Writings on AI governance, software delivery, and the future of engineering decision-making.
We deployed Decision Diffs with our first pilot teams. Here's what we measured, what surprised us, and what we're changing based on real usage.
The U.S. spends $1.7 trillion annually on software and IT. Up to 30% of that is rework from scope miscommunication. Here's why this is a governance problem, not a process problem.
Leading AI engineering at Inspectorio, I saw firsthand how AI transforms supply chain visibility — and why the hardest problems aren't technical.
Most LLM applications generate plausible text. For governance and compliance use cases, plausible isn't enough — you need verifiable citations back to source evidence.
AI coding assistants are making teams ship faster than ever. But speed without governance just means you arrive at the wrong destination sooner.
What running engineering at Vietnam's largest e-commerce platform taught me about the real bottleneck in software delivery — and why I'm now building governance infrastructure for AI-assisted teams.
Every engineering team has felt it: a "small tweak" from product that cascades into missed deadlines, rework, and finger-pointing. The problem isn't the change itself — it's that changes travel through informal channels with no audit trail, no impact analysis, and no accountability.
What if every meaningful scope change in your software project produced an auditable artifact — with citations, required approvals, impact analysis, and automatic propagation? That's the Decision Diff.