From demo to accuracy infrastructure
A series of eight articles on building RAG systems that work reliably on real enterprise data. Read the preface first — it explains what the series is for and how to use it as a diagnostic tool rather than a checklist.
Why a RAG demo succeeds and a production system does not, what this series covers, who it is for, and how to read it as a diagnostic tool rather than a checklist.
4 min · Updated June 2026
Read the preface→What exactly breaks, and why, when you scale a naive RAG system to real enterprise data?
The structural problem with LLM hallucination, how first-generation RAG responds, and the five specific ways it fails at scale — each with the symptom you'll see in your own logs and the tooling that fixes it.
7 min · Updated July 2026
Read this part→What do those four words actually mean, which problem does each one solve, and which one should you reach for first?
Four independent concepts, each solving a distinct failure mode — with the one query each was invented to answer, minimal code for each, and a worked RRF example.
9 min · Updated July 2026
Read this part→What are the specific, named failure modes across ingestion, retrieval, and generation that you need to design against?
A precise, reader-first taxonomy of RAG failure modes across ingestion, retrieval (F1–F8), generation, and operations — each named as the moment you’ve actually lived, with detection code and real cost numbers.
8 min · Updated July 2026
Read this part→How do you organise the system so that each concern has a clear owner, a defined boundary, and a deployable footprint — and what actually happens to one query as it moves through them?
The seven-plane architecture — control, observability, ingestion, tooling, indexing, retrieval, and agentic — with what each plane owns, when you'd touch it, and one real query traced through all seven.
8 min · Updated July 2026
Read this part→How do you parse every document format faithfully, and how do you know when parsing has failed?
Modality-specific parsing, the ten ingestion accuracy patterns grouped by lifecycle stage, how Docling/embeddings/VLMs collaborate, confidence-gated review, and the ingestion observability model — with runnable code for enrichment, embedding validation, and quarantining.
15 min · Updated July 2026
Read this part→What are the eight ways retrieval silently returns the wrong answer, and what is the specific pattern that kills each one?
The eight ways retrieval silently returns the wrong answer — each written as the symptom you'll see — and the nine patterns that kill them, with runnable code for rerank, adaptive routing, and context reordering.
11 min · Updated July 2026
Read this part→How does an agent self-correct after retrieval and generation, and how does the system get more accurate the longer it runs in production?
Why agents earn their cost, seven agentic patterns framed as the problems they solve (router, CRAG, Self-RAG, decomposition, citation, bounded budget, tool fallback), the full LangGraph structure, and the compound accuracy flywheel.
13 min · Updated July 2026
Read this part→What does the confirmed production stack look like, what was deliberately traded away, and what separates a production system from a demo?
The confirmed production stack with every tradeoff stated explicitly, the consequential choices explained, the open-source tool reference (version-checked July 2026), key empirical claims with attribution, and the five things that separate a production system from a demo.
13 min · Updated July 2026
Read this part→