Working references from Certance Advisory: what we measure, what it costs, and what to change first. Every figure traces to a primary source.
October 2026 · Quality engineering
What a passing pipeline proves, what it cannot, and why the difference decides whether AI-written code deserves trust.
Read the article
September 2026 · DORA
DORA's first supervisory year turns on evidence. What Articles 24 and 25 ask for, where automated test suites go blind, and how to map coverage to business risk.
Read the article
October 2026 · The Standard
Edition 2026.6: fourteen chapters, from the five merge gates to testing AI inside the product, and the twelve statements, for teams where AI agents write most of the code.
Read the Standard
Cut what your coding agents spend, keep the quality, and make the bill predictable on GitHub Copilot and Claude Code: shorter agent sessions, a warm prompt cache, filtered tool output and the right model for each task.
Adoption climbs, acceptance rates look healthy, and the board asks what the licences return. The four findings that recur in almost every team we assess, why they happen, and the seven dimensions that measure them.
New pieces go to subscribers first, together with the full AI Token Efficiency Guide.
Subscribe