Platform Architecture

State of Stick separates physical production, connected identity, customer experiences, authoritative business workflows, measurement, AI assistance, and security so each can evolve within its own responsibility boundary.

Layer	Public responsibility
Physical layer	The emblem, display, package element, collectible, or installation
Connected identity layer	The managed reference that associates an approved object with an experience
Experience layer	Customer-facing content, stories, instructions, offers, or workflows
Workflow layer	Server-authoritative claim, registration, payment, fulfillment, signing, service, or ownership state
Analytics layer	Policy-scoped measurement of intentional interactions and outcomes
AI layer	Registered, source-aware assistance for research, drafting, interpretation, and planning
Operations layer	Customer provisioning, configuration, production handoff, approvals, support, and audit
Security layer	Access boundaries, validation, authorization, privacy, and operational safeguards

Logical architecture

physical object
      ↓
connected entry point
      ↓
identity resolution ─────→ lifecycle and trust evaluation
      ↓
approved experience
      ↓
authoritative workflow ──→ payment / claim / fulfillment / service
      ↓
policy-scoped events
      ↓
quality controls ────────→ customer and internal reporting

Administrative control follows a separate path:

named identity
      ↓
session + role + capability
      ↓
customer or internal scope
      ↓
validated state transition
      ↓
audit record

Public interaction and administrative control do not share one universal trust boundary.

Why the layers are separate

• Experience content can change without remaking every physical object.
• Programs can choose an interaction and trust level appropriate to their use case.
• Customer access can remain separated from internal administration.
• Measurement can follow a defined privacy policy rather than being embedded indiscriminately.
• Sensitive business rules stay controlled outside the public page and physical identifier.
• Physical interactions, participant identity, and business authorization can be evaluated separately.
• Raw event history can remain distinct from quality-filtered reporting.
• AI providers can change behind a stable, registered task boundary.

Architectural properties

Change without physical replacement

The object identity is managed separately from destination content. An approved experience can be updated, paused, or retired without remaking the physical object.

Lowest sufficient trust

Standard interactions handle content and ordinary engagement. A verified tier is selected only when the workflow needs stronger evidence of authentic physical presence.

Server authority

Eligibility, prices, permissions, payments, fulfillment, ownership transitions, and publication are determined by server-side state—not by URL parameters, browser claims, static copy, or AI output.

Scoped intelligence

Analytics follow the applicable collection and retention policy. AI tasks follow a registered purpose, model lane, sensitivity, schema, and review rule. Neither layer receives general authority over the application.

Replaceable infrastructure, durable domain model

The public system contract is expressed in customers, programs, objects, identities, experiences, interactions, workflows, outcomes, and reports. Specific infrastructure components may evolve without changing those core responsibilities.

Deployment posture

The web and edge platform is designed around Cloudflare services, including edge application execution, access boundaries, managed data services, AI routing, and deployment controls. This reduces the distance between an object interaction and the policy or experience that must be evaluated.

The public architecture does not depend on exposing infrastructure identifiers or presenting a vendor list as the product. State of Stick’s product is the coordinated physical-and-digital operating model implemented on that infrastructure.

For the domain and state model, see Data and Control Model. This is a public-safe architecture for evaluation and planning. Database schemas, infrastructure topology, credentials, secure-device configuration, internal routes, anti-abuse thresholds, and operational security procedures are intentionally excluded.