Quality Assurance Process
Drive continuous improvement in quality of products or services.
Principles
Quality control is a reactive part of quality management focused on detecting problems after they have occurred. It involves activities such as testing, measuring, and inspecting to ensure specified quality standards are met.
Quality assurance is a proactive process that focuses on preventing defects from occurring in the first place.
Deming's Checklist
14 Point checklist to incorporate quality control principles into the manufacturing process. With a focus on creating systems that got manufacturing right the first time.
- Create constancy of purpose for improving products and services.
- Adopt the new philosophy.
- Cease dependence on inspection to achieve quality.
- End the practice of awarding business on price alone; instead, minimize total cost by working with a single supplier.
- Improve constantly and forever every process for planning, production and service.
- Institute training on the job.
- Adopt and institute leadership.
- Drive out fear.
- Break down barriers between staff areas.
- Eliminate slogans, exhortations and targets for the workforce.
- Eliminate numerical quotas for the workforce and numerical goals for management.
- Remove barriers that rob people of pride of workmanship, and eliminate the annual rating or merit system.
- Institute a vigorous program of education and self-improvement for everyone.
- Put everybody in the company to work accomplishing the transformation.
Generator QA
Deming's Point 3 — "cease dependence on inspection" — applies to automated output too. A generator that produces code, content, or artifacts must pass the same gates as hand-crafted work. If a human commit would fail the pre-commit hook, a generated commit must fail the same hook.
| Principle | What it means |
|---|---|
| Same gates | Generated output runs through identical linting, validation, and hook checks as manual output |
| Fail fast | If the generator can't pass the gates, the template driving the generator is broken — fix the template, not the gate |
| No inspection bypass | "It works at runtime" is not "it ships." The gap between runtime success and commit success is where defects hide |
| Instrument calibration | The tools that measure quality must themselves be measured. A broken issue tracker means the feedback loop is open |
The test: run lint-staged simulation on generated files. If any file fails, the generator template is at Process Maturity level D (draft), not A (approved).
Context
- Process Optimisation — The improvement loop these principles feed
- Standard Templates — Template readiness gates enforce these principles upstream
- Control System — PID mechanics: the gauge must work before the controller can correct
Questions
When your generator passes all runtime tests but fails at commit time, where is the gap — in the generator, the gates, or the assumption that they're equivalent?
- What percentage of your automated output has never been tested against the same hooks that gate manual work?
- If the tool that logs defects is itself defective, how do you close the feedback loop?
- Which of Deming's 14 points is hardest to apply to AI-generated output — and what does that tell you about your process?