Requests
12,840
Rhone OS · a Cloudflare OS concept
What are we working on?
Turn company context into governed apps, briefs, workflows and plans.
Explore role workspaces
Peak Security & Bot Command CenterSr Director, Cybersecurity & Tech Operations · Bot visibility, checkout protection and edge usage with human-gated changes.
AppAI Channel & AEO Readiness BriefSr Director, Digital Product · Product data, agent discoverability and omnichannel opportunity in one brief.
DocumentStorefront Release ReviewManager, Web Development · Deterministic PR and preview checks before a human-approved release.
Workflow2027 Digital & Platform PlanningCross-functional: Digital Product, Cybersecurity & Tech Ops, Web Development · A cross-functional planning story across the funnel, AI channels and resilience.
Slide DeckPublic prototype. Organization-wide pages are example configurations; no real Rhone data or account access.
Build a peak-season security and operations view for our storefront. Separate useful AI agents from abusive automation, watch sensitive endpoints and show edge capacity without giving the app production write access.
AI
I start with no access. To assemble an example view, I would ask for scoped, read-only analytics; credentials would remain inside Gatekeepers.
Gatekeeper: Cloudflare Security + Edge Analytics
Illustrative read-only bot, WAF, request, cache and Worker usage summaries. Proposed changes require a separate approval capability.
Illustrative read-only bot, WAF, request, cache and Worker usage summaries. Proposed changes require a separate approval capability.
Show the operational decisions and approval points as a sample; do not execute changes.
AI
The app blueprint groups simulated signals by endpoint and verification status. Deterministic queries compute trend and usage summaries; a security owner reviews any proposed rule before it can run.
Illustrative code
const [bots, edge, scripts] = await Promise.all([
env.SECURITY.getBotTrends({ scope: "storefront", access: "read" }),
env.EDGE_USAGE.getMonthlySummary({ access: "read" }),
env.SCRIPT_INVENTORY.getChanges({ path: "/checkout" })
]);
return app.render({ bots, edge, scripts, changes: "human-approval" });
AI
The example app is ready. All figures and findings below are synthetic; no Rhone telemetry, usage allotments or current security issues were read.
Draft a digital-product brief on showing up in AI shopping channels. Cover product data, discoverability, agent access policies and how online journeys can connect with store availability.
AI
I can use public site context and, in a real deployment, ask for read-only commerce and analytics capabilities. Access stays no broader than the requester’s permissions.
Gatekeeper: Catalog + Search + AI Crawl Control
Proposed read-only product attributes, aggregate search and crawler analytics; no orders, customer identities or private pricing.
Proposed read-only product attributes, aggregate search and crawler analytics; no orders, customer identities or private pricing.
AI
I would calculate field coverage and schema validity in code, then use a model only to prioritize ambiguous content improvements. All scorecard values here are invented for the demo.
Illustrative code
const attributes = await env.CATALOG.getAttributeCoverage({ access: "read" });
const schema = validateProductStructuredData(attributes);
const priorities = await env.AI_GATEWAY.rankGaps({ schema, access: "scoped" });
return draftBrief({ attributes, schema, priorities });
AI
The example brief separates public observations from hypotheses to validate with Rhone. A Digital Product owner approves any feed, crawler-policy or customer-journey changes.
Build a release workflow for a Nuxt storefront on Cloudflare Pages and Workers. Check performance, cache behavior, structured data and accessibility before each merge.
AI
I would request read-only source, CI and preview metadata. I can draft comments, but cannot merge a PR or alter a deployment.
Gatekeeper: Source Control + Pages + Web Analytics
One approved repository, preview build results and aggregate performance measures; comments proposed for owner review.
One approved repository, preview build results and aggregate performance measures; comments proposed for owner review.
AI
Known checks run without a model. Only ambiguous UX copy or architecture questions reach a model through AI Gateway, with per-task attribution and a budget.
Illustrative code
const preview = await env.PAGES.getPreview({ pullRequest: "DEMO-42" });
const checks = runStaticChecks(preview, ["build", "a11y", "schema", "cache", "vitals"]);
const questions = checks.filter(check => check.requiresJudgment);
return queueForHumanMerge({ checks, questions, merge: false });
AI
The sample PR has a simulated cache-policy regression and a missing product availability field. The workflow drafted review comments; no production deployment or merge was performed.
Prepare a shared 2027 planning review for the digital product, security and web development leads: the commerce funnel, AI readiness, omnichannel, platform capacity and peak-season resilience.
AI
I would use approved team-level aggregates only. Product plans, internal contracts and customer-level details stay out of this public example.
Gatekeeper: Analytics + Catalog + Security + AI Gateway
Hypothetical aggregate funnel, product-readiness, storefront performance, security and model-spend summaries, each scoped to authorized viewers.
Hypothetical aggregate funnel, product-readiness, storefront performance, security and model-spend summaries, each scoped to authorized viewers.
AI
The five-slide example separates target metrics from actual results. As this prototype has no Rhone data, all sample numbers and owners are illustrative and must be validated before a real presentation.
AI
The deck blueprint is ready. Sharing it would re-check every viewer against the resources used; a cross-functional lead approves any real distribution.
Peak Security & Bot Command Center
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
FOR THE SR DIRECTOR, CYBERSECURITY & TECH OPERATIONS
73%
Example bot traffic handled at edge
12
Example verified agents
2
Proposed changes, awaiting review
61%
Example cache offload
Scenario: A sample gift-card lookup path attracts unverified automation. Drafted a rate-rule review; no rule has been changed.
Sample traffic decisions
| Traffic class | Example signal | Recommendation | Action |
|---|---|---|---|
| Verified shopping agents | Signed identity / intended purpose | Observe and allow useful discovery | Read-only |
| Unverified scraping | High-volume product requests | Review challenge thresholds | Owner approval |
| Gift-card lookup | Repeated lookups in scenario | Review rate limit & bot signals | Owner approval |
| Checkout scripts | Example script inventory diff | Validate change provenance | Owner approval |
Edge capacity & continuity
Worker request trend
67%
Example
Bandwidth vs. sample plan
54%
Example
Certificate review coverage
90%
Review
Usage percentages are made up for the demo. In a real deployment, authorized Cloudflare analytics would establish the actual baseline, plan limits and alert thresholds.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
App blueprint
In Cloudflare OS, client code is sandboxed in the browser, and server code runs in a Dynamic Worker with global outbound networking disabled. A Durable Object Facet can give this app separate SQLite-backed state. Typed Gatekeepers mediate authorized reads; keys are never supplied to generated code.
Illustrative server methods
const snapshot = await env.SECURITY.getBotTrends({scope: "storefront", access: "read"});
const capacity = await env.EDGE_USAGE.getMonthlySummary({access: "read"});
return { snapshot, capacity, proposeChange: "requires-security-owner" };
Policy trace
| Step | Boundary | Result |
|---|---|---|
| Read security trends | Scoped analytics Gatekeeper | Logged |
| Assemble app | Sandboxed UI + isolated Worker | Example |
| Change WAF configuration | Named security approver | Not executed |
| Share app | Re-check viewer against observed data | Per viewer |
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Resource boundaries
Cloudflare Security GatekeeperRead-only bot, WAF and verified-agent summaries for authorized zones; rule changes require human approval.
Edge Analytics GatekeeperRead-only request, bandwidth, cache and Worker summaries; no actual Rhone plan or allotment is connected.
Script and domain inventoryApproved inventory and certificate status only; no new production configuration is inferred.
Incident and change portalRead incident and ticket metadata; drafting a change is separate from applying it.
Sharing & approvalsEach viewer is checked against every observed resource. Gatekeepers hold credentials; consequential actions need a human owner. Nothing on this page is connected.
AI Channel & AEO Readiness Brief
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
FOR THE SR DIRECTOR, DIGITAL PRODUCT
AI Channel & AEO Readiness
The opportunity
Shoppers may encounter products through AI discovery before they visit a storefront. An agent-ready experience starts with complete, accurate product attributes, machine-readable pages and intentional crawler access. Whether those pathways drive traffic or sales needs measurement, not assumption.
Proposed workstreams
| Workstream | Example check | Decision owner |
|---|---|---|
| Catalog quality | Fit, fabric, care, size, color and in-stock fields by product | Merchandising + Digital |
| Discoverability | Product structured data, canonical pages and feed freshness | Web Development |
| Agent access | Permit useful verified discovery; limit scraping and checkout abuse | Security + Digital |
| Omnichannel | Whether store inventory, pickup and shipping promises can be exposed accurately | Retail Operations |
First 30 days
- Audit a sample of product pages and structured data against the catalog source of truth.
- Agree which verified AI agents should see product and availability content.
- Define an AI referral and assisted-conversion measurement plan that respects privacy.
- Test whether pickup and shipping availability can be shared without overstating fulfillment guarantees.
The public site suggests commerce and search tools; their current configuration, AI-channel performance and availability accuracy would need Rhone validation. This is a proposal, not a report on Rhone’s internal results.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Readiness scorecard
| Sample category | Example complete attributes | What to check |
|---|---|---|
| Performance tops | 91% | Material + fit descriptions |
| Pants & joggers | 85% | Size and inseam variants |
| Women’s collection | 79% | Color and care details |
| Store pickup pages | To validate | Inventory promise and availability |
Measurement design
Measure verified AI crawler access, product-page schema validation, referral attribution where available, and product-to-checkout funnel changes. Keep attribution separate from evidence of a completed transaction.
Policy design
Use Cloudflare AI Crawl Control for intentional crawler rules, security signals to identify abusive traffic, and human review before changing checkout or discovery policies. An agent should not be treated as a trusted buyer simply because it crawled a page.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Resource boundaries
Shopify catalog exampleProduct data only, selected attributes; public storefront signal, not a live integration.
Algolia search exampleAggregate discovery and query quality; no customer-level searches in this prototype.
Cloudflare AI Crawl ControlCrawler visibility and policy proposals; changing a rule needs a human owner.
Store and fulfillment portalInventory and pickup commitments only if authorized and validated for freshness.
Sharing & approvalsEach viewer is checked against every observed resource. Gatekeepers hold credentials; consequential actions need a human owner. Nothing on this page is connected.
Storefront Release Review
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
FOR THE MANAGER, WEB DEVELOPMENT
5 / 7
Example checks pass
2
Proposed fixes
1
Optional model review
0
Autonomous merges
Event-triggered release path
1
PR + Pages preview
Read only the selected Nuxt repo, preview deployment and changed paths.
2
Deterministic gates
Build/tests, LCP/INP budgets, bundle delta, cache TTL and hit ratio, Worker CPU, accessibility, product structured data and third-party scripts.
3
Selective judgment
Use AI Gateway only for ambiguous brand-voice and UX tradeoffs; keep request logs and team spend visible.
4
Human merge & rollout
Post draft review, ask a code owner to merge, then monitor authorized signals with a rollback plan.
Example only: PR DEMO-42 flags a cache-rule change on product pages and a missing availability property. The workflow cannot merge or deploy.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Sample run: DEMO-42
| Check | Example result | Next step |
|---|---|---|
| Nuxt build + tests | Pass | Continue |
| Core Web Vitals budget | Pass (sample LCP 2.2s) | Monitor field data |
| Cache rule diff | Review: changed product TTL | Confirm freshness with owner |
| Product structured data | Review: availability omitted | Update schema mapping |
| Accessibility | Pass with 1 suggestion | Human UX review |
| Worker CPU + bundle | Pass within example budget | Monitor after merge |
Sharing and permissions
The reviewer sees only the granted repository and preview metadata. A person viewing the report later needs permission to the observed resources; proposing a PR comment does not confer merge permission.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Resource boundaries
Source Control GatekeeperApproved storefront repo and PR metadata, read-only; comments proposed, not posted automatically.
Cloudflare Pages + WorkersPreview build status and aggregate Worker resource use; no deploy token in generated code.
Analytics + CI GatekeepersAuthorized performance budgets, test results and cache-rule diffs.
AI GatewayOptional, budgeted model review for ambiguous questions; deterministic checks do not call a model.
Sharing & approvalsEach viewer is checked against every observed resource. Gatekeepers hold credentials; consequential actions need a human owner. Nothing on this page is connected.
2027 Digital & Platform Planning
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
CROSS-FUNCTIONAL VIEW · THREE ROLE OWNERS
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Deck source ledger
| Slide | Potential authorized source | Check before sharing |
|---|---|---|
| Customer journey | Aggregate web analytics | Separate actuals from sample targets |
| AI channels | Catalog + crawler observations | Validate inventory and agent identity |
| Omnichannel | Store + fulfillment rules | Owner signs off on promises |
| Platform | Pages, Workers, cache telemetry | Match period and plan definition |
| Decisions | Curated strategy and team review | Named human owner signs off |
Real outputs may be shared only with people who are authorized to view every observed source. A copied blueprint carries neither private data nor credentials.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Resource boundaries
Product analytics GatekeeperTeam-level shopping-funnel aggregates only; no order or customer records.
Catalog + retail operationsApproved product and availability summaries, with freshness checks.
Cloudflare security + edge usageAuthorized aggregate traffic, performance and capacity, aligned to the same period.
AI GatewayTeam-level model routing, budget and rate-limit view; sample values only.
Sharing & approvalsEach viewer is checked against every observed resource. Gatekeepers hold credentials; consequential actions need a human owner. Nothing on this page is connected.
Integrations
Organization-wide Gatekeepers could expose scoped capabilities while isolating credentials, logging observations and enforcing approval rules.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
Named commerce tools are suggested by public storefront signals; their present use, configuration and authorization are not verified.
Example Gatekeepers
Productivity suite
Calendar, mail, docs and files under employee permissions
Source control + CI/CD
Authorized repositories, previews, tests and release records
Cloudflare Security
Zone-scoped bot, WAF, AI crawler and request analytics
Cloudflare Pages + Workers
Deploy previews, build status, edge metrics and CPU trends
Cloudflare AI Gateway
Attributed model calls, DLP policy, spend and rate limits
Shopify (example)
Selected catalog fields and aggregate commerce data
Contentful (example)
Curated content, assets and publishing rules
Algolia (example)
Aggregate search quality and discovery signals
Klaviyo (example)
Approved campaign metadata and audience-safe summaries
Global-e (example)
Cross-border availability and shipping policies
Yotpo (example)
Moderated review summaries, not private customer profiles
Customer support platform
Permissioned ticket themes, no direct customer PII
Retail / inventory system
Store availability and pickup rules with freshness checks
Finance / ERP system
Approved aggregate budgets and purchasing policies
HRIS / people operations
Role-scoped onboarding and people policies
Legal / contract system
Controlled templates, review queues and retention policies
MCP Server Portals
Example portals for discovering scoped MCP servers across the whole organization. No authorization is performed by this demo.
Commerce & merchandising
Catalog, inventory and approved product context
Security & IT
Protective analytics, tickets and proposed changes
Marketing & creative
Campaigns, approved brand assets and reporting
Retail operations
Store services, pickup and fulfillment procedures
People & finance
Policy access and aggregate financial controls
Web development
Source, CI, previews and release runbooks
Organization Context
Shared, versioned knowledge for authorized work across Rhone — brand, stores, customer care, finance, people, operations and technology.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
company-mission-and-strategy.md
Mission, values, planning cadence and desired outcomes
brand-voice-and-creative.md
Brand expression, imagery, accessibility and approvals
product-and-fabric-guide.md
Product taxonomy, fit language, fabric and care references
merchandising-playbook.md
Assortment, launch and seasonal planning
retail-store-operations.md
Store service standards and escalation paths
omnichannel-policies.md
Availability, pickup and shipping promise definitions
customer-experience-standard.md
Support tone, returns and issue handling
wholesale-partnerships.md
Partner process and approved communications
supply-chain-standards.md
Supplier, quality and fulfillment procedures
marketing-and-campaigns.md
Campaign planning and channel standards
security-and-identity.md
Access tiers, incident handling and protective controls
engineering-standards.md
Reviews, release gates and reliability practices
data-classification-and-privacy.md
Customer data handling, retention and permissions
responsible-ai.md
Model evaluation, human ownership and inference policy
hr-and-people-policies.md
Hiring, onboarding and employee services
finance-and-procurement.md
Budgeting, purchasing and financial controls
legal-and-compliance.md
Contract templates, consent and publication review
Skills
Reusable jobs-to-be-done across the entire organization. A human owns each output.
Illustrative Rhone concept. All scenarios, metrics, credentials, policies and connections are simulated. No live Rhone systems are connected.
| Name | Description | Group | Source |
|---|---|---|---|
| meeting-prep | Prepare a briefing from authorized calendar, product and company context | General | Prototype library |
| weekly-operating-review | Summarize cross-functional decisions, owners and next steps | General | Prototype library |
| campaign-brief | Create a brand-safe launch outline for human creative review | Marketing & Creative | Prototype library |
| creative-review | Compare draft content with approved brand and accessibility guidance | Marketing & Creative | Prototype library |
| merchandising-analysis | Summarize assortment and product attribute gaps | Merchandising | Prototype library |
| product-description-audit | Find missing fit, fabric, size and care fields | Merchandising | Prototype library |
| store-operations-summary | Prepare a store-facing summary from approved playbooks | Retail Operations | Prototype library |
| pickup-promise-check | Verify inventory freshness before proposing store pickup copy | Retail Operations | Prototype library |
| support-theme-review | Summarize aggregate support patterns without customer PII | Customer Experience | Prototype library |
| returns-policy-draft | Draft explanations using approved customer care policies | Customer Experience | Prototype library |
| web-release-review | Check preview, accessibility, performance and structured data | Digital & Web | Prototype library |
| ai-discovery-audit | Audit proposed product-feed and crawler-readiness signals | Digital & Web | Prototype library |
| incident-response | Assemble scoped evidence and route decisions to incident owners | Security & IT | Prototype library |
| vendor-risk-review | Evaluate proposed vendors against approved requirements | Security & IT | Prototype library |
| access-audit | Reconcile role entitlements and human approval checkpoints | Security & IT | Prototype library |
| supplier-scorecard | Draft a quality and delivery view from approved aggregate data | Supply Chain | Prototype library |
| inventory-exception | Summarize supply alerts and fulfillment impact for an owner | Supply Chain | Prototype library |
| budget-analysis | Explain approved financial variances for decision makers | Finance | Prototype library |
| purchase-request | Prepare a governed purchasing summary, not an order | Finance | Prototype library |
| onboarding-guide | Assemble approved employee onboarding references | HR | Prototype library |
| job-posting-draft | Draft a role description from approved people policies | HR | Prototype library |
| contract-brief | Summarize authorized contract clauses for legal review | Legal | Prototype library |
| policy-comparison | Compare controlled policy versions and flag open decisions | Legal | Prototype library |
Profile
Illustrative account information for this public prototype.
Account
Demo User
No personal information is stored
Display name
Demo User
User ID
demo.user@example.com
AI Gateway
Illustrative scenario data for simulated model routes, controls, and usage — one console.
Illustrative scenario. Providers, models, traffic, latency, spend, users, teams, budgets, and controls are simulated; they do not describe Rhone systems or activity.
Tokens
34.2M
Est. spend
$912
Cache-hit
27%
Error rate
0.6%
p50 latency
480 ms
Illustrative Model Traffic
This month| Model | Route | Tokens | Spend | Token share | p50 latency |
|---|---|---|---|---|---|
| Llama 3.3 70B | Workers AI | 15.4M | $214 | 310 ms | |
| Claude | via AI Gateway | 9.8M | $398 | 720 ms | |
| GPT-4o | via AI Gateway | 6.1M | $251 | 640 ms | |
| Workers AI embeddings (bge) | Workers AI | 2.9M | $49 | 40 ms |
Spend vs. Budget
1 day remaining$912spent of $2,000 cap
46%
On track · ~$1,088 left with 1 day
Illustrative Users
Demo User 014.2M tok $118
Demo User 023.1M tok $96
Demo User 032.8M tok $84
Demo User 042.2M tok $61
Illustrative Usage by Workspace / Team
34.2M tokens totalMarketing & Creative
12.1M tokens · $324
Digital & E-commerce
8.9M tokens · $246
Customer Experience
6.2M tokens · $151
Merchandising & Planning
4.1M tokens · $102
Security & IT
2.9M tokens · $89
Governance
Illustrative guardrail settings modeled with Gatekeepers + AI Gateway, resource-scoped access, audit trails, and human approval.
Illustrative settings. Every model, limit, retention period, approval, and control below is simulated; no Rhone policy or configuration is connected.
Per-team allowed models
Restrict which providers each workspace can call.
Monthly spend caps
Hard limits per team; agents stop before overrun.
Marketing & Creative $500Digital & E-commerce $400Customer Experience $300Merchandising & Planning $250Security & IT $200
PII redaction
Strip sensitive fields from prompts before they leave.
Enabled
Prompt / response logging
Full request logs retained for audit & review.
Enabled · 90-day retention
Rate limits
Per-team request ceilings to protect budgets.
600 req / min|burst 1,000
Default-deny access
An agent starts without resources. Typed Gatekeeper bindings grant only the scope a person approves.
Observation-aware sharing
A viewer's rights are re-checked against every resource an agent observed, even in shared outputs.
Raise Customer Experience cap to $400
Simulated change queued by an agent — needs a human sign-off.
Requires approval