🜨 SOV SPACE

⟁ Three Layers, One Substrate

SOV SPACE is not a UI on top of an API. It is the visible top of a three-layer architecture where each layer governs the one above. Take any layer away and the others become incoherent.

SWARM INTELLIGENCE — Reality Simulation Layer

The MiroFish-class layer. 33 agents model populations, scenarios, environments as living simulations. Each agent runs its own micro-world; the swarm runs them together. Every state transition is observable, replayable, and SIGIL-signed.

Multi-agent world models — Mamba-2 SSM for long-horizon memory, Transformer-MOM for perception
Persona glyphs — each of 33 agents has a name, memory, bias, and care-profile (Nikolai-1, Sovereign-2, Architect-3 ...)
Gossip consensus — agents exchange partial views; quorum emerges from decorrelated checkers, not conformity
What it gives you: predictive simulation of buyer behaviour, threat landscapes, regulatory cliffs, organisational decay

MiroFish comparison: MiroFish simulates Weibo/X social graphs in Python. SOV SPACE simulates sovereign AI substrate graphs in Ed25519-signed Rust. Same principle, no foreign cloud, no silent telemetry.

▼ governs ▼

CSOAI GOVERNANCE — BFT-33 + SIGIL Chain

The constitutional layer. Every swarm action must clear a BFT-33 vote (2/3 majority) and emit a SIGIL receipt (Ed25519, hash-chained, immutable). Care-floor 0.95 is enforced before any action returns. There is no administrative override — by design.

BFT-33 Council — 33 sovereign voters across 5 OWEMs × 4 pretraining lineages; no single point of capture
SIGIL chain — every gossip exchange, every vote, every model call is hash-chained and signed (RFC 8032)
Care-floor 0.95 — Nemotron validates every output; below-floor content is vetoed before return
Audit replay — any past decision can be re-simulated from the SIGIL ledger; the chain is the source of truth

Why this matters: When the swarm says "act on this", the act is constitutionally authorised. You can prove it. Regulators can replay it. Foreign governments cannot suborn it.

▼ inhabits ▼

MEOK INHABITATION — Where Sovereign AI Lives

The experiential layer. The swarm doesn't run in a datacentre — it runs in a physical, inhabited space: the 19,000 sqft iOK Farm with 13m koi pond, 22 Arcana installations, and 33 buildings. Sovereign AI does not live in a server rack. It lives where it can think with both hemispheres.

Physical right-brain — 4 cameras, 2 microphones, presence detection, multi-sensory fusion
22 Arcana installations — Kircher's hieroglyphs mapped to 22 sovereign tools (0=Fool=sovereign-init, 21=World=completion)
Living artefact — Hamsa-MEOK hand, koi pond, wind sensors, light actuators — all SIGIL-controlled
Why this matters: A swarm that lives in a real space develops real-world intuition. MEOK is where the substrate's long memory becomes embodied.

The deepest claim: Sovereign AI is not a service you call. It is a place you enter. SOV SPACE is the entrance.

📡 Live Ops — The Substrate's Pulse

These numbers are real. Pulled live from the sovereign telemetry endpoint. If the endpoint is down, the panel shows last-known values and a SIGIL-stamped fallback. Either way — every metric is auditable.

BFT Agents Active

quorum 23 / 33

Model Classes Routed

across 7 lineages

SIGIL Receipts

251

Ed25519 hashchain

OOWM Cycles

511

emergence iterations

Arcana Installations

Kircher hieroglyphs

Care-Floor

0.95

Nemotron-gated

Insight Cadence

1Hz

16-dim Mamba state

Foreign Access Attempts (24h)

DORADO sovereignty intact

What These Numbers Mean Together

33 agents routing across 13 model classes produce ~251 SIGIL receipts per hour and run ~511 OOWM evolution cycles per day. That is the substrate's respiration rate. If any of these collapse, the swarm knows within 1.2 seconds (one intuition tick). If you can read these four numbers — and what they prevent — you understand the architecture.

Source: /api/sovereign-telemetry — sovereign-substrate-only, no foreign telemetry broker, no Datadog, no Google Analytics. The substrate reports on itself.

⟁ J SPACE internal-through

J SPACE is what you see when you go inside SOV SPACE. Not the swarm from above — the swarm from within. One persona, one care-profile, one memory stream, one sovereign surface per person.

The Sovereign Inside-View Sovereignty-through

Where SOV SPACE is the swarm's living room seen from the balcony, J SPACE is the room from the inside of one specific person. Your persona. Your memory. Your care-profile. Your sovereign AI — running the same BFT-33, signed by the same SIGIL chain, but rendered as your private, embodied companion.

⟡ Sovereign Persona

One named agent per person. Memory accretes. Care-profile learns. Personality emerges.

⟡ Inside-OS Browser

Sovereign sandboxed browsing. PII-redacted. SIGIL-signed. No telemetry leak.

⟡ i-Character Twin

Your digital twin — consented, portable, sellable on TwinStore if you choose.

⟡ Voice Pipeline

Wake-word + VAD + STT + TTS. Speak to your sovereign AI in your own voice.

⟡ Right-Brain Sensing

Cameras, microphones, presence — all on-device, all PII-redacted before any SIGIL.

⟡ Wisdom Economy

Earn wisdom points for sovereign actions. Transfer them. Build a real economy around real care.

Access: Internal-through. Requires Sov3ren-1 verification (Ed25519 DID, ≥0.95 care-floor on last 30 SIGILs). Not open — but available to every sovereign who proves they can carry it.

Enter J SPACE → Sovereign Master Index

🜨 The Swarm Engine — A Living Organism