# IOTA Web Guardian — Pitch Deck Source
**Top 5 Winner — MasterZ × IOTA Hackathon 2026 (66 projects, 200 European developers)**

> Source of truth for the IOTA Berlin pitch. Every number below is cited and verifiable.
> Structure is slide-ready: each `##` section = one slide. Each bullet = one line/box.
>
> **Revision 7 (2026-04-21):** Claude Design handoff after tokens exhausted.
> Audit of the handoff bundle revealed: (a) 3 of 4 last-batch fixes applied
> (slide 6 bar chart, slide 7 line break, slide 14 "unlock") — MISSING the
> slide 9 body "under 3 seconds" → "~2.8 seconds" fix, applied manually to the
> main HTML. (b) The bundle's `*-print.html` version was severely stale
> (pre-Rev-3, missing slide 4 and almost all subsequent revisions) — discarded.
> Generated the final 16-page PDF directly from the main HTML via Chrome
> headless (main HTML has `@media print` CSS in deck-stage.js that handles
> per-slide pagination natively). Final deliverables on Desktop in
> `IOTA-Web-Guardian-Deck/`: `IOTA Web Guardian Pitch.html` (interactive,
> all 16 slides, arrow-key navigation), `deck-stage.js` (required sibling),
> `IOTA-Web-Guardian-Pitch.pdf` (16 pages, 1920×1080, 3.4 MB, shareable).
>
> **Revision 6 (2026-04-21):** Slide 6 bar-chart fix — the two ZERO-CLICK SHARE
> bars were hanging from the top (same top Y, different bottom Y) instead of
> standing on the x-axis. Added `justify-content:flex-end` to the inner columns
> so both bar bases land on the same Y coordinate (proper bar-chart semantics).
>
> **Revision 5 (2026-04-21):** Post-Rev-4 render audit — 3 micro polish items:
> (A) Slide 7 pull-quote line-break between the two sentences, (B) Slide 9 "under
> 3 seconds" aligned to "~2.8 seconds" for internal consistency, (C) Slide 14
> "the surface we monetize" softened to "the surface we unlock" (more accurate).
>
> **Revision 4 (2026-04-21):** Review of the rendered Claude Design deck surfaced
> five credibility/factuality issues and five polish items. Corrections below.
>
> **Revision 3 (2026-04-21):** Slide 11 (Why IOTA) — replaced "Move VM for custom
> logic" row (technically imprecise: Ethereum L2 has EVM for custom logic) with
> "Unified identity + payments on one chain" — a stronger, more accurate differentiator.
>
> **Revision 2 (2026-04-21):** added slide 4 ("What It Means in Dollars") with concrete
> per-publisher revenue scenarios; expanded Stereogum reference with founder context;
> replaced single-number fetch timing with the full JWT fast-path breakdown (first
> fetch / cached fetch / average over 100). Slides 5–16 renumbered accordingly.

---

## SLIDE 1 — The One-Line Pitch

**IOTA Web Guardian turns AI scraping into AI licensing.**

*(Use the brand name, not "We" — the cover slide faces a cold audience.)*

The open web is being harvested for training data. Creators lose the traffic, the revenue, and the leverage.
IOTA Web Guardian puts a verifiable identity + micropayment layer in front of every page — so AI agents pay per fetch, and creators get paid in real time on IOTA.

**Tagline options:**
- *"Pay-per-fetch, not pay-per-lawsuit."*
- *"The open web isn't free anymore. It's just uncompensated."*
- *"401 → VP → 402 → Paid. ~50 ms after the first handshake."*

---

## SLIDE 2 — The Hook

> **"Google scrapes 18 pages for every 1 visitor it sends back. ChatGPT scrapes 1,500. Claude scrapes 60,000."**
> — Cloudflare Radar, 2025

The economic model of the open web — "you give us content, we send you readers" — is **dead**.
AI doesn't send readers. AI keeps them.

---

## SLIDE 3 — The Problem: Traffic Apocalypse

Publishers are hemorrhaging traffic. This is not a forecast — **it already happened in 2025**.

| Metric | 2024 → 2025 |
|---|---|
| Global publisher traffic from Google | **−33%** (−38% in US) |
| Monthly visits to news sites (worldwide) | **2.3B → 1.7B** (−600M in one year) |
| Small publishers (1k–10k daily views) | **−60%** over 2 years |
| Medium publishers (10k–100k daily) | **−47%** |
| Large publishers (100k+ daily) | **−22%** |
| Business Insider organic search | **−55%**, 21% of staff laid off |
| HuffPost search referrals | **−50%** |
| **Stereogum** (indie music magazine, 20+ years of history) ad revenue | **−70%** in 2025 alone — founder Scott Lapatine named AI Overviews as primary cause |
| Dictionary.com visits | 34.1M → 21.3M |
| Chegg non-subscriber traffic | **−49%** YoY |

**Bottom line:** publishers have lost between **20% and 90%** of traffic and revenue in the past 12 months. Smaller outlets are shutting down.

*Sources: Press Gazette, Digiday, AdExchanger, ALM Corp, Chartbeat, Similarweb.*

---

## SLIDE 4 — What It Means in Dollars

Percentages are abstract. **Here's what the traffic collapse looks like on a single publisher's P&L.** Traffic loss maps almost 1:1 to revenue loss — confirmed by Charleston Crafted (−70% traffic → −65% ad revenue) and Stereogum (−70% revenue).

### Solo Blogger / Niche Blog
*(tech, cooking, travel, gaming — one author)*

| | Before (2023) | Today (2025) |
|---|---|---|
| Monthly views | **150,000** | **60,000** |
| RPM | $10 | $10 |
| Monthly revenue | $1,500 | $600 |
| **Annual revenue** | **$18,000** | **$7,200** |

→ **Losing $10,800/year** (−60%). *For many, that's half an income. From "rent covered" to "not worth it."*

### Medium Publisher
*(regional outlet, vertical magazine, small team of 3–10)*

| | Before | Today |
|---|---|---|
| Monthly views | **1,500,000** | **795,000** |
| RPM | $15 | $15 |
| Monthly revenue | $22,500 | $11,925 |
| **Annual revenue** | **$270,000** | **$143,100** |

→ **Losing ~$127,000/year** (−47%). *Two to three full-time editors, gone.*

### Large Publisher
*(100M-view tier — Business Insider / HuffPost scale, **display ad revenue only**)*

| | Before | Today |
|---|---|---|
| Monthly views | **100M** | **78M** |
| RPM | $20 | $20 |
| Monthly display revenue | $2M | $1.56M |
| **Annual display revenue** | **$24M** | **$18.7M** |

→ **Losing ~$5.3M/year** (−22%). *Business Insider laid off 21% of its staff in May 2025.*

*(Note: total publisher revenue includes subscriptions, licensing, commerce. This row isolates the display-ad impact to keep the math clean and defensible.)*

> **Traffic doesn't vanish. It's absorbed by AI. Revenue moves with it.**

*RPM benchmarks: Aditude + Publift 2025. Tier loss data: Chartbeat via Axios, March 2026.*

---

## SLIDE 5 — The Problem: The Crawl-to-Click Gap

The *real* story isn't that AI summaries replace clicks. It's that AI bots take **everything** and return **nothing**.

**Pages scraped per 1 visitor sent back** *(consolidate to one row per company — Claude, Anthropic, ChatGPT, OpenAI were duplicated in Rev 3 render):*

```
Google        ▇▇                             18 : 1
OpenAI        ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇             1,700 : 1   (June 2025)
Anthropic     ████████████████████████████   73,000 : 1   (July 2025, peak 286,000 : 1 in Jan 2025)
```

**This is not search. This is extraction.**

- AI crawlers now drive **8.7%** of all HTML traffic on the web (Cloudflare, 2025)
- AI bot scraping grew **+225% YoY** (Akamai)
- **80%** of AI bot activity is for **LLM training** — not user queries, not referrals
- **30%** of AI bot requests **ignore robots.txt** (Cloudflare Q4 2025)
- AI surfaces send an average of **0.12%** of publishers' referral traffic

*robots.txt is a polite request. It is being openly disregarded. There is no enforcement layer on the open web — until now.*

---

## SLIDE 6 — The Problem: Zero-Click Searches

Even when users do search, they don't click through anymore.

| Metric | 2024 | 2025 |
|---|---|---|
| Zero-click searches | 56% | **69%** |
| CTR when AI Overview present | — | **0.6%** (vs 1.6% without) |
| AI Overview click reduction | 34.5% (Apr) | **58%** (Dec) |
| Click rate on #1 ranked result below an AI Overview | baseline | **−79%** |
| Users clicking cited links in AI Overviews (Pew) | — | **1%** |

**Publisher survey:** news publishers expect search referrals to drop **another 43% by 2029**.

*Sources: Similarweb, Pew Research, Search Engine Land, Digiday, Press Gazette.*

---

## SLIDE 7 — Why This Matters (for AI too)

If creators stop publishing, the AI training pipeline dries up.

- AI quality depends on fresh, reliable, long-tail human content
- Without a compensation layer, we enter a **content death spiral**:
  *creators quit → fewer good sources → AI quality degrades → trust collapses*
- The web runs on an **unwritten contract**: you crawl, you send readers, everyone wins.
  That contract has been unilaterally broken.
- The only sustainable fix: **make the exchange explicit and verifiable.**

> **"If content dies, AI dies with it. The question isn't whether AI should pay — it's how."**

*(This is the strongest single line in the deck. Render it as a full-width callout at the bottom of Slide 7, not just body text.)*

---

## SLIDE 8 — Why Current "Solutions" Fail

| Approach | Fatal flaw |
|---|---|
| **robots.txt** | Voluntary, ignored by 30% of AI bots |
| **Cloudflare AI Labyrinth / Bot blocking** | Defensive only — blocks crawl but monetizes nothing |
| **Paywalls** | Break the open web for humans too, kill SEO |
| **Tollbit / Scalepost** | Centralized gatekeepers, per-publisher deals, US-only |
| **Lawsuits (NYT vs OpenAI, etc.)** | Years long, enterprise-scale only, bloggers left out |
| **Licensing deals (Reddit-Google, AP-OpenAI)** | Bespoke, opaque, exclusive to the top 0.01% of publishers |

**What's missing:** a neutral, permissionless, programmatic layer anyone can deploy — from the NYT to a solo blogger.

---

## SLIDE 9 — Our Solution: IOTA Web Guardian

**A programmable paywall that speaks agent.**

A 5-step pipeline — full first-fetch handshake in **~2.8 seconds**, amortized to **~77 ms** via a 1-hour JWT.

Every protected URL is guarded by a 5-step pipeline:

```
1.  Agent GET /article          → 401 + VP challenge (nonce)
2.  Agent presents VP           → CVS verifies DID chain + signature
3.  Server issues 402 + quote   → "Pay 1,000 nanos to addr X, nonce Y"
4.  Agent pays on IOTA L1       → returns tx digest
5.  x402 verifies on-chain      → 200 OK + content + Guardian JWT
```

**Performance (per-agent, per-site):**

| | Time | What runs |
|---|---|---|
| **First fetch** | **~2.8 s** | Full pipeline: DID resolution + VC verify + L1 payment + x402 settlement |
| **Cached fetch** (JWT fast path) | **~50 ms** | Signature check only — no chain calls |
| **Average over 100 fetches** | **~77 ms** | Realistic scraper workload, JWT TTL = 1h |

> *"First fetch registers the agent. Every subsequent fetch in the next hour is a signature check. Like a toll booth with an electronic pass: register once, beep-beep forever after."*

- **For creators:** drop-in WordPress plugin. Monetize every AI agent visit. Real-time dashboard.
- **For AI agents:** standard HTTP 401/402 flow. One-time identity setup. Pay-as-you-fetch.
- **For the ecosystem:** every payment is on IOTA L1 — **public, verifiable, feeless, instant.**

---

## SLIDE 10 — How It Works (Architecture)

**Three decoupled services, one open standard:**

- **CVS (Credential Verification Service)** — resolves DIDs, validates VC chains, checks challenge nonces. Rust + IOTA Identity SDK.
- **x402 Service** — verifies on-chain IOTA L1 payment proofs. Returns settlement tokens.
- **Guardian Plugin** — the enforcement point. Ships as a WordPress plugin today; the middleware logic is portable to any HTTP stack.

**Identity layer:** W3C Decentralized Identifiers + Verifiable Credentials on **IOTA Rebased (Move VM)**.
**Payment layer:** native IOTA L1 transfers (sub-second finality, feeless, Base58 tx digests).
**Standard:** HTTP **402 Payment Required** + x402 scheme (the emerging web standard for agentic commerce, backed by Anthropic / Base / Stripe).

---

## SLIDE 11 — Why IOTA (and not anything else)

| Requirement | IOTA | Ethereum L2 | Bitcoin LN |
|---|---|---|---|
| Sub-second finality | ✅ | ⚠ ~2s | ⚠ routing dependent |
| **Feeless** micropayments | ✅ | ❌ (gas > transfer value) | ⚠ channel setup cost |
| Native DID / VC support | ✅ (IOTA Identity) | ⚠ (off-chain add-on) | ❌ |
| On-chain auditable for regulators | ✅ | ✅ | ⚠ |
| **Unified identity + payments on one chain** | ✅ | ❌ (DID off-chain + L2 payment, 2 stacks) | ❌ |

**No other L1 can deliver all five. To build Guardian anywhere else, you'd have to stitch together 2-3 different chains.**
**IOTA Web Guardian is IOTA-native by technical necessity, not by sponsorship.**

---

## SLIDE 12 — Live Demo: It's Real Money, On Real Chain

The demo is not a mock. Every fetch produces an actual IOTA testnet transaction.

- WordPress admin dashboard shows:
  - Live pipeline flow (401 / VP / 402 / tx / 200) via Server-Sent Events
  - Chart.js graphs of revenue in nanos, tx volume, CVS/x402 call latency
  - Transactions table with **clickable digests → IOTA Explorer**
- Explorer URL format:
  `https://explorer.iota.org/txblock/<base58-digest>?network=testnet`

**End-to-end timing** (see Slide 9 for full breakdown):

| First fetch | Cached fetch | Avg over 100 |
|---|---|---|
| **~2.8 s** | **~50 ms** | **~77 ms** |

- Cost per fetch: **1,000 nanos** (configurable per post type)

> *"The whole flow is provable. Anyone can click the tx link and see their payment on the IOTA Explorer. No trust required."*

---

## SLIDE 13 — Competitive Differentiation

| | **Cloudflare AI Labyrinth** | **Tollbit** | **NYT-style licensing** | **IOTA Web Guardian** |
|---|---|---|---|---|
| Monetizes AI traffic | ❌ blocks only | ✅ | ✅ | ✅ |
| Permissionless (anyone can deploy) | ❌ | ❌ | ❌ | ✅ |
| On-chain auditable | ❌ | ❌ | ❌ | ✅ |
| Per-request micropayments | ❌ | ⚠ batched | ❌ bulk deals | ✅ **native** |
| Works for a solo blogger | ❌ | ❌ | ❌ | ✅ WordPress plugin |
| Verifiable agent identity (W3C DID/VC) | ❌ | ❌ | ❌ | ✅ |
| Open standard (HTTP 402 / x402) | ❌ | proprietary | N/A | ✅ |

**We're not competing with Cloudflare. We're building the economic layer Cloudflare can't.**

---

## SLIDE 14 — Market Sizing (the "SmartLegacy number")

Using the same framing SmartLegacy used to anchor their pitch ($140B+ lost wallets, 87% of wealth managers unprepared):

- **$140B+** — global digital display-ad **market size** (IAB, 2024 Internet Advertising Revenue Report). *This is the TAM, not a loss figure — the slide render in Rev 3 said "diverted" which is not a defensible claim.*
- **600M+** — monthly publisher visits lost in the last 12 months *(AdExchanger, mid-2024 → May 2025)*
- **−43%** — expected additional drop in news search referrals by 2029 *(publisher survey, Search Engine Land)*
- **8.7%** — share of all web traffic now coming from AI crawlers *(Cloudflare Radar 2025)* — **the surface we unlock** *(not "monetize" — we unlock per-site monetization on this traffic, we don't capture the whole slice)*
- **$0** — what most bloggers currently earn from that crawl volume

**The addressable gap is every page on the web that's currently scraped for free.**

---

## SLIDE 15 — Validation: Top 5 at MasterZ × IOTA Hackathon

Out of **66 projects** and **200 selected European developers**, IOTA Web Guardian placed in the **Top 5 finalists** alongside Textile Tracer, Nplex, DocuNotary, and Tangle Gate.

*(Render the finalists as unordered bullets, NOT numbered 01–05 — the Top 5 are unranked peers, not a leaderboard. Highlight "IOTA Web Guardian" in mint teal / bold so the viewer's eye catches it in the list.)*

> *"IOTA Web Guardian establishes authentication requirements where AI systems must authenticate and pay to access content, enabling fair creator compensation."*
> — IOTA Foundation, MasterZ Hackathon Finale announcement

**Outcome:** invited to the private IOTA ecosystem event in **Berlin (May 2026)** — next step.

---

## SLIDE 16 — Roadmap / Asks

**Shipped (hackathon deliverable):**
- Full 5-step pipeline (Rust services + WP plugin)
- IOTA testnet integration (DIDs + VCs + L1 payments)
- Interactive wizard TUI (9-step self-service setup)
- Agent Tester web UI (replaces CLI for non-devs)
- WordPress admin dashboard with live charts + explorer links

**Next 90 days:**
- Mainnet launch (requires IOTA Identity package deployment)
- Drupal, Ghost, and generic reverse-proxy adapters
- Agent SDK (TypeScript + Python) so AI developers can integrate in one line
- Pilot with 3–5 real publishers

**What we want from IOTA Berlin:**
- Ecosystem endorsement + co-marketing with IOTA Identity team
- Introduction to Web3 publisher networks (Mirror, Paragraph, Farcaster Frames)
- Alignment with emerging x402 standard working groups
- Mainnet launch support + faucet sponsorship for the pilot phase

---

## APPENDIX — Raw Numbers with Sources

### Traffic loss
- Global publisher Google traffic: **−33%** Nov 2024 → Nov 2025 (**−38%** US) — *Press Gazette, Trends Report 2026*
- News site monthly visits: **2.3B → 1.7B** mid-2024 → May 2025 — *AdExchanger*
- Small publishers (1k–10k/day): **−60%** — *Chartbeat via ALM Corp*
- Business Insider: **−55%** organic search Apr 2022 → Apr 2025; 21% staff laid off May 2025 — *Similarweb*
- **Stereogum**: US indie music magazine founded 2002 by Scott Lapatine. One of the last surviving music blogs of the MP3-blog era. Lapatine publicly stated (November 2025) that ad revenue dropped **−70%** in 2025, naming Google AI Overviews as the primary cause.
- **Charleston Crafted**: home-improvement blog. **−70%** traffic March→May 2024 translated into **−65%** ad revenue — confirms ~1:1 traffic-to-revenue ratio
- Chegg: **−49%** non-subscriber traffic Jan 2024 → Jan 2025
- Dictionary.com: 34.1M → 21.3M visits

### Revenue per publisher tier (RPM benchmarks, 2025)
Sources: Aditude, Publift, SmartyAds CPM data.
- RPM typical range: **$5 – $25** for Tier 1 audiences (US/UK/CA/AU)
- Niche blog RPM: **~$10**
- Medium vertical publisher RPM: **~$15**
- Large news RPM: **~$20**
- Tier 1 CPM (display ads): **$8 – $25**
- Blog earnings by scale (Productive Blogging 2026 survey):
  - 10k–50k monthly visitors: **$1k–$5k/month**
  - 50k–100k+ monthly visitors: **$5k–$20k+/month**

### Crawl-to-click gap
- Google: **18 : 1** — Cloudflare crawl-to-refer data
- ChatGPT: **1,500 : 1**
- Claude: **60,000 : 1** per visitor
- OpenAI crawl-to-referral ratio: **1,700 : 1** (June 2025)
- Anthropic: **73,000 : 1** (July), peak **286,000 : 1** (January 2025)
- AI bot scraping growth: **+225% YoY** — Akamai
- 30% of AI bots ignore robots.txt in Q4 2025 — Cloudflare
- Training = 80% of AI bot activity (up from 72%) — Cloudflare Radar 2025

### Zero-click
- Zero-click share: 56% (May 2024) → **69%** (May 2025) — Similarweb
- CTR with AI Overview: **0.6%** vs 1.6% — Search Engine Journal
- Click reduction: 34.5% (Apr 2025) → **58%** (Dec 2025) — PPC Land
- #1 result under AI Overview: **−79%** clicks
- Pew: **1%** of users click AI Overview citations

### AI crawler share
- 8.7% of all HTML requests on the web (2025) — Cloudflare Radar
- GPTBot: 4.7% → 11.7% share of AI crawling traffic
- ClaudeBot: 6% → ~10%

### Guardian performance (measured on IOTA testnet, estimates pending p50/p95 measurement)
- First fetch (full pipeline): **~2.8 s**
  - breakdown: 401 issue (~10 ms) + VP build/send (agent-side) + CVS verify ~500 ms + 402 issue + IOTA L1 payment ~1 s + x402 verify ~500 ms
- Cached fetch (JWT fast path, 1h TTL): **~50 ms** — local signature check only
- Average over a 100-fetch scrape (one agent): **~77 ms** — first fetch amortized across the cache window

### Hackathon
- MasterZ × IOTA: 66 projects, 200 European devs
- Top 5: Textile Tracer, Nplex, **IOTA Web Guardian**, DocuNotary, Tangle Gate
- Community Award: SmartLegacy Passkey

**Source URLs (direct):**
- https://pressgazette.co.uk/media-audience-and-business-data/google-traffic-down-2025-trends-report-2026/
- https://www.adexchanger.com/publishers/the-ai-search-reckoning-is-dismantling-open-web-traffic-and-publishers-may-never-recover/
- https://thedigitalbloom.com/learn/2025-organic-traffic-crisis-analysis-report/
- https://digiday.com/media/google-ai-overviews-linked-to-25-drop-in-publisher-referral-traffic-new-data-shows/
- https://almcorp.com/blog/search-traffic-decline-small-publishers-chartbeat-data/
- https://ppc.land/googles-ai-summaries-now-swallow-58-of-clicks-that-once-went-to-websites/
- https://searchengineland.com/news-publishers-search-referrals-drop-report-467408
- https://blog.cloudflare.com/crawlers-click-ai-bots-training/
- https://radar.cloudflare.com/year-in-review/2025
- https://originality.ai/blog/traffic-apocalypse-ai-content-economy
- https://www.aditude.com/blog/rpm-for-publishers
- https://www.productiveblogging.com/how-much-do-bloggers-earn/
- https://blog.iota.org/masterz-hackathon-finale/

---

## VISUAL SUGGESTIONS FOR CLAUDE DESIGN

**Key visuals to generate:**

1. **Crawl-to-click bar chart** (slide 5) — horizontal bars showing Google 18, ChatGPT 1,500, Claude 60,000 (log scale makes the point viscerally)
2. **Publisher revenue timeline** (slide 3) — line graph 2022 → 2025 showing Business Insider / HuffPost / Stereogum decline overlaid
3. **Three persona cards** (slide 4) — vertical cards with Before/After numbers for solo blogger / medium publisher / large publisher; hero-size loss number (120+pt) in amber/red
4. **Pipeline diagram** (slide 9/10) — left-to-right flow: Agent → 401 → VP → CVS → 402 → IOTA L1 → 200 (with 3-row timing panel: first 2.8s / cached 50ms / avg 77ms)
5. **Competitive matrix** (slide 13) — the table, but as a styled 5×4 grid with ✅/❌ icons
6. **"The gap" visual** (slide 5) — a funnel showing AI crawlers pouring into a site and almost no visitors coming out
7. **Live dashboard screenshot** (slide 12) — real screenshot from `/wp-admin/` Guardian dashboard showing charts + tx table
8. **IOTA Explorer screenshot** (slide 12) — an actual testnet transaction the team can click through live

**Color palette:** IOTA navy + the project's mint teal accent (`#5eead4`) for anything "our solution." Publisher-loss numbers in amber/red.

**Tone:** grave for the problem slides (slides 2–7), solution-confident for slides 8+. Don't soft-pedal the numbers — they're the whole pitch.

---

*File generated 2026-04-21 for the IOTA Berlin preparation call (Friday).*
*Revision 6 (2026-04-21): slide 6 bar-chart base alignment — justify-content:flex-end on inner columns.*
*Revision 5 (2026-04-21): post-Rev-4 render polish — slide 7 pull-quote line-break, slide 9 "under 3s" → "~2.8s", slide 14 "monetize" → "unlock".*
*Revision 4 (2026-04-21): post-render review — slide 5/6 section-number swap, crawl-chart consolidation, $140B reframe, Top-5 unrankng, Large-Publisher display-ad clarification, cover-slide "We" → brand name, slide 7 money-quote as callout, architecture flow relabel, soften "⭐ THE DECIDING REQUIREMENT", add speaker notes.*
*Revision 3 (2026-04-21): Slide 11 Move VM row replaced with Unified identity + payments differentiator.*
*Revision 2 (2026-04-21): added slide 4 (dollars), Stereogum context, JWT fast-path timing.*
*Feed this file to Claude Design to generate the slide deck.*