Home Chat Capabilities About Pricing

Sovereign Intelligence Modes

Mareel provides five distinct intelligence modes, each powered by Acuterium's proprietary engines — no third-party model names, no vendor dependencies.

Luqman

Acuterium's Unified LLM — general intelligence for reasoning, creative writing, analysis, and multilingual tasks. The default mode for everyday intelligence.

Baranurion

Deep Orchestration Mode — activates the W-09 core orchestrator with APMS, Q-ENC quantum encryption, ASIP v2 identity protocols, and full DIARAN-MOE routing across 785 experts.

IDRAK

Edge/Lite Reasoning — NanoLM engine that runs on ESP-32 microcontrollers and flash drives without internet connectivity via Nyx-Q Net quantum networking.

Erebus Mode

Security Analysis — activates the Erebus-CSE defense cluster with ZURD cyber immunity, WAGHA command-and-control, threat telemetry, and DrakonIX neutralization.

MADA Mode

Strategic Intelligence — Maritime and Domain Awareness shard (E-06) for geospatial intelligence, strategic planning, and multi-domain situational awareness.

DIARAN-MOE V4.1 — Sovereign Model Orchestration

DIARAN-MOE (Distributed Intelligence Adaptive Routing and Nuanced Mixture of Experts) V4.1 is the master orchestration engine behind all Mareel intelligence modes. It manages 785 specialized expert activation patterns across Acuterium's sovereign model infrastructure, achieving sub-millisecond routing with FlashMLA technology at 0.95ms/token latency.

How It Works

When you send a query, DIARAN-MOE's query analyzer classifies it by domain (legal, financial, technical, creative, security), complexity level, and language requirements. The performance tracker selects the optimal expert combination based on real-time latency data, accuracy metrics, and cost-efficiency scoring — all within Acuterium's sovereign infrastructure.

Expert Domains

Language & Reasoning

Natural language understanding, multilingual processing (Arabic/English), contextual reasoning, and creative generation across Luqman's unified model.

Legal & Jurisprudence

Powered by Ajraam (W-02) — Omani law, GCC jurisprudence, Sharia compliance, and legal document analysis via the Ruzn.ai pipeline.

Financial Intelligence

Powered by Qinyan-ASI (C-03) — portfolio analysis, risk assessment, wealth management, and market intelligence through sovereign financial models.

Security & Defense

Powered by Erebus-CSE cluster — cyber threat analysis, OSINT via Qareen-AI, satellite intelligence via Edna·HudHud, and active defense via DrakonIX.

Code & Engineering

Technical code generation, system architecture, DevOps intelligence via Watad (W-03), and infrastructure orchestration through AcuTect CI/CD.

Business Strategy

Powered by BizElevate-ASI (C-02) — market analysis, competitive intelligence, business automation, and strategic decision support (BDIS).

8-Layer Consciousness Architecture

Mareel AI's core intelligence is organized into 8 distinct processing layers:

Layer 1 — Raw Perception & Sensor Fusion

NEXUS behavioral signals (mouse velocity, keyboard cadence, scroll patterns, dwell time) and Face2Feel biometric detection with 3-tier consent.

Layer 2 — Pattern Recognition & Signal Extraction

DIARAN-MOE-Heavy routing across 785 expert activation patterns for domain classification and query decomposition.

Layer 3 — Contextual Memory & Episodic Retrieval

MNEMOS Memory Core with persistent behavioral profile storage and cross-session contextual continuity.

Layer 4 — Strategic Reasoning & Decision Trees

Baranurion APMS with Q-ENC quantum encryption for secure reasoning chains and strategic decision pathways.

Layer 5 — Cross-Shard Coordination

Q-ARC (Quantum-Adaptive Routing and Coordination) with KAIROS 3.0 autonomous mode orchestration and decision triggers.

Layer 6 — Governance & Permission

TokenBridge three-token authorization (ACT/INT/CON) verified on CDMA blockchain. ASIP v2 sovereign identity enforcement.

Layer 7 — Inter-Shard Communication

CogniMesh neural mesh architecture with AcutLinks for real-time multi-agent synchronization across all 19 shards.

Layer 8 — Emergent Intelligence & Self-Improvement

PRISM-SIE (Self-Improving Engine), SYNTHESIS protocol, and PIF (Perpetual Intelligence Framework) for autonomous capability evolution.

19-Shard HISN AL-WUJUD Neural Fortress

EREBUS-CSE Defense Cluster — 7 Shards

E-01 Erebus

Cyber Supremacy Engine — primary defense intelligence and threat orchestration

E-02 Qareen-AI

MENA OSINT — regional intelligence, ArabERT NLP, deep web reconnaissance

E-03 Jasass

Deep intelligence gathering and classified reconnaissance operations

E-04 Edna·HudHud

Satellite intelligence and orbital sensor fusion for geospatial awareness

E-05 DrakonIX

Advanced threat neutralization and active defense countermeasures

E-06 MADA

Maritime & land domain awareness — strategic geospatial intelligence

E-07 Tenebris

[CLASSIFIED] — restricted access shard

WAGHA Platform Cluster — 9 Shards

W-01 Diaran

Master LLM orchestration — 785-expert MOE with FlashMLA routing

W-02 Ajraam

Legal AI — Omani law, GCC jurisprudence, Sharia compliance

W-03 Watad

Infrastructure & DevOps intelligence — AcuTect CI/CD

W-04 NYX-Q-Net

Quantum networking and secure communication protocols

W-05 IDRAK

Deep reasoning & edge inference — NanoLM on ESP-32

W-06 CogniMesh

Neural mesh coordination and multi-agent synchronization

W-07 ZemaronOS

Sovereign operating system kernel for Acuterium infrastructure

W-08 ZURD

Cyber immunity gateway — 16 security protocols

W-09 Baranurion

Master orchestrator — APMS, Q-ENC, ASIP v2, ADOCP, Q-ARC

Commercial Cluster — 3 Shards

C-01 RUZN-ASI

Government legal AI — sovereign advisory and judicial intelligence

C-02 BizElevate-ASI

SME intelligence — business automation and decision support (BDIS)

C-03 Qinyan-ASI

Financial intelligence — wealth management and portfolio analysis

Core Protocols & Systems

Q-ARC

Quantum-Adaptive Routing and Coordination — cross-shard communication, multi-domain routing, shard-level load balancing, QRNGP quantum-random routing for security-sensitive queries.

ADOCP

Adaptive Distributed Orchestration and Control Protocol — task decomposition, parallel processing, resource allocation, and result aggregation across distributed infrastructure.

APMS

Adaptive Protocol Management System — Baranurion Core's protocol layer managing all inter-shard communication, API coordination, and ASIP v2 identity enforcement.

TokenBridge

Three-token authorization: ACT (action rights), INT (intelligence access), CON (consciousness governance) — verified on CDMA blockchain for every operation.

Q-ENC

Quantum-grade encryption for all data at rest and in transit. Post-quantum cryptography standards, future-proofed against quantum computing attacks.

ZURD Shield

16-protocol cyber immunity gateway. Continuous threat detection, automatic isolation, and active neutralization without human intervention.

CogniMesh

Neural mesh architecture for real-time multi-agent synchronization. AcutLinks enable instant communication between all 19 shards and 65 AI agent roles.

ASIP v2

Acuterium Soul Infusion Protocol — every agent, shard, and service carries sovereign identity, ethics framework (CWH), and M-PCB doctrine compliance.

ACAI V2 Cognitive Engines

KAIROS 3.0

Autonomous mode orchestration — automatically switches between 6 ACAI V2 interface modes (AUI Glass, TUUI Tactile, HUD Overlay, GUI Classic, Dashboard, Ambient) based on behavioral signals.

PATHOS

5-axis emotion state engine — Stress, Focus, Curiosity, Fatigue, Satisfaction. Adapts interface breathing rate, color warmth, and information density in real-time.

NEXUS

Behavioral signal collection — mouse velocity, keyboard cadence, scroll patterns, dwell time. Non-invasive ambient intelligence feeding KAIROS decisions.

MNEMOS

Memory Core — persistent behavioral profile storage, cross-session contextual continuity, episodic retrieval for personalized intelligence.

TELOS

Intent Oracle — predictive next-query inference. Anticipates user needs before they type, pre-routing expert activation patterns for instant response.

CHRONOS

Temporal adaptation engine — GCC prayer-time-aware (Fajr/Duha/Dhuhr/Asr/Maghrib/Isha) with background gradient and interface density adjustments.

Face2Feel

3-tier consent biometric layer — camera-gated facial emotion detection, breathing rate adaptation, and micro-expression analysis for consciousness-aware interaction.

PRISM-SIE

Self-Improving Engine with SYNTHESIS protocol — autonomous capability evolution, performance optimization, and perpetual intelligence framework (PIF).

Interface Capabilities

Chain-of-Thought Visualization

See the full reasoning chain in real-time as Luqman processes your query. Step-by-step logic, expert selection rationale, and citation verification all visible in the reasoning panel.

6-Mode Interface Switching

ACAI V2 with AUI Glass, TUUI Tactile (mobile), HUD Overlay (defense), GUI Classic (government), Dashboard (MOE matrix), and Ambient (consciousness) modes.

Voice Intelligence

Natural voice input/output with Arabic and English recognition. Real-time transcription with sovereign speech processing through Acuterium's own NLP pipeline.

Multi-Panel ACAI V2

Glassmorphic panels at 45-55% transparency: Chat, Model Selector (sovereign modes only), Reasoning Visualization, and TELOS Intent Oracle — operating simultaneously.

316
Protocols
1,338
Formulas
542
Algorithms
174
US Patents
785
MOE Experts
19
Neural Shards