Analysis

Longer-form technical writing for product managers and software leaders — white papers, design papers, and strategy papers drawn from production work. Each paper distills a hard, ambiguous problem into a framework you can apply: how to decide, what to measure, which tradeoffs are real, and which sophistication the literature overrates. The companion summaries below capture the argument; the PDFs carry the detail.


White Paper · April 2026

Evaluating RAG Systems

A Product Manager’s Practical Guide to Measuring Retrieval Quality

A framework for evaluating retrieval quality without a PhD in information retrieval: how to build a ground-truth test set, which metrics actually matter for tuning decisions (and which just detect regressions), and how to structure a configuration sweep that isolates one variable at a time. Drawn from a production RAG system where this approach lifted retrieval precision by 85% and cut per-query token cost by 65%.

White Paper · April 2026

Tuning RAG Systems for Production

A Decision Framework for Eight Optimization Options and One Production-UX Pattern

Eleven controlled experiments against a production RAG deployment over 85,351 sections of U.S. federal regulations. Which “best practices” move the needle, which are neutral, and which are actively harmful — plus an inference-time confidence signal that reframes reliability in high-stakes domains. Spoiler: only two of eight optimizations produced unambiguous wins, and the sophisticated ones clustered near zero.