Countervail helps legal, compliance, and technical teams understand where AI interaction records may exist across an organization’s stack. Start with code, known vendors, a governance questionnaire, an AI export, policy material, or an active matter profile.
The practical question is simple: if a subpoena, litigation hold, regulator, insurer, or opposing counsel asked tomorrow, what would you actually have to preserve?
Countervail is a guided diagnostic for AI record exposure. Choose the intake path that matches what you have in hand.
Use this when a technical team can provide a GitHub repository or ZIP archive. Expose detects AI providers, observability vendors, agent frameworks, storage layers, and downstream processors.
Use this when you already know the vendors in use and want to model how records may move across AI providers, tracing tools, vector databases, auth systems, analytics, and storage layers.
Evaluate whether the organization can explain its retention, deletion, access, audit, downstream transfer, and legal hold posture.
Use this when the records are not in a company repository but inside a ChatGPT or Claude export, transcript, prompt file, or project archive.
Use this when a scan or assessment needs to become a deliverable: control status, gaps, implementation notes, and a plain record of what was found. Shows the difference between perimeter controls and records found outside the perimeter.
Countervail does not decide legal privilege, produce legal advice, or determine whether a record must be produced. It helps teams find and describe the record custody surface before counsel, regulators, insurers, or opposing parties ask for it.