AI can do impressive things, but it cannot compensate for missing structure. In RAG, agentic coding, and AI music workflows, input quality still determines output quality....
Read more
A web GUI with authentication, day/night theme support, and NetBSD cross-build for 40 EUR in token costs sounds like the end of classic development. This article shows why the opposite is true: what gets devalued is boilerplate, not architecture, accountability, and real system quality....
Read more
This article explains in a structured and practical way how to build a local, auditable knowledge pipeline for LLMs: Markdown → Chunking → Embeddings → SQLite → Retrieval → Prompt → llama-server. A minimalistic, reproducible, cloud-free, blackbox-free RAG implementation....
Read more
Ollama is convenient, but abstracts a lot of details. This article shows step by step how to control the entire LLM supply chain yourself - from your own llama.cpp build to transparent GGUF models to the lean local API....
Read more
A Practical Attempt to Run a Large Language Model Independent of Cloud Services – with Ollama as an open runtime environment and mixed feelings between technical fascination and physical reality....
Read more
Introduction Lunaria is not a framework in the classic sense, but a practical learning and prototyping system....
Read more