Trust & Legal
How Zeopilot uses AI, deterministic controls, evidence and human approval.
Effective October 5, 2026Zeopilot uses AI and deterministic systems to understand websites, identify problems and opportunities, create Candidate Futures, explain recommendations and support governed execution. AI-generated proposals are not automatically treated as verified truth.
Proposal agents, independent verification, challengers, approval controls and production execution are separate responsibilities. The product is designed so an agent that creates a candidate cannot simply certify its own work.
At launch, important production writes require customer approval. Approval is intended to bind to the exact artifact, source state and permission scope shown to the approving user. Customers can revoke connections and use emergency controls independently of the normal orchestration path.
Zeopilot labels evidence and confidence and can choose no winner when evidence is weak or conflicting. Business impact estimates are not presented as certain revenue unless the evidence supports that level of claim.
AI providers may be used to perform limited processing required for product functions. Zeopilot's data-handling obligations, subprocessors and customer-specific commitments are governed by the applicable privacy notice, DPA and customer agreement.