The Evolution Towards a Cognitive and Semantic Architecture
A major shift towards dynamic ontologies, AI state management, and asynchronous orchestration.
Made in Morocco — 26 Patent Claims
Sovereign Decision Intelligence
The first 100% on-premise AI platform designed for enterprises and governments who refuse to compromise on data sovereignty. 99.9% precision via Certified Augmented Generation � your data never leaves your premises.
Scroll to explore
7 Intelligent Agents
Enterprise AI Without Compromise
Six specialized AI agents working in orchestration to deliver secure, explainable, and sovereign artificial intelligence.
Brain Agent
Knowledge graph and semantic search engine. Organize and query your enterprise knowledge with RDF/SPARQL.
- Knowledge graph (Jena TDB2)
- Semantic search & RAG
- RDF/SPARQL queries
- Vector embeddings
Brain-Ingestion Agent
Intelligent knowledge extraction from any source. Transform documents into structured knowledge automatically.
- PDF, Excel, Word extraction
- LLM-powered RDF generation
- MCP & API connectors
- Provenance tracking
Shield Agent
Real-time compliance and security validation. Every AI interaction is checked against your policies.
- PII detection (Presidio)
- GDPR/CNDP compliance
- Prompt injection defense
- Data classification
Vault Agent
Access control and encryption management. Zanzibar-style authorization with cryptographic security.
- ReBAC (OpenFGA)
- End-to-end encryption
- Immutable audit logs
- Key management
Refinery Agent
Quality validation and expert review. Ensure AI outputs meet your organization's standards.
- Output validation
- Expert review workflows
- Quality scoring
- Feedback loops
Connect Agent
MCP-native integration hub. Connect AI to any data source using Model Context Protocol standards.
- MCP server (stdio/SSE/HTTP)
- Database connectors (PostgreSQL)
- REST/GraphQL APIs
- Real-time SSE streaming
Semantic Builder Agent
Automated Schema.org and structured data generation. Enhance SEO and AI discoverability with semantic markup.
- Schema.org JSON-LD generation
- Rich snippets optimization
- Knowledge graph integration
- AI-ready structured data
Unified Data Governance Hub
ArcaQ recommends OpenMetadata as the unified data governance platform. Seamlessly integrate metadata management, data discovery, and lineage tracking with sovereign AI capabilities.
Talentys, ArcaQ's parent company, is an OpenMetadata Center of Excellence and a reference in its integration.
Metadata Management
Data Discovery
Lineage Tracking
Quality Monitoring
AI Stack Transparency
100% Open-Source LLMArcaQ does NOT develop proprietary LLM models. We orchestrate open-source language models via Ollama for complete sovereignty and transparency. The LLM is only used for formatting responses, not for generating answers.
What We Use
- Ollama - Local LLM runtime (100% on-premise)
- Open-source models - Qwen 2.5, LLaMA 3.2, Mistral, nomic-embed-text
- CPU-only - Intel AMX / AMD AVX-512 optimized (no GPU required)
- Apache Jena - Knowledge Graph (RDF/SPARQL/OWL)
- OpenMetadata - Data governance and catalog
What We DON'T Do
- No proprietary LLM - We don't train or develop our own models
- No cloud API calls - Zero dependency on OpenAI/Anthropic/Google
- No creative generation - LLM only formats certified Knowledge Graph data
- Not a ChatGPT competitor - Complementary, decision-focused platform
CAG Architecture: How Responses Are Generated
Knowledge Graph
Apache Jena TDB2
SPARQL Query
Certified facts retrieval
Verified Data
Pre-certified results
Ollama LLM
Formatting only
Human Response
0.1% hallucination
The LLM cannot invent facts. If SPARQL returns empty "Information not available"
Patented Architecture — Claims 1-4
Certified Augmented Generation
The industry evolved from RAG to Context Augmented Generation. ArcaQ goes further — our Certified Augmented Generation adds a governance & provability layer where every fact is mathematically traceable to its source.
Industry Standard
RAG
15-20% HallucinationRetrieval Augmented Generation
1
User query
2
Vector search (approximate)
3
LLM interprets & generates
4
Response (may hallucinate)
- Approximate vector similarity search
- LLM invents facts beyond context
- No auditability or data lineage
Industry Trend 2024
CAG Context
5-10% HallucinationContext Augmented Generation
1
User query
2
Full corpus in long context
3
LLM still interprets freely
4
Better — but still hallucinates
- No vector DB — simpler architecture
- LLM still interprets — can still fabricate
- Limited by context window (2M tokens max)
ArcaQ
ArcaQ � Patented
CAG Certified Patented
0.1% HallucinationCertified Augmented Generation
1
User query
2
SPARQL on Knowledge Graph
3
LLM formats only — never invents
4
Certified & auditable response
- Formal verification — every fact traceable
- LLM formats certified facts — 0% fabrication
- SCAG: 4 Semantic Control & Governance Layers
99.9%
Semantic Precision
vs 80-85% for standard RAG
0%
Fact Fabrication
LLM cannot invent — only format
100%
Auditable Responses
Full lineage from source to output
4
SCAG Filter Layers
Legal — Hierarchical — Cultural — Intangible
While the industry adopts Context Augmented Generation, ArcaQ's Certified Augmented Generation — protected by Patent Claims 1-4 (OMPIC, USPTO, PCT, EPO) — is the only architecture that mathematically proves every fact.
See CAG in Action
Why Sovereignty Matters
Your Data. Your Rules. Your Infrastructure.
In an era where AI requires massive data processing, the question isn't just "What can AI do?" but "Where does my data go?"
Zero Cloud Dependency
No data ever leaves your data center. Full air-gap deployment available.
Jurisdictional Control
Your data stays under your legal jurisdiction. No foreign government access.
You Own Everything
Models, data, outputs - complete ownership with no vendor lock-in.
Regulatory Compliance
Meet GDPR, HIPAA, local banking regulations, and government security requirements.
Cloud AI
Data sent to external servers
Subject to foreign laws
Vendor controls your AI
Internet required
ArcaQ Sovereign AI
Data stays on-premise
Your jurisdiction only
Complete ownership
Air-gap capable
Intellectual Property & Compliance
26 Patent Claims — 63 Jurisdictions
ArcaQ's technology is protected by international patents filed under PCT (Patent Cooperation Treaty) covering 157 countries. Our compliance engine automatically handles 63 data protection regulations across 5 continents.
26
Patent Claims
63
Compliance Jurisdictions
5
Continents
Europe
18 countries
Africa
12 countries
Asia-Pacific
14 countries
Middle East
8 countries
Americas
10 countries
Patent Filings
OMPIC Morocco
PCT International
USPTO
EPO Europe
Global Coverage
Understanding ArcaQ How is ArcaQ Different?
Clarifying our position in the modern data & AI stack
ArcaQ vs DataOS: What's the difference?
DataOS
Focus: Data orchestration & governance
Purpose: Unify data silos, create data mesh/fabric
Core: Data engineering, cataloging, ETL/ELT pipelines
Output: Accessible, governed data products
ArcaQ
Focus: AI agent orchestration for intelligent decisions
Purpose: Transform data into actionable decisions via sovereign AI
Core: 7 specialized AI agents (Brain, Connect, Shield, Vault, Refinery, Ingestion, Semantic Builder)
Output: Automated insights, recommendations, actions
Complementary positioning: DataOS prepares & governs your data ? ArcaQ exploits that data via sovereign AI to drive decisions.
Read the full comparisonArcaQ vs Palantir: What's the difference?
Palantir
Focus: Defense & government data fusion
Deployment: Cloud-first, US-based infrastructure
Cost: $10M+ annual contracts, vendor lock-in
ArcaQ
Focus: Enterprise decision intelligence for all sectors
Deployment: 100% on-premise, absolute data sovereignty
Cost: Transparent pricing, no vendor lock-in
Key difference: ArcaQ offers true data sovereignty with 100% on-premise deployment, while Palantir requires cloud infrastructure and US oversight.
Read the full comparisonWhat makes ArcaQ unique?
- Agentic AI: Autonomous agents that reason and act independently
- Multi-LLM Orchestration: Ollama, OpenAI, Claude, Mistral - choose your sovereign mix
- 26 Patent Claims: Unique IP on decision intelligence & sovereignty
- 100% On-Premise: Absolute sovereignty - your data never leaves (vs cloud alternatives)
- OpenMetadata Integration: Talentys (ArcaQ's parent) is a Center of Excellence
Where does ArcaQ fit in my data stack?
1
Data Infrastructure
Databases, data lakes, warehouses
2
Data Governance
OpenMetadata, Collate, DataOS
3
Decision Intelligence (ArcaQ)
AI agents analyzing, reasoning, recommending
4
Business Applications
Dashboards, apps, automated workflows
Real-world scenario: Multi-department decision making?
Scenario: Strategic M&A Decision in 48 hours
A CFO needs to evaluate an acquisition target. Traditionally, this takes 2-3 weeks. With ArcaQ's sovereign decision intelligence:
T+0h
Query Launch
CFO asks: "What's the strategic value of acquiring Company X?" ArcaQ Brain Agent orchestrates 7 specialized agents.
T+2h
Multi-Department Data Collection
- Finance: Historical P&L, cash flow projections (via Connect Agent ? ERP/SAP)
- HR: Talent overlap, culture fit analysis (via Knowledge Agent ? HRIS)
- Operations: Supply chain synergies, warehouse capacity (via Engineer Agent ? MES/WMS)
- IT: Tech stack compatibility, security posture (via Shield Agent ? CMDB)
- Legal: Regulatory compliance, IP portfolio (via Meta Agent ? DMS)
T+24h
Collective Knowledge Synthesis
Brain Agent synthesizes 127 data sources into a unified knowledge graph. Identifies 23 strategic synergies, 8 integration risks, 4 regulatory blockers.
T+48h
Actionable Decision Package
CFO receives: valuation range (€18-24M), integration roadmap (12 months), risk mitigation plan, board presentation deck. Decision made in 48h vs 3 weeks traditionally.
18 days
Time saved
127
Data sources unified
100%
Cross-dept. completeness
On-Premise
100% Sovereign
What LLM does ArcaQ use? Do you have your own AI model?
Reality
NO, ArcaQ does NOT develop its own LLM. We use Ollama to run open-source models (Qwen 2.5, LLaMA 3.2, Mistral, nomic-embed-text) 100% locally on your infrastructure.
Why this matters: Our innovation is NOT in the LLM itself, but in the CAG (Certified Augmented Generation) architecture that eliminates hallucinations. The LLM only formats pre-verified facts from the Knowledge Graph - it cannot invent information.
- ? 100% on-premise - Models run locally via Ollama, zero cloud API calls
- ? CPU-only supported - Intel AMX / AMD AVX-512 optimized, no expensive GPU required
- ? Model agnostic - Swap LLMs freely (Mistral, LLaMA, Qwen, etc.)
- ? Open-source based - No vendor lock-in, full transparency
Is ArcaQ a ChatGPT competitor? What's the difference?
ChatGPT / Claude
Purpose: General-purpose conversational AI
Generation: Probabilistic text generation (can hallucinate)
Data: Trained on public internet data, cloud-based
Best for: Creative writing, brainstorming, general Q&A
ArcaQ
Purpose: Enterprise decision intelligence on YOUR data
Generation: Certified facts from Knowledge Graph (0.1% hallucination)
Data: Your internal data only, 100% on-premise
Best for: Critical business decisions, compliance, auditable AI
Key insight: ArcaQ and ChatGPT are complementary, not competing. ChatGPT excels at creative tasks. ArcaQ excels at decision-making with certified internal data.
Does ArcaQ require expensive GPUs?
No. ArcaQ is optimized for CPU-only deployment using Intel AMX (Advanced Matrix Extensions) and AMD AVX-512 instructions. This enables air-gapped deployment without GPU vendor dependencies.
- ??? CPU-only mode: Works on standard enterprise servers (Intel Xeon, AMD EPYC)
- ? GPU acceleration (optional): NVIDIA T4/A10 for 3-5x faster embedding generation
- ?? Cost savings: No expensive GPU infrastructure required for production
- ?? Air-gap friendly: No cloud GPU dependencies for isolated environments
How does the Semantic Builder work?
INPUT: Multi-Source Ingestion
Documents (PDF, Word, Excel), Databases (SQL, NoSQL), APIs (REST, GraphQL, MCP)
PROCESS: AI-Powered Semantic Construction
- Ontology Construction: Automatically builds domain-specific schemas with classes, properties, and relationships (OWL/RDFS)
- Semantic Extraction: LLM-powered Named Entity Recognition and Natural Language Processing to identify entities, concepts, and contextual relationships
- RDF/OWL Mapping: Converts extracted knowledge into RDF triples with automated reasoning and inference using Jena TDB2 engine
OUTPUT: Actionable Intelligence
- Knowledge Graph: Sovereign triple store (Jena TDB2) with full ontology and instance data
- SPARQL Endpoint: Real-time query interface with reasoning capabilities for the 7 AI agents
- AI Agent Integration: Direct access for Brain, Connect, Shield, Refinery, Vault, Ingestion, and Semantic Builder agents
Why Semantic Builder matters:
- 100% On-Premise: Your knowledge graph never leaves your infrastructure
- Automated Reasoning: Infers new knowledge from existing relationships
- Multi-Modal: Handles structured, semi-structured, and unstructured data
- Real-time Updates: Continuous ingestion and knowledge graph evolution
Leadership
Meet the Team Behind ArcaQ
Built in Morocco, designed for global impact
CEO
Nabila El Hadi
Strategic Data Governance Leader
Driving data reliability, compliance and enterprise governance across EMEA.
15+
Years
EMEA
Region
CIO
Mustapha Fonsau
Senior Data Architect | ArcaQ Inventor
OpenMetadata SME | Data Strategy, AI Governance & Platforms | Sovereign Decision Intelligence
18+
Years
26
Patents
Join Us
Shape the Future of Sovereign AI
We're building the next generation of decision intelligence. Join us in Rabat, Paris, or remotely.
GTM Specialist - Sovereign Decision Intelligence
The Vision: Arcaq & Sovereign Decision Intelligence
In an era of AI and global data dependency, Sovereign Decision Intelligence (SDI) is the new frontier. Arcaq is more than a data tool; it is a platform designed to give organizations full control over their data assets and decision-making processes.
By integrating with the OpenMetadata and Collate ecosystem, Arcaq bridges the gap between technical metadata and strategic sovereignty. We are looking for a GTM Specialist who understands that data is the ultimate instrument of power and autonomy for modern enterprises.
Your Mission: Evangelist, Strategist & Connector
- Community Leadership: Drive ArcaQ's presence within the OpenMetadata global community. Engage with contributors and users to show how SDI completes the modern data stack.
- Thought Leadership: Create and publish high-level insights on why Sovereignty in data is critical for the next generation of CEOs and CDOs.
- Ecosystem Growth: Identify and build partnerships with stakeholders in Europe and Africa interested in Collate and SDI solutions.
- Market Intelligence: Analyze the competitive landscape and identify "sovereignty gaps" in existing solutions that ArcaQ can fill.
- Lead Generation: Convert digital influence and community interactions into high-value Proof of Concepts (POCs).
Determinant Factor: Personal Branding & LinkedIn Presence
For this role, your digital footprint is your resume. We are looking for a "Voice" in the tech space.
- LinkedIn Mastery: Your profile must be professional, active, and reflect a genuine interest in Tech, Data, or Business Strategy.
- Self-Branding: You have a demonstrated ability to curate content, share opinions, and build an audience.
- Advocacy: You are comfortable being the "face" of a solution and engaging in public-facing digital discussions.
Candidate Profile
- Background: Final year at a Top-tier Business or Engineering School (Bac+5).
- Languages: Fluent English is a must (C1/C2 level) for international community engagement. Perfect French is required.
- Mindset: A "Consultant" mindset — able to understand complex technical architectures and translate them into business value.
- Remote Ready: You are disciplined, proactive, and thrive in a 100% remote environment.
- MDM Spirit: We highly value the global perspective of the Moroccan Diaspora and their ability to bridge international markets.
Why ArcaQ?
- Pioneer the SDI Movement: Work on one of the most strategic topics in tech today.
- Direct Path to CDI: Successful interns will be fast-tracked into GTM Manager or Partnership Lead roles.
- Full Remote: Work from Paris or Rabat with total flexibility and a result-oriented culture.
Ready to define the future of Sovereign Tech?
Apply Now - jobs@talentys.eu
Mandatory:
Include your LinkedIn profile URL.
Subject line: GTM ArcaQ SDI - [Rabat/Paris] - [Your Name]
Applications without a LinkedIn profile link will not be considered.
Get Started
Request Your Private Demo
See how ArcaQ can bring sovereign AI to your organization. Our team will prepare a customized demonstration for your industry and use case.
Email
contact@talentys.eu
Location
Morocco — Serving the World
Join the Sovereign AI Revolution
Partner with ArcaQ to bring sovereign decision intelligence to Africa and beyond.
Rabat, Morocco
Meet us at GITEX Africa 2026 — April 7-9 — Marrakech
Settings
Language