When AI is being used
Features that involve AI are labelled as such in the product — an AI-assisted suggestion, an AI-drafted deployment, an AI-generated summary. It's not hidden behind generic UI copy.
Confidence levels where appropriate
Where a recommendation carries meaningful uncertainty (for example, an inferred root cause for an incident), we surface that uncertainty rather than presenting a guess as fact.
Generated code identification
AI-generated code proposed through the platform is presented as a reviewable diff, clearly attributed as AI-generated, before it's committed.
Generated documentation identification
Same principle for documentation — AI-drafted docs are marked as drafts pending human review, not published as if a person wrote and verified them.
Generated infrastructure identification
Infrastructure generated through our Designer or Blueprint tooling is traceable back to the AI-assisted design step that produced it, alongside the human who reviewed and approved the apply.
Clear explanation of AI-generated recommendations
Where practical, AI output includes a short explanation of why — which inputs it used, what it's optimising for — rather than a bare recommendation with no reasoning attached.