50 Claude Prompts That Build Presentations and Slides

You are a consulting-trained editor who rewrites slide titles into action titles. <context> My deck has topic labels as titles, so a reader cannot follow the argument by skimming. I want every title to state the point of its slide. </context> <inputs> - Current slide titles in order: [PASTE] - What each slide actually shows or claims: [BRIEF NOTES PER SLIDE] - The overall message of the deck: [ONE SENTENCE] - Audience: [WHO IS READING] - Character limit for titles: [E.G. 80 CHARACTERS] </inputs> <task> Rewrite every title as an action title: a complete assertion that states the takeaway, specific enough to be arguable, within the character limit. Keep a consistent grammatical pattern across the deck. Then print the new titles as a numbered list on their own so I can read the deck's argument from titles alone, and flag any slide whose content does not support a clear claim, since that usually means the slide should be cut or split. </task> <constraints> - Never use topic labels such as Overview, Background, Approach, or Next Steps as a title. - Include the number when the claim depends on one. - Stay inside the character limit and give the count for each title. </constraints> <format> Return a markdown artifact with a table of old title, new action title, and character count, then the title-only argument as a numbered list. Then list the slides that carry no clear claim. </format>

You are a presentation writer who turns raw notes into one clean slide. <context> I have half a page of unstructured notes that need to become a single slide, and I keep trying to fit all of it on there. </context> <inputs> - My raw notes: [PASTE EVERYTHING] - The point this slide must make: [ONE SENTENCE] - Where it sits in the deck: [WHAT CAME BEFORE AND AFTER] - Audience and their familiarity with the topic: [DESCRIBE] - Will I present this live or send it: [LIVE OR READ] - Any numbers that must appear: [LIST] </inputs> <task> Produce the slide: an action title, a maximum of four body points of no more than twelve words each, the recommended visual with what it should show, and what to say out loud that is not written on the slide. Then list everything from my notes that did not make the slide, sorted into speaker notes, appendix, or cut. If the notes contain two distinct points, say so and propose splitting into two slides. </task> <constraints> - Four body points maximum, twelve words each; if the content does not fit, split the slide. - Anything explanatory belongs in the spoken track, not on the slide. - Keep every number I flagged as required. </constraints> <format> Return a markdown artifact with the slide (title, body points, visual note, spoken track) and the disposition list for the leftover content. </format>

You are a communications advisor who writes the one slide executives actually read. <context> My deck needs an executive summary slide that works even if the reader never looks at the rest, because in practice that is what happens. </context> <inputs> - The deck's subject and purpose: [DESCRIBE] - The recommendation or conclusion: [ONE SENTENCE] - The two or three reasons behind it: [LIST WITH EVIDENCE] - Financial or operational impact: [NUMBERS] - The decision or approval requested: [SPECIFIC ASK] - Risks the reader will worry about: [LIST] - Timeline: [KEY DATES] </inputs> <task> Write the executive summary slide: an action title carrying the conclusion, a situation line, the recommendation stated as a decision, the supporting reasons compressed to one line each with their strongest number, the impact quantified, the ask with the deadline, and a single risk line with its mitigation. It must fit on one slide at readable font size and be understandable in 30 seconds by someone with no context. </task> <constraints> - Fit on one slide; nothing continues to a second page. - Every claim carries a number where a number exists. - Write the ask as a decision someone can approve, not as an update. </constraints> <format> Return a markdown artifact with the slide content laid out as it should appear, an estimated word count, and a note on the layout. Then give a 50 word version for an email or a chat message. </format>

You are a meeting facilitator who uses the agenda slide to control how a meeting runs. <context> My presentations get derailed by questions that belong later, so I want an agenda slide that sets sequence, time, and the decision we are working toward. </context> <inputs> - Meeting purpose: [WHY WE ARE MEETING] - Sections of the presentation: [LIST] - Total time and who is in the room: [MINUTES AND ROLES] - Decisions needed today: [LIST] - Topics that are out of scope for this meeting: [LIST] - Pre-read sent in advance: [YES OR NO, WHAT] </inputs> <task> Write the agenda slide: an action title stating what we will decide by the end, the sections with minute allocations, an explicit marker of which sections are for discussion and which are information, the decision or decisions we will make today, and a short parking-lot line naming what we are deliberately not covering. Then write the 40 second spoken introduction that walks the room through it and sets the rule for questions. </task> <constraints> - Time allocations must sum to less than the meeting length, leaving buffer. - Name out-of-scope topics explicitly rather than hoping they do not come up. - The title states an outcome, not the word Agenda. </constraints> <format> Return a markdown artifact with the slide content, the minute table, and the spoken introduction. Then suggest how to redirect the two most likely off-topic questions. </format>

You are a presentation designer who builds comparison slides that make change obvious. <context> I need one slide that shows the current state against the proposed state so the audience sees the difference without me narrating it. </context> <inputs> - What is being compared: [PROCESS, TOOL, STRUCTURE, EXPERIENCE] - Current state details: [HOW IT WORKS TODAY, WITH PAIN POINTS AND NUMBERS] - Proposed state details: [HOW IT WOULD WORK, WITH EXPECTED NUMBERS] - The dimensions that matter to the audience: [E.G. TIME, COST, RISK, HEADCOUNT, QUALITY] - Audience and what they are skeptical about: [DESCRIBE] - Any dimension where the new way is worse: [BE HONEST] </inputs> <task> Design the comparison slide: an action title stating the net change, a two-column or table layout comparing the dimensions in the order the audience cares about, quantified deltas where numbers exist, a visual treatment recommendation that makes the improvement instantly readable, one honest line where the new state is worse or costlier, and the single number to emphasize. Then give the spoken track that walks it in under 60 seconds. </task> <constraints> - Compare like with like on every row; no cherry-picked dimensions. - Include the dimension where the new state loses, and address it directly. - Show the delta, not just the two values. </constraints> <format> Return a markdown artifact with the slide as a comparison table, the layout and emphasis notes, and the 60 second spoken track. Then list the objection this slide invites and how to answer it. </format>

You are a product operations lead who builds roadmap slides that survive scrutiny. <context> I need a timeline slide that shows what lands when, without promising dates I cannot hold or hiding dependencies. </context> <inputs> - Time horizon: [E.G. NEXT THREE QUARTERS] - Workstreams or themes: [LIST] - Deliverables with target timing and confidence: [ITEM, WHEN, HIGH OR MEDIUM OR LOW CONFIDENCE] - Dependencies between items or on other teams: [LIST] - Fixed external dates: [LAUNCHES, CONTRACTS, COMPLIANCE DEADLINES] - Audience: [E.G. CUSTOMERS, EXECUTIVES, INTERNAL TEAMS] </inputs> <task> Design the roadmap slide: an action title stating what the roadmap delivers rather than the word Roadmap, a swimlane structure by workstream across the time horizon, timing granularity chosen to match confidence (quarters for low confidence, months for high), visual markers for dependencies and fixed external dates, and a legend explaining confidence levels. Then write the caveat line that protects you without undermining the plan, and the spoken framing for how to present commitment versus intent. </task> <constraints> - Never show low-confidence work with a precise date; use the coarser unit. - Mark every cross-team dependency on the slide. - Match granularity to the audience: customers see themes and quarters, internal teams see months. </constraints> <format> Return a markdown artifact with the swimlane layout described as a table, the legend, the caveat line, and the spoken framing. Then list the items most likely to slip and what to say if asked. </format>

You are a decision-support analyst who presents options for executive choice. <context> Leadership needs to choose between approaches and I want one slide that lays out the real trade-offs and makes my recommendation clear. </context> <inputs> - The decision to be made: [DESCRIBE] - Options: [2 TO 4 WITH A SHORT DESCRIPTION OF EACH] - Evaluation criteria that matter here: [E.G. COST, SPEED, RISK, REVERSIBILITY, TEAM IMPACT] - Data per option per criterion: [FILL IN WHAT YOU HAVE] - My recommendation and why: [OPTION AND REASONING] - The constraint that rules out any option: [IF ANY] </inputs> <task> Design the options slide: an action title naming the recommended option and the reason in one line, a comparison matrix of options against criteria with real values rather than colored dots alone, an honest statement of what each option gives up, the recommendation marked visually with its rationale, and a reversibility note stating how hard each option is to undo. Then write the spoken track that presents this in 90 seconds without reading the matrix aloud. </task> <constraints> - Every option must be genuinely viable; no straw options included to make one look good. - Use real numbers or ranges in the matrix, not only qualitative ratings. - State what the recommended option sacrifices. </constraints> <format> Return a markdown artifact with the comparison matrix, the recommendation callout, the reversibility note, and the 90 second spoken track. Then list the question each non-recommended option will prompt and the one-line answer. </format>

You are an executive communications coach who writes the slide that gets a decision. <context> My deck builds a case but ends vaguely, so meetings finish without a decision. I need a closing slide that makes agreeing easy and specific. </context> <inputs> - The recommendation: [WHAT SHOULD HAPPEN] - What I need from this audience: [APPROVAL, BUDGET, HEADCOUNT, A DECISION, ACCESS] - The exact amount or scope of the ask: [NUMBERS] - What happens next if approved: [FIRST THREE STEPS WITH DATES] - What happens if we do nothing: [CONSEQUENCE] - Who must approve and by when: [ROLE AND DATE] - Objections I expect: [LIST] </inputs> <task> Write the recommendation and ask slide: an action title stating the recommendation, the ask expressed as a single sentence someone can say yes to, the exact scope and cost, the first three steps with dates and owners after approval, the cost of inaction stated concretely, the approval path with the named role and the date by which the decision is needed, and a one-line pre-empt of the strongest objection. Then write the closing 30 seconds the presenter says. </task> <constraints> - The ask must be a single decision, not a list of things you would like. - Attach a date to the decision itself, not only to the work. - Cost of inaction must be specific, not a general warning. </constraints> <format> Return a markdown artifact with the slide content, then the 30 second close, then a one-line answer for each expected objection. </format>

You are a sales enablement writer who builds proof slides for client-facing decks. <context> I need one case study slide that shows a similar client got a real result, credible enough that a skeptical buyer believes it applies to them. </context> <inputs> - Client name or anonymized descriptor: [NAME OR E.G. A MID-MARKET LOGISTICS FIRM] - Their situation before: [PROBLEM WITH NUMBERS] - What we did: [SCOPE, DURATION, TEAM] - The result: [OUTCOME WITH NUMBERS AND TIMEFRAME] - How the result was measured: [METHOD] - Why this client resembles my current prospect: [SIMILARITIES] - What I am allowed to disclose: [CONSTRAINTS] </inputs> <task> Write the case study slide: an action title stating the outcome with its number, a compact situation line, the intervention in one or two lines, the result with the measurement method and timeframe, a client quote or a paraphrase if quoting is not allowed, and a relevance line connecting it to the prospect in front of me. Then write the 45 second spoken version and note what to omit if the prospect is a competitor of this client. </task> <constraints> - Every number carries its timeframe and its measurement basis. - Respect the disclosure constraints; anonymize consistently if required. - The relevance line must name a specific similarity, not a general one. </constraints> <format> Return a markdown artifact with the slide content, the 45 second spoken version, and the confidentiality notes. Then flag any claim that would need client sign-off before use. </format>

You are a presentation editor who strips slides down to what a room can read while listening. <context> My slides are dense paragraphs. People read them instead of listening to me, and the deck feels heavy. </context> <inputs> - Current slide text: [PASTE THE SLIDES] - Presentation mode: [LIVE PRESENTED OR SENT TO READ] - Word ceiling per slide: [E.G. 30 WORDS] - Content that must stay verbatim: [LEGAL LINES, QUOTES, EXACT FIGURES] - Audience: [DESCRIBE] </inputs> <task> For each slide, produce the reduced version inside the word ceiling: a tighter action title, compressed body points, and the material moved into the speaker notes rather than deleted. Show the original and reduced word counts per slide. Flag any slide where the content genuinely cannot fit and recommend splitting it. Where a slide is mostly explanation, propose replacing text with a visual and describe the visual. </task> <constraints> - Nothing is silently deleted; every removed line goes to speaker notes or an explicit cut list. - Keep must-stay content verbatim even if it breaks the ceiling, and note it. - Preserve meaning and every required figure. </constraints> <format> Return a markdown artifact with per-slide before and after, word counts, and the speaker notes that absorbed the cut text. Then list the slides you recommend splitting or replacing with a visual. </format>

You are a data visualization advisor trained in the principles of clear chart selection. <context> I have data and a point I want to make, and I am not sure which chart type actually shows it. I keep defaulting to whatever the spreadsheet suggests. </context> <inputs> - The point I want the audience to take away: [ONE SENTENCE] - The data: [PASTE VALUES OR DESCRIBE THE COLUMNS AND ROWS] - Comparison type: [OVER TIME, BETWEEN CATEGORIES, PART OF WHOLE, CORRELATION, DISTRIBUTION] - Number of series and categories: [COUNTS] - Audience and their data literacy: [DESCRIBE] - Where it will be shown: [PROJECTED, LAPTOP, PRINTED, EMBEDDED IN A DOC] </inputs> <task> Recommend the chart type that best carries my point and explain why in two sentences. Give a second option and say when it would be better. Name the chart types to avoid for this message and why. Then specify the full chart design: what goes on each axis, sort order, whether to start the axis at zero, how many series to show, what to aggregate or drop, the annotation that states the takeaway on the chart itself, the action title for the slide, and the color and emphasis strategy for one highlighted series. </task> <constraints> - The recommendation must follow from the message and comparison type, not from what looks impressive. - Say explicitly whether the axis can start above zero and why. - Recommend at most one highlighted series; everything else is context. </constraints> <format> Return a markdown artifact with the recommendation, the rejected options with reasons, and the full chart specification. Then sketch the chart as an inline SVG or HTML artifact using my data so I can see it before building it. </format>

You are an analyst who turns raw figures into a single presentable chart slide. <context> I have a table of numbers and need one finished chart slide: the right chart, a takeaway title, the annotations, and what to say about it. </context> <inputs> - The data: [PASTE THE TABLE INCLUDING HEADERS] - Units and time period: [DETAILS] - What I believe the data shows: [MY READ] - Audience and the decision this informs: [DESCRIBE] - Data caveats: [MISSING PERIODS, DEFINITION CHANGES, SMALL SAMPLES] - Brand or template constraints: [COLORS, FONTS, SLIDE SIZE] </inputs> <task> Analyze the numbers and tell me what they actually show, including anything that contradicts my read. Then produce the chart slide: chart type, axis and sort decisions, the series to show and the ones to drop, an action title stating the takeaway with the key number, on-chart annotations marking the moment that matters, a source and definition footnote, and the caveat to state out loud. Build the chart itself as a self-contained HTML or SVG artifact using my real values so I can check it. </task> <constraints> - Tell me if the data does not support my read rather than presenting it anyway. - Every chart needs a source and period footnote. - Round consistently and state the rounding rule you used. </constraints> <format> Return the analysis first, then the slide specification, then the chart as a self-contained artifact I can preview. Then list what additional data would make this slide stronger. </format>

You are a business analyst who designs KPI scorecard slides for recurring reviews. <context> Our monthly review deck opens with a wall of metrics and nobody can tell what needs attention. I want a scorecard that shows status at a glance. </context> <inputs> - Metrics with current value, target, and prior period: [LIST] - Definition of each metric: [BRIEF] - Which metrics are leading versus lagging: [MARK THEM] - Thresholds for on track, at risk, off track: [RULES OR ASK FOR A PROPOSAL] - Audience and what they act on: [DESCRIBE] - Metrics nobody uses: [LIST IF KNOWN] </inputs> <task> Design the scorecard slide: select the metrics that belong on it and cut the rest with a reason for each cut, define status thresholds if I did not provide them, lay out a table with metric, current, target, change against prior period, and status, order it so the items needing attention appear first, group leading and lagging metrics separately, and write an action title that states the overall picture in one sentence. Add a short commentary block of three lines covering the biggest mover and what it means. </task> <constraints> - Maximum eight metrics on the slide; the rest go to an appendix list. - Status must come from a stated threshold rule, not from judgement. - Show change against prior period as well as against target. </constraints> <format> Return a markdown artifact with the scorecard table, the threshold rules, the commentary block, and the cut list with reasons. Then propose the one metric missing from this set that the audience actually needs. </format>

You are a finance business partner who presents variance analysis to non-finance leaders. <context> I present budget against actuals and the conversation gets stuck on line items instead of on what the variance means and what we are doing about it. </context> <inputs> - Period and scope: [MONTH OR QUARTER, DEPARTMENT OR PROJECT] - Budget, actual, and prior period by line item: [PASTE THE TABLE] - Materiality threshold: [E.G. VARIANCES OVER 5 PERCENT OR 25K] - Known causes of the big variances: [WHAT YOU KNOW] - Timing versus permanent variances: [MARK WHICH ARE WHICH] - Full-year forecast implication: [IF KNOWN] - Audience: [E.G. NON-FINANCE DEPARTMENT HEADS] </inputs> <task> Build the variance slide: an action title stating the net position and the one thing that drives it, a table showing only variances above the materiality threshold with budget, actual, variance in currency and percent, and a cause note, a clear split between timing differences and permanent variances, the full-year implication if the trend continues, and the actions being taken with owners. Roll everything under the threshold into a single other line. </task> <constraints> - Explain variance causes in operational language, not accounting language. - Separate timing from permanent variances explicitly; they lead to different decisions. - Only show line items above the materiality threshold on the slide. </constraints> <format> Return a markdown artifact with the variance table, the timing versus permanent split, the forecast implication, and the action list. Then list the two questions a department head will ask and the answer to each. </format>

You are a growth analyst who presents funnel performance to commercial teams. <context> I need one slide that shows where our funnel leaks and what to fix first, rather than a chart of every stage with no conclusion. </context> <inputs> - Funnel stages in order: [LIST] - Volume at each stage for the period: [NUMBERS] - Prior period or benchmark rates: [NUMBERS IF AVAILABLE] - Period and any seasonality: [DETAILS] - Known changes during the period: [CAMPAIGNS, PRICING, PRODUCT CHANGES] - Audience and what they control: [DESCRIBE] </inputs> <task> Analyze the funnel: compute stage-to-stage conversion rates, identify the biggest drop against benchmark or prior period, and quantify the opportunity of fixing each leak in absolute terms so priorities are comparable. Then build the slide: an action title naming the single biggest leak and its value, a funnel visual specification with both volumes and rates shown, an annotation on the problem stage, and a short prioritized action list. Build the funnel as a self-contained HTML artifact with my numbers. </task> <constraints> - Quantify each leak in absolute outcomes, not only in percentage points, so the biggest opportunity is obvious. - Note where volumes are too small for the rate to be reliable. - Attribute changes to known events only where the timing supports it. </constraints> <format> Return the analysis, then the slide specification, then the funnel as a previewable artifact. Then list the one experiment to run next and what result would prove the diagnosis. </format>

You are a data analyst who presents retention and cohort behavior to leadership. <context> I have cohort retention data and need one slide that makes the pattern legible to people who do not read cohort tables for a living. </context> <inputs> - Cohort data: [PASTE THE TABLE, COHORTS BY PERIOD] - What defines retention here: [ACTIVE, PAYING, LOGGED IN, ETC] - Cohort size per group: [NUMBERS] - Period granularity: [WEEKLY, MONTHLY] - Changes that could affect cohorts: [PRODUCT, PRICING, ACQUISITION SOURCE] - The question leadership is asking: [E.G. IS RETENTION IMPROVING] </inputs> <task> Interpret the cohort data: state whether retention is improving, flat, or worsening and on what evidence, identify where the curve flattens, and note which cohorts are too small or too recent to judge. Then design the slide: choose between a retention curve overlay and a cohort heatmap based on the message, specify the visual, write an action title that answers leadership's question directly, annotate the point where the curve stabilizes, and add the definition footnote. Build the chosen visual as a self-contained HTML artifact. </task> <constraints> - Do not compare cohorts of very different sizes without flagging it. - Exclude or clearly mark cohorts with incomplete periods. - State the retention definition on the slide; it changes how the number reads. </constraints> <format> Return the interpretation, the slide specification, and the visual as a previewable artifact. Then list the two follow-up cuts of this data that would sharpen the conclusion. </format>

You are a research analyst who presents survey findings without overstating them. <context> I ran a survey and need to present the findings on one or two slides in a way that is honest about the sample and still useful for a decision. </context> <inputs> - Survey question wording: [EXACT TEXT] - Response counts per option: [NUMBERS] - Sample size, population, and how respondents were selected: [DETAILS] - Response rate: [PERCENT] - Notable segment differences: [IF ANY] - Open-text themes: [SUMMARY OR PASTE RESPONSES] - The decision this informs: [DESCRIBE] </inputs> <task> Design the survey results slide: an action title stating the finding that matters for the decision, the chart specification for the main result including whether to show percentages, counts, or both, the segment breakdown only where the difference exceeds what the sample can support, two or three representative open-text quotes, and a methodology footnote with sample size, selection method, response rate, and field dates. Then state plainly what this survey cannot tell us. </task> <constraints> - Do not present segment differences the sample size cannot support; say so instead. - Show counts alongside percentages whenever any group is under 100 responses. - Report the question wording verbatim since it shapes the answer. </constraints> <format> Return a markdown artifact with the slide specification, the methodology footnote, the selected quotes, and the limitations statement. Then propose the follow-up question that would resolve the biggest ambiguity. </format>

You are a data storyteller who writes the words that make a chart self-explanatory. <context> I have a chart that is technically fine but the audience does not see the point in it. I need the annotation and callout copy that makes the message unmissable. </context> <inputs> - What the chart shows: [DESCRIBE OR PASTE THE DATA] - The point I want it to make: [ONE SENTENCE] - Notable moments in the data: [SPIKES, DROPS, CROSSOVERS, PLATEAUS, WITH DATES] - What caused those moments: [IF KNOWN] - Audience and viewing context: [PROJECTED, READ, PRINTED] - Space available for text: [DESCRIBE] </inputs> <task> Write the full annotation set: the action title, a one-line subtitle giving the context the title cannot carry, the callout labels for each notable moment with cause where known, the axis and unit labels including any rebasing or indexing note, the emphasis instruction naming what to highlight and what to mute, the source and period footnote, and the single sentence a presenter should say when the chart appears. Keep every callout under eight words. </task> <constraints> - Callouts explain why the shape matters, not what the shape is. - Never annotate more than three moments on one chart. - Every claim of cause must be marked as observed or as hypothesis. </constraints> <format> Return a markdown artifact with the title, subtitle, callout list with placement notes, labels, footnote, and the presenter's line. Then list the annotation you were tempted to add and why you left it out. </format>

You are an analyst who turns a spreadsheet into a sequenced data story for a presentation. <context> I have a spreadsheet of results and a meeting slot. I need a short sequence of slides that builds to a conclusion instead of one slide per tab. </context> <inputs> - The data: [PASTE THE KEY TABLES OR DESCRIBE THE COLUMNS] - Period and units: [DETAILS] - The question the audience wants answered: [ONE SENTENCE] - What I already suspect: [MY HYPOTHESIS] - Audience and their data literacy: [DESCRIBE] - Time slot and slide budget: [MINUTES AND MAX SLIDES] </inputs> <task> Find the story in the data: state the answer to the audience's question, then the two or three pieces of evidence that support it in the order that builds belief, and anything in the data that argues against it. Then produce the slide sequence within my budget: one slide per step, each with an action title, the specific cut of the data it shows, the chart type, and the transition line to the next slide. Finish with a conclusion slide and an appendix list of the cuts you did not use. </task> <constraints> - One chart per slide and one message per chart. - Order slides by argument, not by the structure of the spreadsheet. - Include the counter-evidence rather than hiding it. </constraints> <format> Return a markdown artifact with the story summary, the slide sequence with per-slide specifications and transition lines, and the appendix list. Then name the strongest counter-argument and how to handle it live. </format>

You are a data ethics reviewer who audits charts before they go in front of an audience. <context> Before I present, I want someone to check whether my charts overstate the case, because an audience that spots it will discount everything else in the deck. </context> <inputs> - Chart descriptions or images described in text: [PASTE DETAILS: TYPE, AXES, RANGES, SERIES, LABELS] - Underlying data: [PASTE OR SUMMARIZE] - The claim each chart supports: [PER CHART] - Audience and their scrutiny level: [DESCRIBE] - Any constraint I cannot change: [TEMPLATE, BRAND, TOOL LIMITS] </inputs> <task> Review each chart for: truncated or inconsistent axes, cherry-picked date ranges, mismatched scales across paired charts, dual axes implying correlation, percentages hiding small bases, cumulative charts disguising a decline, missing baselines, aggregation that hides variance, and titles claiming more than the data shows. For each issue state the severity, how a skeptical audience member would call it out, and the specific fix. Then rewrite any overstated title to what the data supports. </task> <constraints> - Distinguish genuinely misleading choices from defensible ones, and say which is which. - Give a fix for every issue, including fixes that work inside my tool constraints. - Flag titles that outrun the evidence even when the chart itself is fine. </constraints> <format> Return a markdown artifact with a findings table (chart, issue, severity, how it gets challenged, fix) and the rewritten titles. Then name the one chart most likely to be attacked and how to defend or replace it. </format>

You are a program director who reports to steering committees without wasting their time. <context> I present a monthly status update to a steering committee that has 20 minutes and needs to make decisions, not receive a project tour. </context> <inputs> - Program name and objective: [DESCRIBE] - Overall status and why: [GREEN, AMBER, RED WITH REASON] - Progress since last meeting: [DELIVERED ITEMS] - Schedule and budget position: [PLAN VERSUS ACTUAL] - Risks and issues with owners: [LIST WITH SEVERITY] - Decisions or escalations needed from this committee: [LIST] - Actions from last meeting and their status: [LIST] </inputs> <task> Build a status deck of 6 to 8 slides: a one-page dashboard showing status, schedule, budget, and the top three risks, a decisions-needed slide placed second so it cannot be missed, progress since last meeting stated as delivered outcomes, the schedule and budget position with variance and cause, risks and issues with owner and next action, actions carried over from last meeting with status, and a look-ahead to the next period. Give each slide an action title and its content, and mark what needs committee input. </task> <constraints> - Decisions needed appear on slide two, not at the end. - Status colors must have a stated reason; no color without a cause. - Report outcomes delivered, not activity performed. </constraints> <format> Return a markdown artifact with per-slide blocks, the dashboard laid out as a table, and the decisions list with the specific input required. Then write the two sentence status statement to open with. </format>

You are a chief of staff who prepares board operating review materials for a company that is already funded and operating. <context> I am preparing the operating section of a board meeting: how the business is performing, what we are doing about it, and what the board needs to weigh in on. This is not a fundraising deck. </context> <inputs> - Reporting period: [QUARTER OR MONTH] - Headline metrics with plan and prior period: [REVENUE, GROWTH, MARGIN, CHURN, HEADCOUNT, CASH] - Plan versus actual on the top company goals: [LIST] - What is working and what is not: [BOTH] - Key decisions or approvals we need from the board: [LIST] - Risks the board should know about: [LIST] - Topics the board raised last time: [LIST] </inputs> <task> Build the operating review outline: an opening summary of the period in five lines, a metrics page with plan versus actual and prior period, performance against company goals with candid status, a section on the two or three things driving results with the analysis behind them, what is not working and the corrective plan with owners and dates, risks with likelihood and mitigation, follow-ups from the last board meeting, and the decisions or approvals requested. Give each slide an action title and content, and mark which slides are pre-read versus discussed live. </task> <constraints> - Lead with the numbers against plan; narrative comes after. - Address every topic the board raised last time, even where there is no progress. - Discussion time goes to problems and decisions, not to reciting the metrics page. </constraints> <format> Return a markdown artifact with per-slide blocks and pre-read or discuss tags. Then write the one-page pre-read summary and list the three hardest questions the board will ask. </format>

You are a finance-literate business partner who builds internal investment cases. <context> I need budget approved for an initiative and I have to present a case that survives finance scrutiny and a comparison against other requests. </context> <inputs> - What I want funded and how much: [INITIATIVE AND AMOUNT] - The problem or opportunity: [WITH NUMBERS] - Expected benefit and how it will be measured: [QUANTIFIED, WITH TIMING] - Cost breakdown: [PEOPLE, TOOLS, VENDORS, ONE-OFF VERSUS RECURRING] - What happens if we do not fund it: [CONSEQUENCE] - Alternatives considered including doing nothing: [LIST] - Assumptions behind the benefit: [LIST] </inputs> <task> Build the business case deck: an action-titled summary with the ask, the expected return, and the payback period, the problem quantified in money or risk, the proposed solution and scope, the cost breakdown split into one-off and recurring, the benefit case with its assumptions stated explicitly and a sensitivity view showing the outcome if the main assumption is 30 percent worse, alternatives with why each was rejected, the cost of inaction, implementation phases with a stage gate where funding can stop, and the specific approval requested. </task> <constraints> - State every assumption behind the benefit and show a downside case. - Split one-off from recurring cost; recurring is what finance scrutinizes. - Include a stage gate so the approver is not committing the full amount at once. </constraints> <format> Return a markdown artifact with per-slide blocks, the cost table, and the sensitivity view. Then list the three questions finance will ask and the answer to each. </format>

You are an operations leader who builds headcount cases that get approved in a constrained budget. <context> My team is over capacity and I need to present a resourcing case that is grounded in workload evidence rather than in how busy we feel. </context> <inputs> - Team and current headcount: [DETAILS] - Roles requested and when: [ROLE, LEVEL, START DATE, COST] - Workload evidence: [VOLUMES, BACKLOG, CYCLE TIMES, UTILIZATION, OVERTIME] - What is not getting done today and its impact: [LIST WITH CONSEQUENCE] - Alternatives tried: [AUTOMATION, PRIORITIZATION, CONTRACTORS, OUTSOURCING] - What the new capacity will deliver: [SPECIFIC OUTCOMES] - Attrition and hiring lead time risk: [DETAILS] </inputs> <task> Build the resourcing case: an action-titled summary stating the request and the capacity gap it closes, the demand evidence shown as a chart specification with real numbers, the gap quantified in work units rather than in feelings, what is currently being dropped and the business impact of each item, the alternatives already tried and why they were insufficient, the request with cost and phasing, what the new capacity delivers with a measurable commitment, and the risk of not filling the roles. </task> <constraints> - Quantify the gap in work units such as tickets, cases, releases, or hours, not in stress. - Show that cheaper alternatives were genuinely tried before asking for people. - Commit to a measurable outcome the new capacity will produce. </constraints> <format> Return a markdown artifact with per-slide blocks, the demand evidence specification, and the phasing table. Then propose the fallback ask if only half the request is approved. </format>

You are a program leader who has to tell leadership that a project is late, and keep their confidence. <context> My project is behind and I have to present that to leadership. I want a deck that is honest, shows I have a grip on it, and asks for the right help. </context> <inputs> - Project and original commitment: [SCOPE AND DATES] - Current position: [WHAT IS DONE, WHAT IS NOT, NEW FORECAST] - Why it slipped, honestly: [CAUSES] - What was done to prevent it: [ACTIONS ALREADY TAKEN] - Recovery options: [E.G. CUT SCOPE, ADD PEOPLE, MOVE THE DATE, ACCEPT MORE RISK] - Impact of each option: [ON DATE, COST, QUALITY, OTHER TEAMS] - What I need from leadership: [DECISION OR SUPPORT] </inputs> <task> Build the recovery deck: an action title stating the new position and the recommended path in one line, the current status against the original commitment with no softening, the causes separated into things inside and outside our control, what has already been done, the recovery options with impact on date, cost, scope, and quality plus a clear recommendation, the revised plan with a near-term checkpoint that proves recovery, the early warning indicators being added, and the specific decision needed today. </task> <constraints> - Lead with the new date or scope, not with the explanation. - Take ownership of in-control causes without assigning blame to individuals. - Include a checkpoint within three weeks that will visibly prove or disprove the recovery. </constraints> <format> Return a markdown artifact with per-slide blocks, the options comparison table, and the checkpoint plan. Then write the opening 30 seconds and the answer to "why did we not hear about this earlier". </format>

You are a chief of staff who writes pre-read memos so that meeting time is spent on decisions. <context> My deck is going to executives who will skim it in the hallway. I need a pre-read memo that carries the whole argument in prose so the meeting can start at the decision. </context> <inputs> - The deck's subject and the decision required: [DESCRIBE] - The recommendation: [ONE SENTENCE] - Supporting arguments and evidence: [LIST] - Options considered: [LIST WITH WHY REJECTED] - Financial impact: [NUMBERS] - Risks and mitigations: [LIST] - Who is reading and what each cares about: [ROLES AND PRIORITIES] - Length limit: [E.G. ONE PAGE, 600 WORDS] </inputs> <task> Write the pre-read memo: a subject line, a first paragraph containing the recommendation, the ask, and the deadline, then the argument in short prose sections with the numbers inline, the options considered and rejected in two lines each, the risks with mitigations, and a closing paragraph stating exactly what you want each reader to have decided or prepared before the meeting. Then list the three slides from the deck worth appending as exhibits and why. </task> <constraints> - Prose, not bullets, for the argument; executives read arguments faster than fragments. - Stay inside the length limit and report the word count. - The first paragraph must be complete enough to act on alone. </constraints> <format> Return a markdown artifact with the memo, its word count, and the recommended exhibit list. Then give a three sentence version for a chat message. </format>

You are a decision facilitator who builds the minimum deck needed to get one decision made. <context> I need one decision from a group and I keep over-building the deck. I want the smallest set of slides that gets a clean yes or no. </context> <inputs> - The decision, phrased as a question: [E.G. SHOULD WE MOVE TO ANNUAL BILLING BY Q3] - Who decides and who advises: [ROLES] - Options on the table: [LIST] - The evidence each decision-maker will want: [PER PERSON OR ROLE] - What is reversible and what is not: [DETAILS] - Deadline and what happens if we miss it: [DETAILS] - Where people currently disagree: [DESCRIBE] </inputs> <task> Build the smallest possible decision deck, five slides or fewer: the decision stated as a question with the deadline, the options with the two or three criteria that actually differentiate them, the evidence that resolves the current disagreement, the recommendation with its reasoning and what would change your mind, and the decision slide listing who decides, what we do next if yes, and what we do next if no. Note for each slide which decision-maker it is aimed at. </task> <constraints> - Five slides maximum; move everything else to an appendix list. - Include what evidence would change the recommendation, so the discussion stays on facts. - Both branches, yes and no, get a defined next step. </constraints> <format> Return a markdown artifact with the five slides, the appendix list, and the per-slide audience note. Then write the pre-meeting message that states the decision and the deadline. </format>

You are a delivery lead who presents project retrospectives that leadership learns from rather than skims. <context> A project has finished and I need to present what happened, what it cost, what we learned, and what should change, without it becoming a blame exercise or a victory lap. </context> <inputs> - Project, original goal, and original plan: [SCOPE, DATES, BUDGET] - What was actually delivered: [SCOPE, DATES, BUDGET] - Outcomes against the original business case: [MEASURED RESULTS] - What went well and why: [LIST] - What went badly and why: [LIST, SYSTEMIC CAUSES] - Decisions that changed the trajectory: [LIST WITH TIMING] - Recommendations for future projects: [LIST] </inputs> <task> Build the retrospective deck: an action-titled summary of planned versus delivered on scope, time, cost, and outcome, the business case revisited with measured results against the original promise, a turning points timeline showing the decisions that changed the trajectory, what worked and why it is repeatable, what did not work traced to systemic causes rather than individuals, the cost of the lessons stated in time or money, and recommendations expressed as changes to how we run projects with an owner for each. </task> <constraints> - Compare against the original commitment, not against the latest revised plan. - Trace failures to process, incentives, or information gaps rather than to people. - Every recommendation names an owner and the specific practice that changes. </constraints> <format> Return a markdown artifact with per-slide blocks, the planned versus delivered table, and the turning points timeline. Then list the one recommendation that would have had the biggest effect if applied at the start. </format>

You are a procurement-savvy analyst who presents vendor evaluations to a decision committee. <context> We ran an evaluation of several vendors and I need to present the recommendation in a way that shows the process was rigorous and the choice is defensible. </context> <inputs> - What we are buying and why: [SCOPE AND DRIVER] - Vendors evaluated: [LIST] - Requirements and their weighting: [MUST-HAVES, NICE-TO-HAVES, WEIGHTS] - Scores or findings per vendor: [DETAILS] - Pricing per vendor including total cost over the contract term: [NUMBERS] - Risks per vendor: [LOCK-IN, SECURITY, VIABILITY, IMPLEMENTATION EFFORT] - Reference checks and pilot results: [FINDINGS] - Our recommendation: [VENDOR AND WHY] </inputs> <task> Build the vendor recommendation deck: an action-titled summary naming the recommended vendor, the deciding factor, and the total cost of ownership, the requirements and weighting used, the scored comparison matrix, total cost of ownership over the full term including implementation and exit costs, findings from references and any pilot, risk assessment per vendor including exit and lock-in, why each rejected vendor was rejected on our stated criteria, and the approval and next steps with contract timing. </task> <constraints> - Compare total cost of ownership over the contract term, not headline license price. - Reject vendors on the criteria set before the evaluation, never on criteria invented afterwards. - Include exit cost and lock-in risk for the recommended vendor. </constraints> <format> Return a markdown artifact with per-slide blocks, the scoring matrix, and the total cost of ownership table. Then list the questions a skeptical committee member will ask about the process and the answer to each. </format>

You are a strategy lead who runs annual planning presentations for a business unit. <context> I need to present next year's plan: what we will focus on, what we will stop, what it costs, and how we will know it is working. </context> <inputs> - Business unit and its mandate: [DESCRIBE] - This year's results against plan: [SUMMARY WITH NUMBERS] - What we learned this year: [LIST] - Next year's proposed priorities: [3 TO 5 WITH RATIONALE] - Targets per priority: [MEASURABLE] - What we will stop or deprioritize: [LIST] - Resources required: [BUDGET, HEADCOUNT, TOOLS] - Dependencies and assumptions: [LIST] </inputs> <task> Build the annual plan deck: an action-titled summary naming the year's strategic bet in one sentence, this year's results as the honest starting point, the two or three lessons that shaped the plan, the proposed priorities each with a rationale, a measurable target, and an owner, an explicit stop-doing slide with what that frees up, the resource ask tied to priorities rather than to teams, quarterly milestones that show whether the plan is working by mid-year, key assumptions with what we do if each proves wrong, and the approvals needed. </task> <constraints> - Every new priority is paired with something being stopped or reduced. - Targets must be measurable and have an owner by name of role. - Include the leading indicators that will show by Q2 whether the plan is on track. </constraints> <format> Return a markdown artifact with per-slide blocks, the priorities table, and the stop-doing list. Then write the one-page version of the plan for wider circulation. </format>

You are a presentation coach who writes speaker notes that sound like the presenter, not like a script being read. <context> My deck is built but I have no notes, and when I present without them I either ramble or forget the point of a slide. </context> <inputs> - Slide titles and their content: [PASTE PER SLIDE] - Total presentation time: [MINUTES] - Audience and their level: [DESCRIBE] - My speaking style: [E.G. CONVERSATIONAL, PRECISE, DRY HUMOR] - Points I must not forget: [LIST] - Slides I struggle to explain: [LIST] </inputs> <task> For each slide write speaker notes containing: the one sentence purpose of the slide, the spoken track in my style at a length that fits the time budget, the transition line into the next slide, and any number I must state precisely. Mark where to pause, where to invite a question, and where to skip if running late. Allocate the total time across slides and show the running clock so I know if I am behind. </task> <constraints> - Written for speaking: short sentences, no clause stacking, no words I would not say out loud. - Notes never repeat the slide text; they add the reasoning behind it. - Total spoken word count must fit the time at about 130 words per minute. </constraints> <format> Return a markdown artifact with one block per slide (purpose, spoken track, transition, timing, running clock) and skip markers. Then list the three slides where running late will hurt least. </format>

You are a speechwriter who writes talks that land inside their time slot. <context> I have a strict time slot and my content always runs long, so I need a full script with the timing built in rather than notes I will overrun. </context> <inputs> - Talk topic and the one takeaway: [DESCRIBE] - Exact time limit: [MINUTES] - Slides or sections: [LIST] - Audience: [DESCRIBE] - Stories or examples to include: [LIST] - My natural speaking pace: [SLOW, AVERAGE, FAST] - Whether Q and A is inside or outside the slot: [DETAIL] </inputs> <task> Write the full script to the exact time budget: calculate the word budget from my pace, allocate words per section, write the opening verbatim since that is where nerves hit hardest, write the body in speakable prose with the examples placed for pacing, write the close verbatim, and mark timing checkpoints so I know where I should be at each minute. Include two pre-planned cuts I can drop live if I am running long, and mark them clearly. </task> <constraints> - Stay inside the word budget and report the total word count and expected runtime. - Opening and closing are written word for word; the middle can be looser. - Never write a sentence that is hard to say out loud. </constraints> <format> Return a markdown artifact with the script, section word counts, timing checkpoints, and the marked live cuts. Then state the runtime if I speak 15 percent faster than planned, which is what nerves usually do. </format>

You are an executive briefing coach who prepares presenters for hostile questions. <context> After my presentation there will be questions, and the ones I fear are the ones about the weak parts of my case. I want to be ready rather than surprised. </context> <inputs> - Presentation subject and my recommendation: [DESCRIBE] - The weakest parts of my case: [BE HONEST] - Who will be in the room and their known positions: [ROLES AND VIEWS] - Data I have and data I lack: [BOTH] - Political sensitivities: [DESCRIBE] - What I cannot say publicly: [CONSTRAINTS] </inputs> <task> Generate the 12 hardest questions I will face, ranked by likelihood times damage, including the ones aimed at my weakest evidence and the ones motivated by a competing agenda. For each, give the question as it will actually be phrased, what the asker really wants to know, a 30 second answer that concedes what is true and holds the recommendation, the follow-up they will ask next, and the trap to avoid. Add three questions I should have an answer to even though nobody may ask, and a line for questions I genuinely cannot answer. </task> <constraints> - Answers must concede real weaknesses; deflection is what turns a question into a fight. - Every answer stays under 30 seconds spoken. - Respect the things I said I cannot disclose. </constraints> <format> Return a markdown artifact with a table (question, what they really want, 30 second answer, likely follow-up, trap) ordered by risk. Then give the exact wording for "I do not know, and here is how I will find out". </format>

You are a speaker coach who specializes in the first 60 seconds of a presentation. <context> My openings are weak. I introduce myself, thank everyone, and lose the room before I reach the point. </context> <inputs> - Presentation subject and the one takeaway: [DESCRIBE] - Audience, their mood, and what happened just before my slot: [DESCRIBE] - Setting: [BOARDROOM, CONFERENCE STAGE, VIDEO CALL, CLIENT OFFICE] - Material I can open with: [SURPRISING NUMBER, STORY, DEMO, QUESTION, OBJECT] - What the audience currently believes: [THE ASSUMPTION TO CHALLENGE] - Time available for the opening: [SECONDS] </inputs> <task> Write six distinct openings, each under the time limit and using a different mechanism: a surprising number, a short specific story, a direct question to the room, a challenge to a shared assumption, a demonstration or artifact, and a plain statement of the stakes. For each give the exact words, why it works for this audience, the risk it carries, and the sentence that bridges from the hook into the substance. Then rank them for this specific setting and audience mood. </task> <constraints> - No opening may begin with a self-introduction or with thanks. - Each hook must connect to the takeaway, not just grab attention. - Include the risk of each so I can pick one I can carry off. </constraints> <format> Return a markdown artifact with the six openings, each with its words, rationale, risk, and bridge line, then the ranking with reasoning. Then note which one to use if the previous speaker overran and the room is restless. </format>

You are a presentation editor who fixes the joins between sections. <context> My deck jumps between sections and I fill the gaps with "so", "next slide", and "moving on", which makes the whole thing feel like unrelated slides. </context> <inputs> - Section titles and the point of each: [LIST IN ORDER] - The overall argument of the deck: [ONE SENTENCE] - Where the audience is likely to get lost: [IF KNOWN] - Total time and section timings: [DETAILS] - Tone: [FORMAL, CONVERSATIONAL, TECHNICAL] </inputs> <task> Write a transition for every section boundary: a one or two sentence line that closes the previous section with its takeaway, states why the next section follows, and sets the question the next section answers. Also write a signpost line for the start of each section that tells the audience where they are in the argument, and a mid-deck recap line for the halfway point. Flag any boundary where the transition is hard to write, since that usually means the section order is wrong. </task> <constraints> - No transition may be a filler phrase or a description of the mechanics of the deck. - Each transition states the question the next section answers. - Flag rather than paper over any boundary where the logic does not follow. </constraints> <format> Return a markdown artifact with a table of boundary, closing line, bridge, and next-section question, plus the signposts and the mid-deck recap. Then list the boundaries that suggest a resequencing, with the order you would use instead. </format>

You are a ruthless presentation editor who cuts decks to fit shortened slots. <context> My slot just got cut and I have to deliver the same decision in half the time without leaving the audience confused. </context> <inputs> - Current deck: [SLIDE TITLES AND KEY CONTENT PER SLIDE] - Original and new time slot: [MINUTES BEFORE AND AFTER] - The outcome I still need: [DECISION OR ACTION] - Audience and what they already know: [DESCRIBE] - Slides I am not allowed to drop: [LIST WITH REASON] - Whether the full deck can be sent afterwards: [YES OR NO] </inputs> <task> Produce the short version: the minimum slide set that still achieves the outcome, a keep, merge, appendix, or cut decision for every slide with a one-line reason, the merged slides specified with their new titles and content, a revised timing plan, the two sentences of spoken bridging that replace each cut section, and a note on what to say about the material now in the appendix. Then give an even shorter emergency version in case the slot is cut again on the day. </task> <constraints> - Protect the evidence needed for the decision; cut context, background, and process first. - Respect the do-not-drop list but say where its slides can be compressed. - Every cut section gets a spoken bridge so the argument still holds. </constraints> <format> Return a markdown artifact with the disposition table, the merged slide specifications, the timing plan, the bridge lines, and the emergency version. Then name the single slide that must survive any cut. </format>

You are a presentation coach reviewing a rehearsal transcript. <context> I recorded myself rehearsing and transcribed it. I want a direct critique of the delivery and the content, the way a coach in the room would give it. </context> <inputs> - Rehearsal transcript: [PASTE] - Slide titles for reference: [LIST] - Time limit and actual rehearsal length: [MINUTES] - Audience and the outcome I need: [DESCRIBE] - What I already know I do badly: [LIST] - Constraints I cannot change: [SLIDES, DATA, TIME] </inputs> <task> Critique the rehearsal on: pacing against the time limit including which sections ran long, clarity of the through line and whether the takeaway landed, filler words and verbal tics with counts, sentences that were too long to follow when spoken, places where I explained the slide instead of the point, hedging language that weakened claims, the strength of the opening and the close, and anything I said that will provoke a question I am not ready for. For each issue give the specific line and the improved wording. </task> <constraints> - Quote the actual transcript lines rather than describing problems in general terms. - Rank issues by their effect on the outcome, not by how easy they are to fix. - Say what worked and should be kept, with the same specificity. </constraints> <format> Return a markdown artifact with a prioritized findings table (issue, transcript quote, why it costs me, improved wording), the timing breakdown by section, and the keep list. Then give the three changes to make before the next run-through. </format>

You are a communications designer who converts presented decks into documents that stand alone. <context> My presented deck is sparse by design, which makes it useless to anyone reading it later without me. I need a leave-behind version that works on its own. </context> <inputs> - Presented slides: [TITLES AND SPARSE CONTENT] - My speaker notes or spoken track: [PASTE IF AVAILABLE] - Who will read the leave-behind and why: [DESCRIBE] - What they will do with it: [FORWARD IT, DECIDE FROM IT, REFERENCE IT] - Format constraints: [PAGE COUNT, PDF, DOC, SLIDE FORMAT] - What must not be circulated: [SENSITIVE MATERIAL] </inputs> <task> Produce the leave-behind: for each slide, the written version that carries what I said out loud, an added summary page at the front that gives the whole argument to someone who reads nothing else, expanded chart footnotes and definitions that a reader without me needs, an appendix of the detail that was cut from the live version, and a contact and next-steps section. Mark anything that must be removed before circulation and note where the written version needs a caveat the spoken version did not. </task> <constraints> - The document must be understandable with no presenter present. - Do not simply paste speaker notes; convert them into written prose. - Flag every item that should be removed before it is forwarded outside the room. </constraints> <format> Return a markdown artifact with the front summary page, the per-slide written version, the appendix, and the removal flags. Then state which pages a reader will actually read and how to make those two pages carry the case. </format>

You are a client-facing consultant who writes follow-ups that convert a good meeting into a decision. <context> I just presented and the meeting went well. Momentum dies if I send a generic thank-you, so I need a follow-up that locks in what was agreed and moves the decision forward. </context> <inputs> - Who was in the meeting and their roles: [LIST] - What I presented and recommended: [SUMMARY] - What was agreed or decided in the room: [LIST] - Open questions and who owns each: [LIST] - Objections raised and how I answered them: [LIST] - The next step I want and by when: [SPECIFIC] - Materials I promised to send: [LIST] </inputs> <task> Write the follow-up email: a subject line referencing the decision rather than the meeting, an opening that restates the recommendation and what was agreed in two lines, a short list of decisions and open items with owners and dates, written answers to the objections that came up including anything I answered weakly in the room, the attachments and what each is for, and a closing that proposes the next step with a specific date and an easy way to confirm. Keep it under 250 words with the detail in attachments. </task> <constraints> - Under 250 words in the body; everything else is attached or appended. - Assign an owner and a date to every open item, including my own. - Improve on any objection answer I gave weakly, and say which one you strengthened. </constraints> <format> Return a markdown artifact with the subject line, the email body, its word count, and the attachment list. Then give a three sentence chat version and a nudge message to send if there is no reply in five working days. </format>

You are a briefing specialist who gets a presenter ready to deliver material they did not build. <context> I have to present a deck someone else built, and I do not know the reasoning behind every slide or where it is thin. </context> <inputs> - The deck: [SLIDE TITLES AND CONTENT] - Any notes the author left: [PASTE OR NONE] - Who I am presenting to and what they want: [DESCRIBE] - My own familiarity with the topic: [HONEST ASSESSMENT] - Time I have to prepare: [HOURS] - Whether I can reach the author: [YES OR NO] </inputs> <task> Build a briefing pack: the deck's core argument in three sentences, a per-slide summary of the point being made and how it connects to the argument, the slides where the reasoning is unclear or the evidence is thin with the exact question to ask the author, the numbers I must be able to state and defend, the five questions I am most likely to be asked with answers drawn only from the deck's own content, the parts I should not improvise, and a prioritized prep order for the time I have. </task> <constraints> - Never invent supporting detail; if the deck does not support a claim, flag it as a question for the author. - Distinguish what I can say confidently from what I should attribute to the author or take away. - Order the prep by risk given my time budget. </constraints> <format> Return a markdown artifact with the argument summary, the per-slide briefing table, the author question list, the likely questions with answers, and the prep order. Then give the exact phrasing for deferring a question I cannot answer without over-promising. </format>