Clarus News – Architecture at a glance
A scalable, sovereign, and AI-powered media platform
Executive Summary
Clarus News is a modular, AI-assisted news and analysis platform that combines editorial quality with automation. The architecture has been designed to remain scalable, adaptable, and vendor-independent – particularly in the field of artificial intelligence.
The core idea: AI is a tool – not the owner of content or processes.
Architecture principles (What sets us apart)
1. Clear separation of responsibilities
The platform follows a clean separation between:
- Editorial & business logic
- AI-powered support
- Data storage & search
This ensures editorial control always remains with the operator – not with the model.
2. Multi-model strategy instead of AI monoculture
Clarus News deliberately doesn't rely on a single AI model. Depending on the task, different models are deployed:
- fast, cost-effective models for routine tasks
- powerful models for analysis, research, and complex texts
- on request, private models for sensitive content
Result: cost control, quality management, and technological independence.
3. MCP – Controlled AI integration
AI integration is done via the Model Context Protocol (MCP). This means:
- AI only receives explicitly authorized context
- System actions (e.g., database changes) are clearly separated from AI reasoning
- Workflows are reproducible, auditable, and configurable
👉 Particularly relevant for governance, compliance, and data protection.
4. Data stays in the system
The platform uses PostgreSQL as its central data backbone:
- Content
- Metadata
- Search indexes
- Crawler results
Full-text search is directly integrated into the database. No external search infrastructure, no dependencies.
Technical structure (simplified)
Web application
Server-side rendered, fast, SEO-friendly
→ Presentation, navigation, editorial interface
Automated content pipeline
Background processes for:
- Research & crawling
- Text processing
- Translation & metadata
- AI-powered analysis
AI layer (MCP)
Unified control of external and internal AI models
Database & search
Centralized, performant, and controllable data storage
Clarus News – Technical overview & architecture
Overview
Clarus News is a multilingual, modular news and analysis platform with integrated editorial curation logic and AI-assisted content creation.
Core components
- ASP.NET (.NET) web app (Razor Views)
- PostgreSQL with full-text search (tsvector)
- CLI-based content pipeline (crawler & prompt jobs)
- LLM integration via the Model Context Protocol (MCP)
MCP & private LLMs
With MCP, you can connect both external LLM providers and private, self-hosted models. This enables maximum control over data, cost, and compliance.
Benefits
- Scalable content creation
- Multilingual support (DE / FR / EN)
- Technological sovereignty
Strategic advantages for decision-makers
✅ Cost control
- AI deployment where it brings real value
- Models interchangeable based on budget & requirements
✅ Independence
- No lock-in with individual AI providers
- Option for private LLM operation (on-premise / private cloud)
✅ Scalability
- More content ≠ linearly more staff
- Automation relieves the editorial team, doesn't replace it
✅ Governance & trust
- Clear rules, reproducible processes
- AI as assistance, not as an uncontrolled black box
In short
Clarus News is not an experimental AI playground, but a productive media platform with:
- controlled AI integration
- long-term technological sovereignty
- clear editorial responsibility
The architecture is designed to remain viable in five years – regardless of which AI model is currently "state of the art."