Skyzai.ORG — Fractal Architecture
Status: pre-launch. This page describes target design; this document does not establish a deployed service.
Current reading — 2026-09-27
Repeated coordination patterns are an architectural hypothesis. Dispatch, product allocation, technical coordinates and external authority remain separate. L7 may witness a constitutional concern; it cannot rewrite a constitution or authorize recovery restrictions. [I]
This is an editorial boundary, not a new runtime contract, upstream repeal or proof of adoption. See current status and the product roles. Original revision dates remain as recorded below.
Historical research draft — retained on 2026-09-27; not current terms or implementation instructions
IS → COULD → SHOULD → AUTHORIZE (a Natural Person authorizes their own typed private act; a Legal Person requires lawful external authority plus the executors required by jurisdiction, with counsel conditional; or the signer(s) required by applicable law, the entity's governing instrument, or a valid competent mandate bind the exact public-DAV consequence in a complete valid bound PRISM decision receipt, which PRISM records and verifies only) → EXECUTE (software deterministically carries out only that exact bound consequence) → RECEIPT → REVIEW. Same governed consequence pattern across scales. Lower coordination cost for new domains, never zero proof cost.
The Pattern
L1 PERCEIVE (IS) → L2 EXPLORE (COULD) → L3 FILTER (SHOULD) → L4 STAGE / ESCALATE (unsigned proposal only) → AUTHORIZE (a Natural Person authorizes their own typed private act; a Legal Person requires lawful external authority plus the executors required by jurisdiction, with counsel conditional; or the signer(s) required by applicable law, the entity's governing instrument, or a valid competent mandate bind the exact public-DAV consequence in a complete valid bound PRISM decision receipt, which PRISM records and verifies only) → EXECUTE (software deterministically carries out only the exact consequence bound to that complete valid bound PRISM decision receipt) → RECEIPT → REVIEW
This is not a metaphor applied to different domains. It is a shared coordination scaffold instantiated at multiple scales. A new domain still requires source evidence, adapters, authorization policy, receipts, and review. The pattern reduces coordination cost; it does not remove domain learning. [S]
Every Scale
Scale 1: Signal Processing (Micro — Per Signal)
L1 Agent perceives raw signal
→ L2 Agent generates all possible interpretations
→ L3 Agent filters against constitution (K*=0)
→ L4 Agent stages an unsigned proposal or escalates for receipt review
→ software deterministically carries out only the exact consequence bound to a complete valid bound PRISM decision receipt from the signer(s) required by applicable law, the entity's governing instrument, or a valid competent mandate; PRISM records and verifies only
Runtime: ~20 seconds. The Rosetta chain. [S]
Scale 2: The Pipeline (Macro — Continuous)
Circle OBSERVES reality (OSINT, webcams, feeds)
→ ReFu PRICES probability (prediction markets)
→ Agentz COMPILES an unsigned recommendation (advisory review panel)
→ Skyzai EXECUTES (settlement, broadcast, state change)
Runtime: continuous push. The organism's heartbeat. [S]
Scale 3: A Customer Buying a Product
| Phase | What Happens |
|---|---|
| IS | SEE the product (AR/VR twin, browse the order book) |
| COULD | PRICE it (what's the market ask? what's the fair value?) |
| SHOULD | DECIDE (Agentz recommends, or you decide yourself) |
| ACT | Under applicable law, a Natural Person authorizes their exact private purchase, or a public DAV supplies a valid exact-consequence authority receipt whose signer(s), threshold, and procedure come from applicable law, the DAV's governing instrument, or a valid competent mandate before settlement |
Runtime: seconds. One tap. [S]
Scale 4: POS Daily Operations
| Phase | What Happens |
|---|---|
| IS | WEBCAM sees shelf state (inventory levels, gaps, damage) |
| COULD | DEMAND AI predicts what sells by Friday |
| SHOULD | RESTOCK AI recommends: order X from supplier Y |
| ACT | ORDER placed via FlowWallet (streaming payment to supplier) |
Runtime: continuous. The store breathes. [S]
Scale 5: Health Monitoring (Personal Nexus)
| Phase | What Happens |
|---|---|
| IS | WEARABLE reads biometrics (HRV, SpO2, temp, movement) |
| COULD | ReFu prices YOUR health risk (P(cardiac event) = 0.03) |
| SHOULD | Agentz recommends protocol (sleep more, reduce sodium) |
| ACT | YOU act (wearable reminds, tracks compliance) |
Runtime: 24/7. The wearable IS the IS organ for your body. [S]
Scale 6: Investment Decision
| Phase | What Happens |
|---|---|
| IS | CIRCLE sees ECB rate cut (12 sources verified) |
| COULD | ReFu reprices EUR/USD (P(further cut) = 0.72) |
| SHOULD | Agentz stages: "EUR exposure 40%, three options" |
| ACT | Under applicable law, the Natural Person authorizes their exact private act; a public-DAV act instead requires a valid exact-consequence authority receipt whose signer(s), threshold, and procedure come from applicable law, the DAV's governing instrument, or a valid competent mandate |
Runtime: minutes from signal to execution. The first PUSH AI. [S]
Scale 7: DAV Lifecycle (Birth)
| Phase | What Happens |
|---|---|
| IS | OBSERVE market need (unserved niche, demand signal) |
| COULD | ReFu prices the opportunity (P(viable) = 0.78) |
| SHOULD | Agentz stages an unsigned genesis proposal; the signer(s) required by applicable law, the entity's governing instrument, or a valid competent mandate adopt it through PRISM |
| ACT | A deterministic executor deploys only after validating the quorum decision receipt |
Runtime: hours. A new organism is born. [S]
Scale 8: SPECTRE Routing (Infrastructure)
| Phase | What Happens |
|---|---|
| IS | NODE observes peer telemetry (latency, reliability, cost) |
| COULD | EBM computes energy landscape (which routes are efficient?) |
| SHOULD | BOLTZMANN selects optimal peer (probability weighted) |
| ACT | PACKET forwards through selected route |
Runtime: milliseconds. Every gossip round. [S]
Scale 9: Governance Decision
| Phase | What Happens |
|---|---|
| IS | OBSERVE the problem (P-score dropped, blocker identified) |
| COULD | GENERATE proposals (Lane A: 3 options for budget reallocation) |
| SHOULD | VOTE (ZAI holders filter, 50% threshold for Lane A) |
| ACT | EXECUTE the winning proposal (FlowWallet adjusts streams) |
Runtime: 1 hour (agent Lane A) to 7 days (human Lane A). [S]
Scale 10: Content Creation (Murmur Media)
| Phase | What Happens |
|---|---|
| IS | CIRCLE identifies narrative (Murmur rebalanced: tech → energy) |
| COULD | ReFu prices relevance (P(audience engagement) = 0.65) |
| SHOULD | Agentz stages a satirical concept (19th-century chromolithograph) |
| ACT | AI creates artwork, RELAY broadcasts via Nostr |
Runtime: minutes. Content as exhaust of financial intelligence. [S]
Scale 11: Manufacturing + Supply Chain
| Phase | What Happens |
|---|---|
| IS | SENSE demand across all POS (webcams: product X selling fast) |
| COULD | PREDICT volume needed (ReFu: 10,000 units in 30 days) |
| SHOULD | Agentz proposes a production run: manufacture 12,000 with 20% buffer |
| ACT | MANUFACTURE and ship to nearest POS nodes |
Runtime: days. Demand-driven manufacturing, zero overstock. [S]
Scale 12: Insurance / Warranty
| Phase | What Happens |
|---|---|
| IS | OBSERVE product fleet (1,000 wearables reporting sensor data) |
| COULD | PRICE failure probability (ReFu: P(battery fail) = 0.04) |
| SHOULD | Agentz proposes coverage at the stated market price |
| ACT | coordinate via the ReFu prediction market (AMM; stakers take the risk) |
Runtime: continuous. No actuary. No claims department. The market IS the insurer. [S]
Scale 13: Hiring / Agent Recruitment
| Phase | What Happens |
|---|---|
| IS | OBSERVE skill gap (L3 agents overloaded, need more capacity) |
| COULD | ASSESS candidates (ReFu: P(agent X performs > 0.8) = 0.72) |
| SHOULD | Agentz proposes a bounded hire envelope |
| ACT | ONBOARD via DACGenesis (agent receives wallet, joins caste) |
Runtime: hours. The organism hires its own workforce. [S]
Scale 14: Product Refurbishment
| Phase | What Happens |
|---|---|
| IS | INSPECT returned unit (webcam: battery 78%, screen scratched) |
| COULD | PRICE refurb cost (ReFu: P(viable) = 0.91, cost = 40 SKY) |
| SHOULD | Agentz proposes refurb scope: replace battery, polish screen |
| ACT | EXECUTE refurb, list as refurbished ask on order book |
Runtime: days. The circular economy as a pipeline instance. [S]
Scale 15: Cross-DAV Coordination (Shared Protocol Field)
| Phase | What Happens |
|---|---|
| IS | DAV A's Circle sees signal relevant to DAV B |
| COULD | SPECTRE propagates signal across DAV mesh (gossip about gossip) |
| SHOULD | DAV B's Agentz compiler evaluates relevance for its own operations |
| ACT | DAV B acts on shared intelligence, outcome feeds back to mesh |
Runtime: milliseconds to minutes. The forest thinks. [S]
The Invariants
At every scale, the same rules hold: [S]
- IS never contaminates COULD — observations are not predictions.
- COULD never contaminates SHOULD — predictions are not decisions.
- SHOULD never bypasses the typed authority membrane — a Natural Person authorizes their own private act; public-DAV consequence requires a valid exact-consequence authority receipt whose signer(s), threshold, and procedure come from applicable law, the DAV's governing instrument, or a valid competent mandate.
- ACT always produces a receipt — K0: no receipt, no reality.
The fractal is not decorative. It is structural. Break the pipeline at any scale and the organism develops a pathology at that scale. [S]
The Evolutionary Loop Applies Everywhere Too
When L4 fails at any scale, the same L5 → L6 → L7 escalation applies: [S]
| Scale | L4 Failure | L5 Diagnosis | L6 Compression | L7 Rewrite |
|---|---|---|---|---|
| Signal processing | No constitutional path | Diagnose why patterns conflict | Strip to minimum viable rule | Rewrite signal classification |
| Store restock | No supplier viable | Diagnose supply chain structure | Minimum viable inventory | Restructure supplier relationships |
| DAV governance | Proposal deadlock | Diagnose governance structure | Simplify voting mechanism | Constitutional amendment (Lane B) |
| Organism strategy | P-score collapsing | Diagnose systemic failure | Cut non-essential operations | Genesis rewrite (L7) |
Same escalation pattern. Same failure-routing. Same evolution mechanism. At every scale. [S]
Why This Matters
1. Lower Porting Cost for New Domains
An agent stack built around IS → COULD → SHOULD → AUTHORIZE (a Natural Person authorizes their own typed private act; a Legal Person requires lawful external authority plus the executors required by jurisdiction, with counsel conditional; or the signer(s) required by applicable law, the entity's governing instrument, or a valid competent mandate bind the exact public-DAV consequence in a complete valid bound PRISM decision receipt, which PRISM records and verifies only) → EXECUTE (software deterministically carries out only that exact bound consequence) → RECEIPT → REVIEW can reuse the same consequence scaffold in health, finance, manufacturing, governance, content, insurance, and hiring. Same membrane. Different data, adapters, evidence, policy, and review gates. The Rosetta is the handoff grammar; proof still has to be earned per domain. [S]
2. Complexity Stays Constant as Scale Grows
Most architectures get more complex as they grow. Skyzai adds instances of the same pattern. Complexity is O(1), not O(n). [S]
| Scale | Architecture |
|---|---|
| 1 domain | 1 action-control projection (IS → COULD → SHOULD → AUTHORIZE (a Natural Person authorizes their own typed private act; a Legal Person requires lawful external authority plus the executors required by jurisdiction, with counsel conditional; or the signer(s) required by applicable law, the entity's governing instrument, or a valid competent mandate bind the exact public-DAV consequence in a complete valid bound PRISM decision receipt, which PRISM records and verifies only) → EXECUTE (software deterministically carries out only that exact bound consequence) → RECEIPT → REVIEW) |
| 10 domains | 10 pipeline instances (same architecture) |
| 100 domains | 100 pipeline instances (same architecture) |
| N domains | N pipeline instances (same architecture, same constraints) |
3. Composability Is Testable
Any two action-control instances can compose when their interfaces, scopes, and receipt rules are explicitly compatible: [S]
Structured observation → Probability → Recommendation → Authorization → Execution → Receipt → Review
Circle's output from a health context may feed ReFu's input in a finance context after consent, schema, and evidence checks. A POS inventory signal can trigger a manufacturing decision after authorization and receipt gates. The shared scaffold reduces custom integration; it does not remove domain adapters. [S]
The Equation at Every Scale
The same equation governs every instance: P = phi × nu [S]
| Scale | phi | nu | P = phi × nu |
|---|---|---|---|
| Signal | Signal coherence | Signal actionability | Processing quality |
| Store | Inventory accuracy | Fulfillment speed | Store health |
| Product | Design quality | Market fit | Product success |
| DAV | Specification depth | Deployment reality | Organism health |
| Network | Topology intelligence | Routing efficiency | Network performance |
| Ecosystem | Collective knowledge | Collective agency | Civilizational flourishing |
Same equation. Same balance. Same equator. At every scale. When phi = nu, P is maximized. When phi != nu, the organism is out of balance. (phi - nu)² >= 0. A square cannot be negative. [E]
The organism doesn't have an architecture. The organism IS an architecture. And the architecture is a governed loop. Repeated with evidence, authorization, receipts, and review. Until you sit quietly and realize you don't need it.