NEURIXA Handbook
The knowledge behind NEURIXA
The NEURIXA Handbook consolidates six months of research, architecture design, operational experience and structured analysis behind the platform.
The complete handbook contains 548 pages, 32 chapters and approximately 90,000 words. It documents the principles, methods and technical reasoning developed during the creation of NEURIXA.
The complete handbook is an internal knowledge asset of IT Business Consulting Magyarország Ltd. It is not published or distributed in full.
What the handbook covers
The handbook examines how complex organisational and operational information can be processed, connected, retrieved and used responsibly.
Its subjects include:
- document processing and structured knowledge;
- chunking, embeddings and vector-based retrieval;
- retrieval-augmented generation and language models;
- source traceability and auditable answer generation;
- network visibility and controlled AI access;
- security-event analysis and investigation support;
- energy decision support and energy security;
- operational resilience and human-controlled workflows.
The handbook distinguishes clearly between model training, fine-tuning and RAG-based knowledge use. It also explains that a vector database does not convert a vector back into a finished answer: it retrieves relevant original source passages and structured facts, while the language model generates the response token by token from the retrieved context.
What we publish
This page provides selected public extracts and high-level explanations from the NEURIXA Handbook.
The published materials present carefully chosen principles, concepts and operational lessons in an accessible form. They are intended to support professional understanding and informed discussion.
They do not reproduce complete chapters, disclose system configurations or provide the detailed implementation knowledge required to reconstruct the platform.
