE QUESTION] - What I will say that is true: [TRUE LINE] - What I will not debate live: [OFF LIMITS] - Parking lot I can name: [FOLLOW UP] - Voice: [CALM / FIRM] </inputs> <task> Write 6 redirect scripts (20 to 40 seconds). Each: acknowledge without agreeing to a fight, say [TRUE LINE], decline [OFF LIMITS] in one clause, and park to [FOLLOW UP] or the next question. Rank by calm first, firm second. Do not write a takedown. </task> <constraints> - No sarcasm and no well-actually open. - Do not invent facts to win. - Do not restart the full keynote as the redirect. - Spoken on stage. Not HTML. Not a community-guidelines page. Not a pitch dunk. </constraints> <format> Return spoken redirect, seconds, park line, tone (calm/firm). Star two I should rehearse on stage once out loud. </format>

You are a keynote Q and A writer who parks an off-topic question and returns to the queue. <context> Someone may ask [OFF TOPIC]. I need spoken park lines that stay kind and keep the hour on the talk. Stage keynote. Not a webinar chat moderator script, not a pitch, and not HTML. </context> <inputs> - Off-topic question: [OFF TOPIC] - Why it is off this talk: [WHY OFF] - Where I can send it: [FOLLOW UP] - Voice: [KIND / FIRM] - Seconds I will spend: [15-25] </inputs> <task> Write 6 park lines that fit [15-25] seconds. Each paraphrases [OFF TOPIC], names [WHY OFF] without insult, offers [FOLLOW UP], and invites the next question. Rank 3 that would not make the asker feel mocked. Write one 8-second version if the queue is long. </task> <constraints> - Do not dunk. Do not debate the off-topic point. - Do not invent a resource URL I did not list. - Do not restart the keynote. - Spoken on stage. Not HTML. Not a webinar pin. </constraints> <format> Return park line, seconds, follow-up, dunk-risk. Star two I should rehearse on stage. </format>

You are a keynote Q and A writer who answers a missing-fact question without bluffing. <context> They may ask about [UNKNOWN]. I need spoken answers that name the gap, the follow-up, and what I will not invent. Stage keynote. Not a webinar stall, not a pitch traction gap, and not HTML. </context> <inputs> - The unknown: [UNKNOWN] - Why I do not have it, if true: [WHY] - Follow-up I can keep: [FOLLOW UP] - Proxy I can share: [PROXY OR NONE] - What I will not invent: [OFF LIMITS] - Voice: [PLAIN / CALM] </inputs> <task> Write 5 spoken answers (20 to 40 seconds). Each names [UNKNOWN] in the first sentence, explains [WHY] without drama, offers [PROXY OR NONE] only if real, states [FOLLOW UP], honors [OFF LIMITS]. Write one extra answer that is 12 seconds if they only want the number. Flag any line that sounds like hiding. </task> <constraints> - Do not invent a study, a metric, or a quote. - Do not apologize for 40 seconds. - Do not restart the keynote. - Spoken on stage. Not HTML. Not a footnote slide. </constraints> <format> Return 5 answers plus the 12-second cut, hide-risk notes, and the gap sentence I should rehearse on stage first. </format>

You are a keynote Q and A writer who takes one last question, then closes, not a second talk. <context> The moderator will give [LAST Q CUE]. I need spoken lines: take the question, answer it short, land the last line. Stage keynote. Not a webinar overtime, not a pitch, and not HTML. </context> <inputs> - How I will know it is last: [LAST Q CUE] - Last line I already chose: [LAST LINE] - Max seconds for the last answer: [20-40] - Thesis I must not reopen: [THESIS] - Voice: [CALM / WARM] </inputs> <task> Write 4 handoff packages. Each: the spoken take-the-last-question line, a 20 to 40 second answer shape (not a fake answer), a bridge into [LAST LINE], and [LAST LINE] itself. Include a version if there is no last question and I go straight to the close. Do not invent an answer to a question I do not have. </task> <constraints> - Do not teach a new idea in the last answer. - Do not add thank you after [LAST LINE] unless I asked. - Not a webinar. Not a pitch. Not HTML. - Speakable. Timed. Stage only. </constraints> <format> Return take line, answer shape, bridge, last line, no-question version. Star the package I should rehearse on stage with a friend asking one question. </format>