Engine 64 is designed for customer-facing AI, where visitor input, business content, integrations, and model output all cross important boundaries. Our approach keeps those boundaries explicit from ingestion to response.
Product data is scoped to the owning organisation and, where applicable, the individual agent. Authenticated workflows validate membership or authorised support access before returning tenant-owned data.
Widget requests resolve their scope from a publishable key lookup, then verify agent ownership and allowed website domains. Secret material is not embedded in the public widget.
Website pages, uploaded documents, and visitor messages are treated as untrusted input. The assistant is instructed to use retrieved content as evidence while preserving operator rules and refusing to follow conflicting instructions found inside that content.
Customer-configured URLs pass through SSRF controls, redirect checks, timeout handling, and destination policies. Outbound integration delivery uses guarded requests and provider-specific destination restrictions where available.
Production rate limiting uses a shared backend. Operational health checks, structured logs, error monitoring, audit records, and explicit rollout controls support safe changes without silently widening access.
Our privacy policy explains how personal information is handled, our terms set the service boundary, and our sub-processor page lists the providers used to deliver the platform. Security or procurement questions can be sent directly to our team.
Tell us about your requirements and we will answer plainly, with the relevant technical or policy detail.