Company Deck - #22
Open
grandchildrice wants to merge 81 commits into
Open
Conversation
- 14枚構成の会社説明スライドを作成 - SL01: 表紙 - SL02: 組織概要 - SL03: ミッションとビジョン（3つの柱） - SL04: 5つの活動領域 - SL05: 研究実績サマリー（数値ハイライト） - SL06: セキュリティ監査AIエージェント（世界1位） - SL07: Lean形式検証自動化（世界初） - SL08: DEX不公平性分析（FC2026採択） - SL09: 定理証明AIエージェント（SOTA） - SL10: 学会採択・論文一覧 - SL11: Ethereumエコシステムへの貢献 - SL12: サービス提供（AI監査・共同研究） - SL13: 支援者・パートナー - SL14: お問い合わせ Sources: - ascon-proposalブランチの既存スライド（実績情報・画像） - nyx.foundation ホームページ（活動領域・スポンサー・ニュース） - PR Times プレスリリース（各研究成果の詳細）
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
presentation-agent | 1e0b3cd | Commit Preview URL Branch Preview URL |
Jun 09 2026, 06:20 AM |
修正内容:
- SL13: 支援者・パートナーをアイコン+バッジ形式で視覚的に改善
- プレミアムスポンサー/スポンサー/サポーター/学術パートナーを
コンパクトに1スライドに収まるようレイアウト変更
- SL12b: Agentic Hacking Cup（世界初のAI攻防ハッキング大会）
- SL12c: ASCON（AI経済コンペティション）
- SL12d: 暗号×ブロックチェーン Study Group（東大共催）
- SL12e: 研究所「Nanto」建設構想（富山県南砺市）
ソース:
- agentic-hacking-cup ブランチ
- ascon-proposal ブランチ
- presentation/zktokyo-coreprogram2026-sponsors ブランチ
- PR Times プレスリリース（Nanto構想）
- SL12e: Nanto研究所スライドを更新 - 新しい完成イメージ画像（木造建築）をメインに配置 - 既存のhouse_nanto画像は「まず居住・シェアハウス用住宅を購入」の説明として併記 - 段階的アプローチの説明を追加 - SL13: 支援者・パートナーをnyx.foundationの実際のスポンサー画像で更新 - 全スポンサー/サポーターのプロフィール画像をダウンロード - Premium Sponsors / Sponsors / Supporters / Academic Partners の4段構成 - 丸型プロフィール画像で視覚的にインパクトを向上
- Sponsors: w-14 → w-16 に拡大し「キリフダ」等が改行せず収まるように - Sponsors: gap-4 → gap-5 に広げてガタつき解消 - Supporters: Katomasa を削除
- データトラスト・藤本真衣・松本ゆうほ・中城元臣: w-12→w-14, text-[8px]→text-[7px]で改行回避 - Premium Sponsorsの匿名アイコン下に「匿名」ラベルを追加し高さを揃える
- SL12b: ダッシュボードキャプションを削除し、提案資料リンクに差し替え - SL12c: 同上
- nyx.foundation/memberからメンバー情報と画像を取得 - 8名のメンバー（名前・役職・経歴）をグリッドレイアウトで配置 - 組織概要（SL02）の直後に挿入 - Speaker Notesも追加
SL03: 'Why Ethereum' - 6 member quotes with photos in 3x2 grid SL04: 'Build the center of invention here' - 2 missions + 3 Hamming principles SL05: Updated subtitle to connect philosophy with results
…earch institute vision SL03: Replace member quotes with 4 thematic cards showing what's actually happening in Ethereum (consensus, cryptography, DeFi, quantum resistance) with tags and factual descriptions. Open science framing. SL04: Title updated to 'Build center of invention, toward world's best lab'. Added Bell Labs / Nobel Prize aspiration. Fixed Banri quote visibility (white bg + black border instead of black bg).
…iples as 3-col row
Feedback: SL2-8 were small Japanese sentences lined up — childish, not intuitive. Replaced prose with hard, researched numbers as the hero. - SL05 four-storms: each tile now leads with a researched figure — $7.1B post-quantum migration cost, $2.2B crypto stolen in 2024, $2.4T annual loss from poor software quality, $1.8B MEV extracted. - SL06 storm ①: the post-quantum migration is shown to be already exercised on a real large network — a node-field dot visual with "10,000+ nodes / 1,000,000+ validators" (Ethereum mainnet scale). - SL06/SL07 storm cards: paragraphs → a big figure + one line + a tiny source (12-15yr migration; $2.2B; $2.4T & 100x bug-fix cost; $1.8B MEV). - SL02: prose blocks → a chip row (電気/水道/口座/支払い/連絡 → 結論). - SL03/SL04/SL08: paragraphs compressed to key phrases; SL08 is now a 4-storms → one-question → answer flow. Figures: post-quantum migration cost/timeline (US gov estimate), crypto theft 2024 (Chainalysis), software-quality cost 2022 (US), cumulative Ethereum MEV (Flashbots), node count (Etherscan). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- 嵐① (keys age): now leads with a historical comparison — the last algorithm migration (SHA-1→SHA-2) took ~10 years; post-quantum is far larger at 12-15+ years, shown as two comparison bars. - 嵐② (attacks): updated to the 2025 figure — $3.4B stolen, the worst year on record — and adds the AI angle (over half of 2025's blockchain exploits could have been run autonomously by AI). - 嵐③ (correctness): updated to the latest poor-software-quality cost, ~$3.1T/year (2024). - 嵐④ (money skew): replaced the MEV figure with the wealth-gap trajectory the brief asked for — top 1% income share 15%→23% (2000→2023) with a dashed "and rising" projection row. - Every figure now carries an explicit on-slide source line (NIST, Chainalysis, World Inequality Report, Oxfam, Etherscan, etc.). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The "私たちの仕事の作法" paragraph becomes three principle tiles, each with an icon and one line: - 先に解く — solve society's hard problems first, on Ethereum (sunrise). - 知識をぶつけ合う — live together, connect with the world, collide knowledge (linked nodes). - 公共のために — return results to the world as a public good (globe). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- New efficiency band: the same problem costs ×1 to fix at design stage vs ×100 once it has shipped into society (software-quality cost data). This is the fact-based answer to "compared to what" — solving before society notices is the highest results-and-speed per yen, and Nyx reached an audit world #1 on ~¥10M in 6 months. - The three engagement cards now list expectable merits as bullet points (donate / sponsor / collaborate) instead of timeline visuals. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Per feedback, the efficiency comparison is now research → PoC → social implementation, in time: - Normal path: ~17 years for research to reach real-world practice (Balas & Boren, 2000 — the canonical research-to-practice gap). - Nyx's path: ~0 years — research is verified from day one on a real production network (Ethereum), collapsing the PoC → deployment valley. That is the fact-based "compared to what": Nyx has the highest results and speed per yen toward social implementation. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- SL10 efficiency comparison now contrasts the normal research path (~17 years to reach practice) with Ethereum's social-implementation cadence — a new spec goes from proposal to live in as little as ~1 year (Ethereum ships network upgrades twice a year). - Removed the faint subtitle line under the title on SL02-SL09 — the action titles already carry the message; declutters every slide. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The design_critic persona now scores low on text too small or faint for an elderly viewer, and on layouts with no clear eye-flow / multiple competing focal points. Added a matching panel-wide rule (legible & eye-flow) so the loop drives accessibility and visual-hierarchy fixes. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Panel refine (design_critic now enforcing elderly legibility and eye-flow) ran 4 iterations, deck score 3.70 → 4.50, converged: - bumped sub-legible 7.5-9px source/caption/label text to ~9-11px and raised contrast (opacity up); - recoloured the SL06/SL07 comparison bars from near-black (which rendered as solid dark boxes) to light grey so the numbers lead; - SL09 reorganised — fewer competing blocks, larger metric cards. The loop also dropped the 仕事の作法 tiles from SL09 while decluttering. Since that section was explicitly requested, it is restored as its own dedicated slide (SL09b) — three large, elderly-readable principle tiles (先に解く / 知識をぶつけ合う / 公共のために). Deck is now 12 slides. Note: the loop changed SL09's "#1" caption from "イーサリアム財団 監査" to "Sherlock監査コンテスト", citing inputs/company-deck-design.md. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- SL08: enlarged the small text — the 昔/これから lines 11→14px with higher contrast, the storm chips 11→13px, the answer eyebrow 10→12px. - SL09: rebuilt for a clear top-down eye-flow — the zig-zag 2x2 data grid becomes one centered identity block, then a single straight path identity → 実績 metrics → partner logos. - SL09b: dropped the "本物の嵐…" text box; in its place a photo of the Nyx research house (house_nanto.png) under the three principles — illustrating "ともに住み、知識をぶつけ合う". Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- SL09b: show both research-house photos side by side (house_hongo + house_nanto) under the three principles. - SL10: title becomes the call to action — "検証可能な街を、共につくりましょう". - SL05 / SL06: storm ② adds the fact that attacks are growing more sophisticated — much of 2025's crypto loss targeted people via social engineering (impersonation, phishing), "code less, people more". - SL09: company-overview line (founding / base / research areas) kept. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- SL08: removed the small 昔は/これからは contrast lines — the slide is now just the four storms → the one question → the answer. The contrast stays in the speaker notes. - SL10: widened .eff-v and set white-space:nowrap so "≈17年" no longer wraps to two lines. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The slides' 3D read as flat, cheap, anime-ish. Upgrade the renderer: - soft shadows (PCFSoftShadowMap) with a shadow-casting key light whose shadow camera is fit to the scene bounds; meshes cast and receive. - a transparent ShadowMaterial ground plane so models are grounded with a contact shadow instead of floating. - ACES filmic tone mapping for richer, non-flat shading. - image-based lighting via RoomEnvironment + PMREM, so metal and glass parts get real reflections. - rebalanced lights (soft ambient + hemisphere + key + cool fill). This lifts all four embedded scenes at once. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Re-ran the recursive self-improvement loop (3 iterations each) so the scenes read as detailed models rather than a few abstract blocks: - two-cities 64 → 146 parts (94/100) - ethereum-city 53 → 81 parts (92/100) - quantum-keys 55 → 72 parts (93/100) - capital-flows 71 → 85 parts (93/100) Combined with the upgraded Scene3D rendering (shadows, tone mapping, image-based lighting), the embedded 3D no longer looks flat or cheap. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The primitive 3D of "post-quantum keys" and "DeFi capital flows" read as unrecognizable piles of blocks — no viewer, least of all a non-expert, could tell what they were. Concept 3D from primitives is the wrong tool for metaphors. Replaced with scenes anyone reads at a glance: - SL06 → besieged-house: a recognizable house, its lock trivially copied (spare keys scattered by the door), with thief figures closing in from every direction — keys aging + attacks that never sleep. - SL07 → fragile-bridge: a properly-built-looking bridge cracked and collapsing mid-span (looks right, fails), beside pedestals where coins pile sky-high on one and the rest sit near-empty — skewed money. quantum-keys.json and capital-flows.json removed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
…cities The JSON-primitive Scene3D approach could not render a city that read as a city — boxes on a plate, not a skyline. Replaced the two city slides with a hand-coded Three.js engine (public/city/engine.js): one seeded, deterministic city — street grid, diagonal river + bridges, ~55 varied buildings with procedural lit-window facades, rooftop kit, parks, street trees, streetlights, animated traffic, sky-gradient IBL and bloom. One generator, three skins of the same town: - base a clear-day stylized cityscape → SL02 (見える街) - invisible a Matrix-green wireframe silhouette city → SL02b (見えない街) - ethereum the same city, futuristic cyan/violet → SL04 (ethereum) The three are now full-bleed immersive slides (layout: full) with a live, draggable, slowly-orbiting city behind the copy. SL02b is a new slide so the "二つの街" beat reads as two immersive reveals — the visible city, then its invisible structure. two-cities.json / ethereum-city.json removed. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The seriph theme styles h1/p at higher specificity than the slide's own classes, so .city-h / .city-p lost their color and fell back to the theme's dark text — unreadable on the dark ethereum and invisible city backgrounds. Switched the headings/body to <div>s with !important colors and deepened the text-side veil so the copy reads cleanly on all three. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The besieged-house and fragile-bridge JSON scenes were primitive-block approximations. Replaced both with a hand-coded Three.js engine (public/storm/engine.js), matching the city scenes' quality: - house (SL06): a cosy house — pitched roof, lit windows, fence with an open gate — whose key keeps peeling off endless translucent copies, while hatted thieves creep in from every route. 鍵 / 攻撃. - bridge (SL07): a normal-looking, working suspension bridge — towers, cables, hangers — that never stops gently swaying. It looks fine; it is fragile (slim legs, hairline cracks on one tower). The wealth on the deck is piled almost entirely onto one side. 正しさ / 偏り. Both embedded as live panels via /storm/index.html?v=<variant>. The bridge caption now reads "動いて見えて、実はもろい橋" — it is not a broken bridge, it is one that works yet is unsound. The two orphaned JSON scenes are removed; nothing imports Scene3D anymore. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
New slide bridges the storm framing (SL05-08) and the org intro (SL09) by showing concrete research products against each storm — SPECA for storm 2, Verity for storms 1+3, DeFi Analytics for storm 4, with Eris as the foundation — plus achievement badges (Sherlock #1, 5+ accepted papers, EF grant, Cambridge EF Retreat invitation). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The "嵐①" tags read as clunky and the foundation/products split broke visual rhythm. Now all four products (Eris, SPECA, Verity, DeFi Analytics) sit in one equal-weight row with larger logos; the storm mapping comes through naturally via descriptions and surrounding deck flow rather than explicit numeric tags. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Swap in higher-quality official artwork sourced from each institution's canonical channel (JMOOC for Waseda, Wikimedia for the 2004–2024 U-Tokyo horizontal mark, U-Toronto exchange for HKUST). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
P10 now drops the four-up metric grid for a single bold callout
("監査コンテスト 世界一・国際学会 5件採択") — the detailed track
record lives on the next slide so this one stays uncluttered.
P11 swaps insider terms for plain Japanese: Sherlock → セキュリティ監査
コンテスト, DeFi Analytics → Market Lab, 形式検証 → 数学で「正しさ」を
証明, MEV → お金の流れの偏り, EF Retreat → イーサリアム財団の合宿.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds three industry & community partners alongside EF and the universities, switching from a 5-wide single row to a 4×2 grid. Hero callout above tightened (smaller padding, 32px headline) so the expanded partner block fits without crowding. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bottom row was being clipped off the slide. Shrunk the hero callout (28px headline, tighter padding/margins), trimmed logo-cell from 56px to 44px, and tightened the section header margin. Also swapped to smbc_nikko2.png (cleaner crop without the tagline) and applied a 1.6× CSS transform so the SMBC mark fills the cell instead of vanishing inside its built-in whitespace. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The new artwork is a tight horizontal lockup with no built-in whitespace, so the 1.6× transform hack is no longer needed. Drops back to smbc_nikko.png (overwriting it) and removes the per-image scale override. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Sourced from the Google Photos share (resolved through the og:image to a lh3.googleusercontent CDN URL). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
p9 (SL08): replace the 嵐①②③④ pill grid + arrow + bordered question box with an editorial layout — the four storm concepts (鍵 · 攻撃 · 正しさ · お金) in a serif row with thin separators, a faded vertical line descending into the quoted question, then a thin divider before the answer headline. No more UI-button pills. p11 (SL09c): rewrite the project blurbs to use the deck's "街" metaphor (政策のシミュレーション基盤 / 街の脆弱性を見つけるAI / 正しさを証明しながら街を作る技術 / 街のお金の偏りを測る). Rename Market Lab → 経済分析 and drop its wordmark for a single larger chart icon. Switch the achievement strip from horizontal fig|text to vertical fig-over-caption so the Japanese labels stop wrapping mid-word. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Sources: