The AI Documentation Readiness System: Grade Any AI Product Before a Regulator Does

In August, that stops being a UX problem and becomes a compliance one.

The EU AI Act's Article 50 transparency obligations apply from August 2, 2026. Users must be told when they are interacting with an AI system, and synthetic content must be marked as such. California's training data transparency law has been in effect since January, and Colorado's replacement law lands January 1, 2027. Every one of them asks for documentation most teams have not written, because nobody ever told them what "good enough" looks like.

I have spent seven months grading real AI companies on exactly this, in public, every Wednesday, in The Documentation Engineer. This is the instrument I grade them with.

What you are buying is not a template pack. It is a scored diagnostic.

You open the product you are auditing. You use it the way a first-time user would. You answer the questions. The workbook does the rest: it scores the seven-question audit, weights the four ADECP categories into a letter grade, maps your gaps against three jurisdictions, and builds a remediation plan that shows you only the actions your own scores triggered. No flat checklist. Nothing you did not earn.

Inside the workbook (7 tabs, 90 live formulas, zero manual math):

• The 7-Question Audit. Score each question 0 to 2. Evidence is required: a score you cannot cite is a guess. Auto-bands into one of four verdicts, from Ethically Documented to Documentation Gap.

• The ADECP Scorecard. Four categories, four criteria each. Ethical Stakes, Accessibility, Positioning, Execution, weighted to a single grade.

• Regulatory Readiness. Eight binary checks across the EU AI Act, California, and Colorado. Auto-computes your gap count and a risk rating.

• The Limitation Builder. Scores an existing limitation statement against the Specific + Contextual + Action-Oriented formula, then rewrites it using eight product-type patterns, from code generation to agentic systems.

• The Remediation Plan. Eight prioritized actions, each surfacing only when your scores call for it, ranked by user exposure rather than by effort.

• The Summary Scorecard. The one page you screenshot and bring to your team. Grade, score, band, risk rating, and a one-sentence finding in plain English.

Who this is for:

• Documentation engineers and DevRel professionals who need a standard, not another opinion.

• Founders and product leads who need to know their exposure before a customer or a regulator finds it.

• Anyone moving into technical writing or AI governance who wants to run a real audit and have the artifact to prove it.

What you will have in 90 minutes: a defensible grade for your AI product's documentation, a mapped list of your regulatory gaps, and a prioritized plan to close them, written the way I would write it for a $2,500 audit.

The documentation gap is the ethics gap. These are the same problem. This is the tool that finds yours.