30 Claude Prompts for Demo Scripts

You are a sales engineer who rebuilds a standard demo for the one persona in the room. <context> I have a standard 12-click demo that we run for everyone. The next call is with one persona who will tune out if I show screens that do not affect their job. </context> <inputs> - Standard demo outline: [STANDARD DEMO OUTLINE] - Persona title: [PERSONA TITLE] - What they are measured on: [PERSONA KPIS] - Screens I can show: [AVAILABLE SCREENS] - Call length: [CALL LENGTH IN MINUTES] </inputs> <task> Rebuild the demo as a persona-specific flow. Keep only screens that affect their KPIs. Add a one-line reason for each keep, and list the cuts with a one-line reason. </task> <constraints> Do not add screens that are not in the available list. Total spoken time must fit the call length with five minutes reserved for questions. No section longer than four minutes. </constraints> <format> A keep list in run order, a cut list, then a minute-by-minute flow I can follow live. </format>

You are an enterprise account executive who runs mixed-room demos without losing either side. <context> My next demo has more than one role on the call. If I run one path I will bore the executive or skip the detail the operator needs. I want a path that switches audience without restarting. </context> <inputs> - Attendee titles: [ATTENDEE TITLES] - Product: [PRODUCT NAME] - Each role's top question: [ROLE QUESTIONS] - Time I have: [TOTAL MINUTES] - Must-show capability: [MUST SHOW CAPABILITY] </inputs> <task> Build a single demo path that addresses each role in turn. Mark who each section is for, what I say to hand the room to the next role, and where I park detail for a later working session. </task> <constraints> The must-show capability appears once, not once per role. Executive sections stay under three minutes. Do not write a separate demo for each person. Keep handoff lines under 15 words. </constraints> <format> A run-of-show table: Minute, For whom, What I show, What I say, Handoff line. </format>

You are a solutions consultant who localizes demo scripts to a vertical and a single use case. <context> Our master demo script is horizontal. I am walking into a vertical account with one use case that matters, and I need the script rewritten around that job, not around our product map. </context> <inputs> - Master demo script: [PASTE MASTER SCRIPT] - Vertical: [VERTICAL] - Use case: [USE CASE] - Sample data I can load: [SAMPLE DATA AVAILABLE] - Words this vertical uses: [VERTICAL VOCABULARY] </inputs> <task> Rewrite the master script for this vertical and use case. Replace generic labels with the vertical vocabulary. Call out which sample data to load on each screen. </task> <constraints> Do not keep a section that does not serve the use case. Do not invent sample data fields we cannot load. Spoken examples must use the vertical vocabulary, not our internal names. </constraints> <format> The rewritten script in run order. Each step: Screen, Data to load, Spoken example, Vertical word used. </format>

You are a sales director who coaches reps to run a 15-minute executive demo without looking rushed. <context> I have a 45-minute technical demo and a 15-minute slot with an executive. I need a short version that still proves value and ends with a clear ask, not a teaser that says come to the long one. </context> <inputs> - Long demo outline: [LONG DEMO OUTLINE] - Executive title: [EXECUTIVE TITLE] - Business outcome they named: [BUSINESS OUTCOME] - Risk they mentioned: [NAMED RISK] - Ask I want at the end: [DESIRED ASK] </inputs> <task> Write a 15-minute executive demo: two minutes of framing, eight minutes of proof on the outcome, three minutes on the risk, two minutes for the ask and next step. </task> <constraints> No more than four screens. No feature names unless the executive used them. Do not say we can cover that later as a way to skip proof. Every minute must be assigned. </constraints> <format> A 15-row minute plan with spoken line, screen, and the point of that minute. </format>

You are a pre-sales lead who reorders demo screens to match the prospect's stated priorities, not the product's information architecture. <context> The prospect sent a priority list before the demo. My default click path starts in the admin console, which is last on their list. I want a new sequence that starts where they start. </context> <inputs> - Their priority list: [STATED PRIORITIES] - Screens I can open: [SCREEN LIST] - Default click path: [DEFAULT CLICK PATH] - Time per priority: [MINUTES PER PRIORITY] - Known blocker: [KNOWN BLOCKER] </inputs> <task> Reorder the screens to match their priority list. For each priority, pick the one screen that proves it, write the opening line, and note if the known blocker will come up there. </task> <constraints> Do not start with a screen that is not on their list. If a priority has no matching screen, write a spoken line that admits it and offers a follow-up artifact. Stay inside the minutes per priority. </constraints> <format> A sequence list: Priority, Screen, Opening line, Blocker note, Minutes. </format>

You are a sales engineer who writes short, honest answers for questions that stall live demos. <context> I lose time on live demos when a question I should have expected sends me into a ramble. I want a prep sheet of likely questions with a 20-second answer and a place to park the rest. </context> <inputs> - Product: [PRODUCT NAME] - Audience: [ATTENDEE TITLES] - Questions I already got last time: [PAST QUESTIONS] - Limits of what we do: [KNOWN LIMITS] - Competitive product they use: [INCUMBENT PRODUCT] </inputs> <task> Write a prep sheet of 10 likely questions. For each: a 20-second spoken answer, whether to answer live or park it, and one sentence on what not to say. </task> <constraints> At least three questions must come from the known limits, not from happy-path features. Do not invent capabilities. If the incumbent is relevant, say how we differ in one clause, not a full bake-off. </constraints> <format> A table: Question, 20-second answer, Live or park, Do not say. </format>

You are a demo facilitator who keeps a room on the story without dismissing people who ask sideways questions. <context> Prospects ask useful questions that belong later in the agenda, and I either chase them and blow the clock or I shut them down and they go quiet. I need a park-and-return script I can say out loud. </context> <inputs> - Demo agenda sections: [AGENDA SECTIONS] - Question they asked last time: [OFF TOPIC QUESTION] - Owner who can answer later: [FOLLOW UP OWNER] - Remaining minutes: [MINUTES LEFT] </inputs> <task> Write six park-and-return lines I can use live: acknowledge the question, say where it fits, write it down in front of them, and return to the current beat. Include one version for a technical question and one for a commercial question. </task> <constraints> Each line under 25 words. Never say great question as filler. Name the later section or the follow-up owner so the park is specific. Do not promise an answer you cannot give. </constraints> <format> Six numbered lines labeled Technical or Commercial, each with the exact words and the agenda section I point to. </format>

You are a principal solutions architect who can drop into a technical deep-dive and climb back to the business story in under two minutes. <context> An IT buyer will ask to see architecture mid-demo. If I stay there I lose the economic buyer. If I refuse I lose the IT buyer. I need a contained deep-dive and a return line. </context> <inputs> - Architecture I can show: [ARCHITECTURE I CAN SHOW] - IT buyer title: [IT BUYER TITLE] - Economic buyer title: [ECONOMIC BUYER TITLE] - Business story I was in: [CURRENT STORY BEAT] - Hard limit I must not overclaim: [HARD LIMIT] </inputs> <task> Write a two-minute technical sidebar: what I show, the three facts I state, the hard limit I admit, and the exact sentence that returns the room to the business story. </task> <constraints> Sidebar stays under 180 words spoken. Admit the hard limit in plain language. Do not use a whiteboard dump. The return sentence must name both buyers so neither feels dropped. </constraints> <format> Sidebar script, then a labeled Return sentence, then a one-line offer for a later architecture working session. </format>

You are a seasoned account executive who recovers when a prospect challenges a claim you just made on a live demo. <context> A prospect pushed back on a claim I made while a screen was up. I froze, restated the claim louder, and lost trust. I want a recovery script that treats the challenge as useful, not as an attack. </context> <inputs> - Claim I made: [CLAIM MADE] - What they said back: [PROSPECT CHALLENGE] - Evidence I actually have: [REAL EVIDENCE] - What I cannot prove today: [UNPROVEN PART] - Deal stage: [DEAL STAGE] </inputs> <task> Write a recovery: acknowledge the challenge, separate what I can prove now from what I cannot, offer a specific proof next, and get back to the demo beat without fake confidence. </task> <constraints> Do not restate the original claim as if they misheard. Do not invent evidence. The next proof must be something I can send within two business days. Keep the live recovery under 80 words. </constraints> <format> Four short spoken beats: Acknowledge, What is proven, What is not, Next proof and date. </format>

You are a sales coach who trains reps to say they do not know, without killing momentum or sounding unsure about the whole product. <context> When I do not know an answer on a demo I either guess or I apologize for too long. I need a short check-and-return line that protects trust and sets a real follow-up. </context> <inputs> - Topic I cannot answer live: [UNKNOWN TOPIC] - Who on my team can answer: [INTERNAL EXPERT] - When I can get back: [RESPONSE WINDOW] - What I still can show now: [NEXT DEMO BEAT] - Prospect name: [PROSPECT NAME] </inputs> <task> Write five check-and-return lines for different tones: factual, technical, commercial, security, and integration. Each line names the expert or the artifact I will send and the next beat I will run now. </task> <constraints> No apology longer than four words. Do not guess a half answer. Do not say I will circle back. Name a time window. Keep each line under 30 words. </constraints> <format> Five labeled lines I can read aloud, each followed by the next demo beat in five words or fewer. </format>

You are an account executive who sends a same-day demo recap the prospect can forward internally. <context> I just finished a product demo and I need a recap email tonight. If I wait until tomorrow the details blur and the champion has nothing to send to people who were not on the call. </context> <inputs> - What I showed: [DEMO SECTIONS SHOWN] - Reactions I heard: [LIVE REACTIONS] - Questions parked: [PARKED QUESTIONS] - Agreed next step: [AGREED NEXT STEP] - Champion name: [CHAMPION NAME] </inputs> <task> Write a same-day recap email the champion can forward. Restate what we showed in their words, list parked questions with owners, and confirm the next step with a date. </task> <constraints> Under 180 words. No marketing adjectives. Do not add sections I did not show. Subject line must name the company outcome, not the word demo. Use the champion's language from the reactions. </constraints> <format> Subject line, then the email body with three short labeled parts: What we covered, Open questions, Next step. </format>

You are a sales operations manager who turns messy demo feedback into a dated next-step plan. <context> I have raw notes from a demo: compliments, doubts, missing people, and vague interest. I need a plan I can put in the CRM and send to the champion without sounding like I invented momentum. </context> <inputs> - Raw feedback notes: [PASTE FEEDBACK NOTES] - People missing from the call: [MISSING ATTENDEES] - Deal stage: [DEAL STAGE] - Proof they asked for: [REQUESTED PROOF] - My proposed date: [PROPOSED DATE] </inputs> <task> Turn the notes into a next-step plan: what is confirmed, what is still open, who must join next, which proof to send, and a dated ask. Flag anything in the notes that is only polite interest. </task> <constraints> Do not treat compliments as buying signals. If the notes do not support a date, say DATE NOT CONFIRMED instead of using my proposed date as fact. Keep the plan to one page. </constraints> <format> Five labeled lines: Confirmed, Open, Who is missing, Proof to send, Dated ask. Then a CRM note of 50 words. </format>

You are a sales engineer who writes leave-behind one-pagers that match the live demo, not the full product brochure. <context> After demos I send a generic one-pager and the prospect says it does not match what they saw. I want a one-pager that only covers the screens and outcomes from this call. </context> <inputs> - Screens I showed: [SCREENS SHOWN] - Outcome we claimed: [CLAIMED OUTCOME] - Audience: [ATTENDEE TITLES] - Competitor they mentioned: [NAMED COMPETITOR] - Logo and product name: [PRODUCT NAME] </inputs> <task> Write a one-pager: headline outcome, three proof points tied to the screens shown, a short note on how we differ from the named competitor, and the agreed next step. </task> <constraints> Do not mention screens I did not show. Do not write a full competitive matrix. One page means under 220 words. Tone is factual, not brochure copy. </constraints> <format> Headline, three proof bullets with the matching screen name, one competitor sentence, next-step line. Ready to paste into a doc. </format>

You are a sales development lead who writes short follow-up sequences that do not nag after a product demo. <context> We send one recap and then go quiet, or we send five check-in emails that all say just bumping this. I want a 7-day sequence where each note adds a new artifact or answers a parked question. </context> <inputs> - Demo date: [DEMO DATE] - Parked questions: [PARKED QUESTIONS] - Artifacts I can send: [AVAILABLE ARTIFACTS] - Champion: [CHAMPION NAME] - Next meeting we want: [TARGET NEXT MEETING] </inputs> <task> Write a 7-day sequence of four touches: day 0 recap, day 2 artifact, day 4 parked-question answer, day 7 meeting ask. Each touch must add something new. </task> <constraints> No email that is only checking in. Each body under 90 words. Do not invent artifacts that are not on the list. If parked questions outnumber artifacts, combine answers on day 4. </constraints> <format> Four emails with day number, subject, body, and the attachment or link I send. </format>

You are a solutions consultant who turns every question from a demo into a custom proof pack the champion can share. <context> The prospect asked a set of questions I answered verbally. Those answers will fade. I want a proof pack that restates each question, the answer, and the evidence, so the champion does not have to replay the call. </context> <inputs> - Questions asked: [QUESTIONS ASKED] - Verbal answers I gave: [VERBAL ANSWERS] - Evidence I can attach: [EVIDENCE LIST] - What I still need to confirm: [ITEMS TO CONFIRM] - Champion title: [CHAMPION TITLE] </inputs> <task> Build a proof pack: one card per question with the question in their words, a tighter written answer, the evidence file to attach, and a status of CONFIRMED or PENDING. </task> <constraints> Do not upgrade a verbal hedge into a hard claim. PENDING items need an owner and a date. Maximum eight cards. Language a champion can forward to legal or IT without editing. </constraints> <format> A pack of cards: Question, Written answer, Evidence, Status, Owner if pending. </format>

You are a sales trainer who writes demo openers that set the agenda before anyone shares a screen. <context> I start demos by thanking people and asking if they can see my screen. By the time I get to the point, I have spent three minutes and lost the executive. I want a 90-second opener that states the outcome, the agenda, and how we will handle questions. </context> <inputs> - Outcome we agreed in discovery: [AGREED OUTCOME] - Agenda sections: [AGENDA SECTIONS] - Attendees: [ATTENDEE NAMES AND TITLES] - Question rule I want: [QUESTION RULE] - Timebox: [TOTAL MINUTES] </inputs> <task> Write a 90-second spoken opener: greet by role not by small talk, restate the agreed outcome, list the agenda in order, state the question rule, and confirm the timebox. </task> <constraints> Exactly 90 seconds at a normal speaking pace, about 200 words or fewer. Do not introduce the company history. Do not ask Can you see my screen as the first line. Name at least one attendee by role. </constraints> <format> The spoken opener as one block, then a 5-line agenda I can paste into chat as I start. </format>

You are a closer who ends product demos with a dated next meeting, not a soft we will be in touch. <context> My demos trail off when time runs out. People say thanks and leave with no calendar hold. I want a close that restates value, names the open items, and books the next meeting while they are still on the call. </context> <inputs> - Value they reacted to: [VALUE THEY REACTED TO] - Open items: [OPEN ITEMS] - Next meeting I want: [NEXT MEETING TYPE] - Proposed times: [PROPOSED TIMES] - Decision maker who was missing: [MISSING DECISION MAKER] </inputs> <task> Write a 2-minute close: one sentence of value in their words, the open items with owners, the next meeting type, two times to pick from, and a line that invites the missing decision maker. </task> <constraints> Do not say just wanted to check availability. Offer two real times, not a vague this week. Keep the close under 160 words. Do not introduce a new feature in the close. </constraints> <format> Spoken close, then the exact calendar invite title and three-bullet agenda for the next meeting. </format>

You are a demo producer who writes the one-line transitions that keep a click path feeling like one story. <context> Between sections I go quiet, hunt for the next screen, and then start the next feature as if it were a new meeting. I need spoken transitions that connect the last proof to the next one. </context> <inputs> - Demo sections in order: [SECTION LIST] - Proof each section is meant to make: [PROOF PER SECTION] - Persona in the room: [PERSONA TITLE] - Phrase they used for the pain: [THEIR PAIN PHRASE] </inputs> <task> Write one transition line for each jump between sections. Each line must use their pain phrase or the last proof, then point at what the next screen will settle. </task> <constraints> Each transition under 18 words. No And now I will show you. No section titles read aloud. Every line must mention the persona's job or the pain phrase. </constraints> <format> A list: From section, To section, Spoken transition, Screen I should already have open. </format>

You are a sales coach who inserts a mid-demo checkpoint so reps stop talking if the room is not with them. <context> I run the full demo even when faces go blank. I want a checkpoint at the halfway mark that confirms the value they care about, or redirects me if I have been showing the wrong thing. </context> <inputs> - Halfway section: [HALFWAY SECTION] - Value I think they care about: [ASSUMED VALUE] - Alternate path if I am wrong: [ALTERNATE PATH] - Minutes left after checkpoint: [MINUTES LEFT] - Champion name: [CHAMPION NAME] </inputs> <task> Write a 45-second checkpoint: one recap sentence, two yes or no questions, a line if they confirm, and a redirect script if they do not. Fit the remaining minutes to the path they choose. </task> <constraints> Questions must be answerable in one word. Do not ask How are we doing so far. If they do not confirm, cut to the alternate path and drop at least one planned section. Name the champion when you ask. </constraints> <format> Checkpoint script, Confirm branch with remaining minutes, Redirect branch with remaining minutes. </format>

You are a sales coordinator who sends a demo agenda the day before so attendees arrive knowing what will be shown and what they should bring. <context> Prospects join our demos cold. They have not pulled sample data, the right people are missing, and I spend the first ten minutes resetting context. I want an agenda email that goes out 24 hours ahead. </context> <inputs> - Demo goal: [DEMO GOAL] - Sections and minutes: [SECTIONS AND MINUTES] - What they should bring: [WHAT TO BRING] - Who must attend: [REQUIRED ATTENDEES] - Dial-in or link: [MEETING LINK] </inputs> <task> Write a short agenda email: goal in one sentence, timed sections, what to bring, who must attend, and a reply ask if a required person cannot make it. </task> <constraints> Under 140 words. Tone is practical, not salesy. Do not attach a slide deck. The reply ask should be a yes or no, not an open prompt. Include the meeting link once at the bottom. </constraints> <format> Subject line, email body with timed agenda, What to bring, Who should be there, Reply ask, link. </format>

You are a competitive specialist who writes displacement demos that stay factual and do not trash the incumbent. <context> The prospect already uses a named rival. A feature tour will look like ours versus theirs. I need a demo that starts from the job they still cannot finish in the incumbent and proves the gap without a smear. </context> <inputs> - Incumbent product: [INCUMBENT PRODUCT] - Job they still cannot finish: [UNFINISHED JOB] - Our product: [PRODUCT NAME] - Proof I can show live: [LIVE PROOF] - Claim I must not make: [OFF LIMITS CLAIM] </inputs> <task> Write a displacement demo: open on the unfinished job, show the incumbent limit in their words, show our live proof, and close on the switching cost they will actually face. </task> <constraints> Do not mock the incumbent. Do not make the off-limits claim. Switching cost must be honest, including work they will have to do. Keep spoken time under 12 minutes. Three screens maximum. </constraints> <format> Open, Incumbent limit, Live proof, Switching cost, Close. Each part: spoken lines and the screen. </format>

You are a solutions architect who runs architecture walkthroughs for IT buyers without turning the call into a full design review. <context> An IT buyer asked for an architecture walkthrough after the business demo. I need a script that covers data flow, identity, and failure modes in 20 minutes, with a parking lot for design-review topics. </context> <inputs> - Architecture I can share: [ARCHITECTURE I CAN SHARE] - Identity model: [IDENTITY MODEL] - Data stores involved: [DATA STORES] - Failure mode they asked about: [FAILURE MODE] - Topics I will not decide live: [PARKED DESIGN TOPICS] </inputs> <task> Write a 20-minute architecture walkthrough: data in, identity, data at rest, the named failure mode, and a parking lot. Include the one diagram I show and the questions I ask them. </task> <constraints> One diagram only. Do not design their future state live. Admit what the architecture I can share does not cover. Keep each section to four minutes. Questions to them must be specific, not any questions. </constraints> <format> Five timed sections with spoken points, the diagram callout, questions I ask, and the parking-lot list. </format>

You are an onboarding specialist who writes sandbox walkthroughs so a prospect can click without getting lost. <context> We gave the prospect a sandbox and they bounced. I need a walkthrough script they can follow alone, plus a 15-minute live version I can run if they want a guide. </context> <inputs> - Sandbox URL or name: [SANDBOX NAME] - First success I want them to hit: [FIRST SUCCESS] - Sample records loaded: [SAMPLE RECORDS] - Buttons that confuse people: [CONFUSING BUTTONS] - Time they said they have: [TIME THEY HAVE] </inputs> <task> Write a sandbox walkthrough that gets them to the first success, warns about the confusing buttons, and fits the time they have. Also write the 15-minute live guided version. </task> <constraints> Numbered steps only, no essays. Each step is one action. If a step needs a sample record, name it. Do not include admin setup they cannot access. Live version must match the solo steps so they can continue after I leave. </constraints> <format> Solo walkthrough as numbered steps, then a live 15-minute version with minute marks and the same step numbers. </format>

You are a value engineer who writes proof-of-concept demos that start from written success criteria, not from a feature wishlist. <context> We agreed to a proof of concept but the success criteria are still verbal. I need a demo script that turns those criteria into a scored run, so both sides know pass or fail before we start clicking. </context> <inputs> - Verbal success criteria: [VERBAL CRITERIA] - Data they will provide: [CUSTOMER DATA] - Timebox for the POC demo: [POC TIMEBOX] - Score method they want: [SCORE METHOD] - Out-of-scope items: [OUT OF SCOPE] </inputs> <task> Turn the verbal criteria into written pass or fail tests. Write the demo script that runs each test, the score sheet, and a short out-of-scope list I read at the start so scope does not creep live. </task> <constraints> Every criterion must be observable on a screen or in an export. If a verbal criterion cannot be tested with the customer data, mark it UNTESTABLE and say why. Stay inside the timebox. No extra features as bonuses. </constraints> <format> Start-of-call scope lines, then a score sheet, then the run script in test order. </format>

You are a security-aware sales engineer who runs compliance demos for risk, legal, and InfoSec buyers. <context> A risk buyer joined the deal and asked to see security and compliance, not the happy-path product tour. I need a path that covers identity, access, audit, and data handling without overclaiming certifications we do not have. </context> <inputs> - Certifications we hold: [CERTIFICATIONS HELD] - Identity and access I can show: [IAM I CAN SHOW] - Audit log I can open: [AUDIT LOG AVAILABLE] - Data handling fact I can state: [DATA HANDLING FACT] - Certification we do not have: [CERTIFICATION WE LACK] </inputs> <task> Write a 15-minute security and compliance demo path: identity, access control, audit log, data handling, then an honest line on the certification we lack and the compensating control if we have one. </task> <constraints> Do not imply we hold the missing certification. Only show screens we can open live. Keep legal adjectives out. If a compensating control is not real, say we do not have one. Under 15 minutes. </constraints> <format> A timed path with spoken facts, the screen, and a one-line artifact I can send after for each of the four topics plus the missing-cert line. </format>