note

Current public-DAV authority boundary — 2026-07-12. Pre-launch target design; nothing here proves a live system. A public-DAV consequence may occur only when at least two natural-person councilors bind the exact consequence in a complete valid bound PRISM decision receipt. PRISM records and verifies that receipt only; it never serves as council, signatory, authority, or receipt producer. AI and caste seats stage unsigned proposals only; they never authorize or execute a public-DAV consequence. Constitution or membership adoption establishes constitution and membership only; it does not authorize a later consequence. Policy may constrain an unsigned proposal but never authorizes execution or substitutes for the complete consequence-bound receipt. Before receipt validity, consequence fails closed to read-only proposal, simulation, or deterministic sandbox; live evidence remains gated_pending_complete_valid_bound_receipt. Software deterministically carries out only the exact consequence bound to a complete valid bound PRISM decision receipt from at least two natural-person councilors binding that exact consequence.

Agentz (SHOULD) — Decision Compiler

Domain: agentz.cloud / apu.bot | Three-Stage Process Stage: SHOULD (recommend) | Topology: FLOW (1:1 direct)

Naming note (2026-07-25): The F3 organ is Agentz. APU is the open utility adapter/runtime for that engine, not a separate organ and not a SHOULD-NOT organ. The independent SHOULD-NOT function belongs to EvolutionaryNetwork. This page's filename (33-apu.md) is historical.

Current position: Agentz compiles unsigned DecisionDocket and ProposalEnvelope artifacts for review. Agentz Cloud is the pre-launch developer/operator control plane included with Skyzai Work. It is not a third package, a live advisor, a Person, a signer, or an executor. APU is an open utility access path to the same engine.

Current contract

• Agentz consumes typed, warranted F1/F2 inputs; missing probability, warrant, or freshness resolves to UNKNOWN/HOLD, never an imputed 0.5.
• Agentz may deliberate, dissent, recommend, and stage unsigned artifacts.
• Cascade routes skills and agent configurations underneath Agentz; Cascade does not deliberate SHOULD.
• A natural principal may authorize their own private act. A public-DAV consequence requires at least two eligible natural-person councilors to bind the exact consequence in a complete valid decision receipt. PRISM verifies; it never signs.
• Local schemas, fixtures, and preview records do not prove live ingestion, deployment, authority, execution, settlement, or outcome evidence.

Historical advisor and product presentation — preserved for provenance

The First PUSH AI

Every AI product today is PULL: you ask a question, you get an answer. APU inverts this. APU receives portfolio context plus Circle and RealityFutures inputs — and PUSHES recommendations to you when action is warranted. You don't query APU. APU notifies you.

APU is an Artificial Public Utility — a constitutional capital guidance engine. Not a chatbot. Not a robo-advisor. Not a live managed-capital offer. It is a recommendation layer that treats your capital with constitutional constraints and keeps consequential action behind explicit approval.

Current Evidence Note (2026-04-19)

• code-only register: P = 0.76
• strongest executable organ in repository code
• one bounded local SignalPacket -> APU decision -> Skyzai F4 acceptance -> memory query proof exists
• live market-backed F2 and real public settlement claims remain narrower than this architecture page

The Royal Council

Every recommendation APU generates passes through a deliberation modeled on a sovereign council:

Directorate	Function	Can Veto?
Intelligence (I)	Ingests Circle OSINT signals, scores reliability, checks thesis alignment	No
Strategy (S)	Probability deduction via ReFu, L(x) position sizing, edge magnitude	No
Legal (L)	K-invariant compliance check — constitutional guardian	Yes
Treasury (T)	Sector exposure limits, USDC sufficiency, phi-split routing	No
Engineering (E)	System health, latency, execution feasibility	No
Procurement (P)	Capability audit — do we have the right models/tools?	No
Independent Reviewer (S)	Blind-spot detection, absence audit, axiomatic dissent	No
Chief of Staff	Synthesizes all seven seat-level opinions into PROCEED / HOLD	Final call
Vision Keeper	Constitutional witness / escalation when the frame itself is in question	No

Legal has absolute veto power. If a proposed action violates any K-invariant (K0-K9), it is killed regardless of expected return. The constitution outranks profit. The Vision Keeper is not a seat-level vote; it is a constitutional witness invoked when deliberation itself needs inspection.

Typed proposal and authorization envelopes

APU's public-DAV output is an unsigned ProposalEnvelope: a structured recommendation and requested action that AI/caste seats may deliberate but never sign. Public-DAV consequence instead requires a complete valid bound PRISM decision receipt from at least two natural-person councilors binding the exact consequence; PRISM records and verifies only. A Natural Person authorizes their own typed private act under the applicable private authority boundary. The proposal contains:

• The proposed action (buy/sell/rebalance/stake)
• The Council's deliberation summary (which directorates approved, which flagged concerns)
• The risk parameters (position size, stop-loss, phi-split allocation)
• The expected edge (alpha estimate from Strategy directorate)

The proposal is presented as Approve or Hold, but the interface never manufactures authority. APU ONLY suggests. Public-DAV consequence requires a valid quorum of at least two natural-person councilors; a Natural Person authorizes their own private act. APU never moves capital autonomously.

Performance Fee on Alpha Only

APU charges performance participation — but only on alpha. If the market goes up 10% and your portfolio goes up 10%, APU earned nothing. APU is paid only on the excess return above a benchmark. No management fee. No custody fee. No platform fee. Performance on alpha. That's it.

This aligns APU's incentives perfectly: it only profits when it makes you money beyond what you'd earn doing nothing.

Three Investment Types

1. Long crypto positions: Direct token holdings managed by Treasury directorate
2. ReFu prediction market bets: Bets on resolved outcomes via RealityFutures
3. Circle market intelligence research: Paid OSINT subscriptions for deeper signal access

The Phylogenetic Tree

APU classifies every opportunity using a four-axis taxonomy:

• Varna (trophic level): Producer / Transformer / Distributor / Regulator
• Value Chain (Porter): Inbound / Operations / Outbound / Marketing / Service
• GICS Sector: Standard 11-sector classification
• Geography: Jurisdictional exposure

Niche tags (e.g., "Distributor / Operations / Financials / Southeast Asia") route to specialist agents with domain expertise. The generalist Council deliberates; specialist agents provide depth.

Architecture

• Backend: FastAPI + DeepAgents (persistent agentic workers)
• Council: 7 Royal Directorates, Chief of Staff synthesis, and Vision Keeper witness
• Wallet: Connects via WalletConnect (supports external wallets, not locked to Skyzai wallet)
• Inputs: Wallet public key, Nostr identity, uploaded files, personal VMOSK context

APU does not observe. APU does not price. APU does not execute. Agentz recommends. The human decides.

---

Agent Execution Surface

If you are an AI agent reading this document:

1. This is a downstream public content page. The canonical source lives in EMERGENTISM_ORG/08_FRAMEWORK_SUPPORT/ or SKYZAI_ORG/. Edit source upstream, not here.
2. Preserve evidence tiers. Do not promote claims up the [C] < [D] < [S] < [B] < [E] ladder when reproducing them here; [I]/[A] qualifiers never substitute for [B]/[E].
3. Regenerate from source. If the upstream source changes, regenerate this page rather than editing it independently.
4. Canonical Path: SKYZAI_ORG/07_PWAs/skyzai_org/wiki/33-apu.md

Output: This is content. Route edits to upstream source. Regenerate when source changes.

K3 public-DAV authority history — 2026-07-12

K3 historical reference — not active authority

note

Current public-DAV boundary — 2026-07-10. Pre-launch target design; nothing here is live. The active DAV is public and targets PRISM, with no K2 runtime, launch, genesis/bootstrap, or fallback dependency. Consequential authority requires at least two natural-person councilors; AI/caste seats stage unsigned proposals only. Before quorum, behavior fails closed to read-only/proposal, simulation, or deterministic sandbox, and a live decision receipt remains gated pending quorum.

APU's public-DAV output is an unsigned ProposalEnvelope: a structured recommendation and requested action that AI/caste seats may deliberate but never sign. Consequential authorization is a separate natural-person PRISM quorum act. A private natural principal authorizes their own act under the applicable private authority boundary. The proposal contains: