AI Guardrails & Quality Checks

A "Check Quality" button on Requirements and every DRP module, checking real, specific issues tailored to that content — informational, never blocking.

What this is, and what it isn't

AI Suggestions (see that guide topic) drafts and improves content. Quality checks do the opposite job: they validate content you already have, checking it against a set of specific, named dimensions and explaining exactly why each one matters — never rewriting anything themselves.

A quality check is purely informational. It never blocks saving, never blocks marking a module complete, and never runs automatically — you decide when to check, by clicking the button.

Where to find it

Click "Check Quality" on any DRP module's editor page, next to "Get AI Suggestions":

  • Every module is checked against its own specific dimensions. See the DRP Module Reference for what each module actually checks and why. It also compares the module with every other section of the same DRP and lists, separately, anything the two cannot both be true about — e.g. a module requiring something that Solution Scope rules out — naming the other section.
  • Requirements is the one module checked on two levels: its summary, and the requirements linked to this DRP — each one's clarity, completeness, atomicity and testability, plus a likely duplicate or conflict elsewhere in the project and any conflict with something already stated in another section of the same DRP, such as a requirement asking for something Solution Scope explicitly excludes.

Results appear in the same place: a green "meets quality standards" box if nothing was found, or a card per issue naming what failed and why, plus a list of suggested fixes — you apply them yourself, the same as any other AI suggestion. A "Conflicts with other sections" list follows, with one card per conflict naming the section it disagrees with.

Why every module has its own checks

Requirements guardrails check atomicity and testability because those dimensions genuinely mean something for a requirement. They mean nothing for a RAID log, a Cover & Version Control page, or a Traceability Matrix — so instead of applying one generic formula everywhere, every module in Requio has its own specific checks, tailored to what that module actually contains.

A few examples: the RAID Log is checked for real, specific mitigations rather than placeholders like "monitor closely". Stakeholders & RACI is checked for exactly one Accountable per area of work — the single most common RACI failure mode. Traceability Matrix and Test Cases are checked against the real requirements linked to this DRP, so a fabricated Requirement ID gets caught, not just a vague "looks incomplete" verdict.

The full list of what each module checks — and why it matters — is on that module's own guide page, in its "Quality checks" section.

Plan availability

Quality checks — for Requirements and for every DRP module — require the Standard plan or above. On Free, the button is shown but explains it needs an upgrade rather than running. (The AI Knowledgebase, by contrast, is available on every plan, with a document allowance on Free.)