Where AI May Be Used
Depending on the product surface, AI may be used for generation, retrieval, classification, tool use, evaluation, or content adaptation.
The exact model stack depends on the product, deployment environment, and risk posture. We document the automation boundary for operators and reviewers.
How Outputs Should Be Interpreted
Model output can be incomplete, incorrect, stale, or overconfident. It should be treated as an operational input that may require validation, not as an unquestionable source of truth.
Customers remain responsible for approving production decisions, regulated workflows, and externally published content unless a signed agreement explicitly states otherwise.
Transparency Commitment
When AI is materially involved in a delivered workflow, we aim to make the automation boundary understandable to operators, stakeholders, and reviewers.
That includes documenting human review expectations, failure modes, and the limits of what the system is designed to decide automatically.
Last updated: August 20, 2026