Data access and separation
Users authenticate into an organisation-scoped workspace. Customer data is separated from every other customer, ownership is verified before records can be read or changed, and data is encrypted in transit.
Trust and security
Built for the realities of regulated product work
Compound is designed for work that must remain verifiable, traceable, and under your team's control. That includes clear answers about how customer data is handled and what your security and quality teams can review before deployment.
Model training
Customer records are used only to provide Compound to you. They are not used to train our models or third-party foundation models, run advertising, or sell data.
Evidence and verification
Material facts stay linked to their source. Compound keeps AI-assisted work distinct from deterministic checks and surfaces missing, conflicting, or uncertain evidence for review.
Review and approval
Named reviewers approve, reject, or correct outcomes that require judgement. Compound prepares the evidence and decision context while your team retains authority over the decision.
Audit trail
Every action that contributes to or changes governed work is logged with its source, timestamp, AI provenance, checks, corrections, and reviewer decisions, so your team can reconstruct how the outcome was reached.
Security and validation review
We can provide current architecture and data-flow information, model-use details, control descriptions, validation documentation, and our current subprocessor list for review.
Review Compound before you rely on it
Tell us what your quality, IT, or security team needs to assess. We will share the current materials and walk through the controls relevant to your intended use.