The Platform
Every component is in production on enterprise-scale datasets. It's not a prototype: it's a proven system.
Four proprietary components
OSINT Engine
Universal data collection engine, configurable on any domain: regulations, competitors, markets, territory, suppliers, tenders.
- 4 integrated search engines with fallback
- Crawling on target sites with IPv6 rotation
- Change detection: identifies when a source updates
- Multi-format parsing: HTML, PDF, DOCX, XLSX, JSON, XML
- Configurable on any vertical domain
RAG Engine
End-to-end Retrieval-Augmented Generation pipeline with hybrid search and multi-collection support.
- Semantic chunking with configurable overlap
- Vector embeddings (open-weight models or APIs)
- Hybrid search: semantic ranking + full-text (Meilisearch)
- Multi-collection for separate domains
- Support for multiple source types
Knowledge Graph
Multi-type relational graph, battle-tested in production on enterprise-scale datasets.
- Typed nodes and edges
- N-hop search (2, 3, 4-hop relationships)
- Path finding between entities
- Interactive drill-down from graph to source documents
- Scales to tens of thousands of nodes
Explorer & Dashboard
Interactive views and customizable dashboards with integrated AI chat.
- WebGL graph rendering for large-scale graphs
- Thematic views: heatmap, timeline, Sankey, treemap
- Domain-specific dashboards
- Integrated AI chat with tool use on Knowledge Graph + RAG
- Radar and comparative analysis
Three innovations that set us apart
Proprietary MCP Server
We build a custom MCP Server that exposes your data, your tools and your processes to any AI — while keeping full control over what goes in and what comes out. The AI becomes an intelligent interface sitting on top of your systems.
Why it matters: MCP is the emerging standard for AI to communicate with enterprise tools. Having your own MCP Server means you decide how the AI accesses data, tools and processes: you're not tied to a third-party platform that changes the rules whenever it wants.
Intelligent Anonymization
Anonymization and pseudo-anonymization layer integrated into the MCP. Your sensitive data is replaced with consistent aliases before leaving the perimeter, and automatically restored in the result. GDPR-compliant by design.
Why it matters: You can use external AI (Claude, GPT) even on your most sensitive data without ever truly exposing it. Critical if you operate in regulated sectors or in Public Administration: data stays in, AI answers come out.
Intelligent AI Routing
The system automatically decides whether to use the internal AI (on-premise, zero cost) or market AI (Claude, GPT) based on task type, data sensitivity and complexity. Maximum power, minimum risk, optimized costs.
Why it matters: You don't have to choose between on-premise and cloud: the system picks the best option case by case. You get top performance on complex tasks and protect data where it really matters.
Open stack, zero lock-in
PostgreSQL, Meilisearch, Ollama, open-weight models. Zero vendor lock-in. If tomorrow you want to manage everything in-house, you can — no strings attached.
€0
vendor lock-in
-70%
vs API costs after year one
∞
scalability: decreasing cost per query
Let's start here
Every transformation starts with a conversation.
Tell us your challenge: together we'll find the right solution.
