Intelligence infrastructure and data handling

Intelligence runs on Proofenance-owned NVIDIA GPU servers - dedicated Ubuntu machines on a controlled private network. We deploy well-reputed open-weight models on that estate rather than routing sale or governance content through third-party hosted inference APIs.

Operating the full stack ourselves strengthens security: model weights and runtime stay under Proofenance control, traffic does not cross into public cloud inference services, and we patch and harden hosts on our own schedule.

Open-weight models keep customer material off external vendor APIs where retention and training policies are outside our direct oversight.

Intelligence analyses sale and governance-record content already contained on Proofenance: for example, correspondence, diligence, Open Banking exports, investigation workspaces, and training records.

When a sale or governance record is analysed, processing may cover bank patterns, diligence drafts, media relevance, ownership maps, and investigation assembly.

Third-party screening and identity providers remain separate regulated subprocessors; their API usage is documented on the Security hub.

All data flowing through Proofenance and Intelligence is encrypted in transit and at rest. User access is role-based with audit logging on key staff actions.

Client data retention follows the data lifecycle described on the Security hub.

Rather than dictate the output conclusions, Intelligence prepares content for easy and comprehensive review by users. It does not auto-approve or auto-reject compliance decisions.

Where Intelligence has contributed to a sale or governance record, findings can be traced to the relevant source documents and checks.