Introduction: The Ground Has Shifted A little under a year ago, we published a detailed field guide on building a computer vision system for object identification, counting, and ERP integration. The core message of that post still holds: getting from "the AI sees an object" to "the record in your system is correct" is a journey full of non-obvious problems — duplication, occlusion, lighting drift, object similarity, and knowing when to trust the machine. What has changed in t

Software testing has always been the part of the development cycle that everyone agrees is important and almost everyone underinvests in. The reasons are structural. Writing good tests is time-consuming. Maintaining test suites as code changes is even more so. QA engineers are perpetually under-resourced relative to the volume of work they are expected to validate. And under deadline pressure, testing is the discipline that gets compressed first — with consequences that typic

Most engineering teams that are serious about AI have already adopted AI coding tools. Some have restructured their code review process. A smaller number have started experimenting with autonomous AI agents for bounded tasks. Very few have answered the harder question: what does the whole team actually look like when AI is a first-class member of the delivery process — not an add-on tool, but a structural part of how work gets planned, built, tested, and shipped? That is the