These are the third parties that process data on our behalf, what each one sees, and where it is located. We will give notice before adding a sub-processor.
Cloud Run (API and job execution), Cloud SQL for PostgreSQL with pgvector, Cloud
Storage (erasure journal), Secret Manager, and Cloud Logging. Location:
europe-west2 (London).
Sees: encrypted payloads, embeddings, metadata, and operational logs.
DNS, TLS termination, and the edge proxy in front of the API. Location: global edge network.
Sees: request metadata and, because TLS terminates at the edge, the bearer credential in transit. This is why the API also requires a Google identity token from a trusted origin.
Payment processing and usage-based billing. Location: EU/US.
Sees: billing identifiers and metered usage counts. No memory content.
Serves this marketing site. No customer data transits it.
Your embedding provider is chosen and configured by you, in code you run. If you embed locally, no third party sees your plaintext. If you use a hosted embedding API, that provider receives the text being embedded — but it is your processor, not ours.