SuperAPI AI Directory

presentation-architect

Office & productivity

View repository

Updated 10-15 21:03

About this skill

Presentation Architect — Phase 1: Overall Design (★Core Skill)

Overview

PrezenX Suite'sCore Skill. From the presentation title and basic information, design the audience analysis, storyline, slide structure, and time allocation, and generate the design documents that form the foundation for subsequent phases (content creation and HTML generation).

Design Principles

  1. One Slide, One Message Principle — Each slide has only one main message
  2. Audience Concretization Principle — Not "executives" but "Mr. Yamada, General Manager of Sales"
  3. Behavior Change Design — The purpose of a presentation is to "change the audience's behavior"
  4. Evidence-Driven — Link data and sources to claims

Execution Steps

Step 1: Requirements Definition (requirements.md generation)

Collect the following information from the user and generaterequirements.md . If there is missing information,present specific optionsand confirm.

Requirements template: requirements.md

# プレゼンテーション要件定義

## 1. 基本情報
| 項目 | 内容 |
|------|------|
| タイトル | {仮タイトル — Phase 3 で確定} |
| 発表日 | {YYYY-MM-DD} |
| 持ち時間 | {N分}(Q&Aバッファ: {M分}含む) |
| 会場 | {会場名 / オンライン} |
| 発表者 | {氏名・役職} |
| 言語 | {日本語 / 英語 / バイリンガル} |

## 2. 聴衆分析
| 項目 | 内容 |
|------|------|
| 参加者数 | {N名}(推定) |
| 主要人物 | {具体名・役職 — 意思決定権限を持つ人} |
| 知識レベル | {技術: 高/中/低, ビジネス: 高/中/低} |
| 態度 | {好意的 / 中立 / 懐疑的} |
| 意思決定権限 | {予算承認 / 方針決定 / 技術選定 / なし} |
| 関心事項 | {ROI / 技術革新 / リスク / 競争力 / 政策整合性} |

## 3. 目的設定
| 項目 | 内容 |
|------|------|
| 主目的(行動変容) | {聴衆に何をしてほしいか — 動詞で記述} |
| 副目的 | {認知向上 / 関係構築 / 情報共有} |
| 成功基準(定量) | {例: 承認を得る / 予算確保 / 次回会議設定} |
| 失敗基準 | {例: 質問ゼロ / 否定的フィードバック} |

## 4. 制約条件
| 項目 | 内容 |
|------|------|
| コンテンツ制約 | {NDA対象, 未公開データ, 公開範囲} |
| 技術制約 | {投影環境, 解像度, ネットワーク有無} |
| アクセシビリティ | {色覚配慮, フォントサイズ最小値} |

## 5. リサーチ結果参照
{research-notes.md が存在する場合、主要発見を引用・参照}

Step 2: Deepening Audience Analysis

Analyze the following for each key person:

Analysis Item Content
Name/Title Specific name (not "CFO" but "CFO Taro Tanaka")
Areas of Interest What this person cares about most
Concerns Anxieties they may have about this proposal
Messages That Resonate The most effective expressions to prompt action
Knowledge Level Respective levels of technical/business knowledge

Step 3: Storyline Design (storyline.md generation)

Use the following structure rather than the ki-sho-ten-ketsu (introduction, development, twist, conclusion) structure:

Structure for AI for Science / Policy Recommendations

Background → Current State → Gap Analysis → Proposed Approach
→ Technical Architecture → Expected Impact → Timeline → Ask

Structure for Management / Investment Decisions

Why Now → Problem → Solution → Evidence → Microsoft Positioning
→ Competitive Advantage → Roadmap → Call to Action

Storyline Template: storyline.md

# ストーリーライン: {タイトル}

## 全体構造: {構造タイプ名}

| # | スライドタイトル | メインメッセージ | サポートデータ | ビジュアル指示 | トランジション | 想定時間(秒) |
|---|----------------|-----------------|---------------|--------------|---------------|-------------|
| 1 | {title} | {1メッセージのみ} | {データ・出典} | {図解/グラフ/画像の種類} | {前後の接続} | {N秒} |

## 時間配分サマリー
| セクション | スライド数 | 合計時間 | 持ち時間比 |
|-----------|-----------|---------|-----------|
| 導入 | {N} | {M秒} | {X%} |
| 本論 | {N} | {M秒} | {X%} |
| 結論 | {N} | {M秒} | {X%} |
| Q&Aバッファ | — | {M秒} | {5-10%} |
| **合計** | **{N}** | **{M秒}** | **100%** |

Step 4: Slide Outline Generation (slide-outline.md generation)

storyline.md Based on this, define the detailed structure of each slide:

# スライドアウトライン: {タイトル}

## スライド {N}: {タイトル}
- **メインメッセージ**: {1つだけ}
- **ビジュアルタイプ**: {図解 / データグラフ / 画像 / テキスト主体}
- **コンテンツ要素**:
  - {箇条書き / データテーブル / 引用 / コード例}
- **データソース**: {出典 URL or research-notes.md 参照}
- **ノート**: {発表者ノートの要点}
- **トランジション**: {次スライドへの接続文}

Step 5: Verification of Time Allocation

Constraint Rules
Maximum time per slide Consider splitting if exceeding 120 seconds (2 minutes)
Minimum time per slide Consider merging if less than 30 seconds
Q&A Buffer Secure 5–10% of the allotted time
Total time Keep within ±5% of the allotted time
Title Slide 15–30 seconds
Thank You Slide Include but do not add to the time allocation

Typical Slide Structure Patterns

AI for Science (MEXT / Academic): 15–20 min

  1. Title / Self-introduction
  2. Background — Why AI for Science Matters Now
  3. Current State Analysis — Research Trends in Japan and Abroad
  4. Gap Analysis — Challenges and Unsolved Problems
  5. Proposed Approach — Technical Overview
  6. Technical Architecture — System Configuration
  7. Preliminary Results / PoC Data
  8. Expected Impact
  9. Timeline / Roadmap
  10. Ask — Request for Cooperation / Budget Request

Microsoft HQ / For Executives: 30 min

  1. Executive Summary
  2. Why Now — Changes in the Market Environment
  3. Problem — Current Challenges
  4. Solution — Overall Picture of the Proposal
  5. Microsoft Positioning — Competitive Advantage
  6. Technical Architecture (Simplified Version)
  7. Evidence — Track Record and Data
  8. Customer Impact — Adoption Cases
  9. Competitive Landscape
  10. Financial Projections / ROI
  11. Roadmap — Execution Plan
  12. Risk & Mitigation
  13. Call to Action

Validation Loop

Design Validation Check

1. requirements.md を走査:
   - 聴衆が具体名で記載されているか → 抽象的なら修正要求
   - 主目的が動詞で記述されているか → 名詞なら動詞に変換
   - 成功基準が定量的か → 定性的なら具体数値を追加

2. storyline.md を走査:
   - 各スライドのメインメッセージが 1 つか → 複数あれば分割
   - 時間合計が持ち時間の ±5% 以内か → 超過/不足なら調整
   - Q&A バッファが 5-10% 確保されているか → 不足なら追加

3. slide-outline.md を走査:
   - データソースが全スライドに付与されているか
   - ビジュアルタイプが指定されているか
   - トランジションが論理的に繋がっているか

4. 全チェック通過 → ⏸️ ユーザー承認 → Phase 2(コンテンツ作成)へ

Failure Recovery

Failure Pattern Cause Recovery Procedure
Insufficient audience information The user does not know the details Present 3 assumed persona patterns and have the user choose
Time allotment undecided Meeting agenda not finalized Design for 3 patterns: 15 min / 30 min / 60 min
Purpose is ambiguous Only "information sharing" Ask "What do you want to happen as a result of information sharing?"
Too many slides Cramming in content Consolidate and reduce based on the one-message-per-slide principle
Structure type mismatch Choosing a structure that does not fit the audience Re-run audience analysis and reselect the structure

Quality Gates

  • requirements.md All items are filled in (no TBD)
  • The key audience members are listed with specific names and titles
  • The primary objective is described as behavior change (verb)
  • Success criteria are set quantitatively
  • storyline.md Each slide has only one main message
  • The total time allocation is within ±5% of the allotted time
  • A Q&A buffer of 5–10% of the allotted time is secured
  • The storyline structure fits the audience type
  • slide-outline.md Data sources are specified on all slides
  • If research-notes.md exists, key findings are reflected

Gotchas

1. The persona over-optimization trap — excessive adaptation to one specific person

If you optimize too much for one decision-maker (e.g., the budget approver), you leave the other participants behind. Set at most 3 key peopleand cover the interests of each person in a balanced way. Especially for MEXT, be mindful of both "policymakers" and "researchers".

2. Optimistic time estimates — insufficient Q&A buffer

Presenters overestimate their own speed.Always secure 5–10% of the allotted time as a Q&A buffer. For a 30-minute presentation, secure 2–3 minutes of Q&A time. Because Q&A tends to run long in Japanese meetings, 10% is recommended.

3. No abstracting requirements — do not use "executives"

Abstract descriptions such as "for senior management" or "for managers" significantly degrade design quality. Specific names, specific job titles, and specific interestsmust be used in descriptions. If information is unavailable, set tentative names and tentative job titles as an "assumed persona".

4. Over-detailing the storyline — the Phase 1 boundary

In Phase 1structure and flow. Specific wording, expressions, and visual details within slides are the domain of Phase 3 (content creation). If you start writing sentences here, rework will occur.

5. Avoid using the ki-sho-ten-ketsu structure

The "ki-sho-ten-ketsu" structure common in Japanese presentations is unsuitable for executives and policymakers. a structure that puts the conclusion (Ask) first. However, Background-first may be appropriate for academic presentations — judge based on audience analysis.

Memory Persistence

Patterns learned from past designs:

  • For MEXT: A "consistency with national strategy" section is mandatory. For budget requests, linking to the 6th Science, Technology, and Innovation Basic Plan is effective
  • For Microsoft HQ: Convey the conclusion in the first 3 slides. Put details in the Appendix
  • Executives: If presenting ROI, also stating "the cost of not investing" increases persuasiveness
  • Time allocation: In practice, it takes 1.2–1.5 times the plan. It is safe to target 80% at the design stage
  • AI for Science: The two Whys — "why AI" and "why now" — are mandatory in the introduction

Prohibitions

  1. Multiple messages per slide prohibited — Each slide must have exactly one main message
  2. Abstract audience descriptions prohibited — Use specific names and job titles, not "executives" or "stakeholders"
  3. Design without time allocation prohibited — Assign an estimated time (seconds) to every slide
  4. Omitting the Q&A buffer prohibited — Secure at least 5% of the allotted time for Q&A
  5. Unconditional adoption of the ki-sho-ten-ketsu structure prohibited — Choose the structure based on audience analysis
  6. Encroaching on the Phase 3 domain prohibited — Do not develop specific wording or expressions
  7. Ignoring research-notes.md prohibited — If it exists, always refer to and reflect it
  8. Prohibition of referencing copyright-infringing content — Check the rights status of the source