Independence and evidence boundaries.

Absolute neutrality is not a credible promise. Here is what we can prove, what remains a model-generated hypothesis, and where the governed evidence standard begins.

• No pay-to-play, ever

  Vendors cannot pay to appear in the analysis, to influence ordering, or to suppress a competitor. We do not accept vendor sponsorship of specific briefs. If we ever introduce vendor revenue, it will be disclosed on every page it touches — not buried in a footer.

• Category-first reasoning

  The public brief reasons about categories first. It may then name illustrative vendors from model knowledge, but those names are visibly unverified hypotheses — never rankings or evidence-bound recommendations. The governed workspace uses the published vendor library and its source controls.

• Simulated perspectives are labelled honestly

  The confidential brief uses one model to simulate distinct professional lenses — architecture, procurement, org design, and operations. They are useful challenges, not independent agents or expert testimony. Unresolved trade-offs remain visible.

• Two lanes, two different standards

  The confidential lane is directional and uncited. Provisioned workspaces are evidence-bound: material claims carry source locators and status, then pass human approval before an immutable Blueprint can be published. The two are never presented as equal.

• No product-vendor placement channel

  We do not separately share briefs or mention telemetry with product vendors for sales, placement, or lead generation. The disclosed Lovable and Google processors still handle submissions and may themselves offer products.