Legal

Subprocessor Register

Last updated July 14, 2026

Cortex uses the following subprocessors to deliver the platform. This register matches the one published in the Trust Center; customers on the DPA receive 30 days' notice before any new subprocessor goes live.

01Current subprocessors

  • Cloud infrastructure (GCP) — compute, storage, managed Postgres. Data: all tenant data (encrypted). Region: customer-selected.
  • Anthropic — LLM inference via the Router. Data: prompt + context at run time. Region: US / EU.
  • Google (Gemini) — LLM inference via the Router. Data: prompt + context at run time. Region: US / EU.
  • OpenAI — optional LLM inference via the Router. Data: prompt + context at run time. Region: US.
  • Object storage / CDN — artifact and asset delivery. Data: generated artifacts (encrypted). Region: customer-selected.
  • Error & ops telemetry — reliability monitoring. Data: operational metadata only, no payloads. Region: US / EU.

02Notes

Model subprocessors are inference-only and contractually barred from training on customer data. Bring-your-own-key tenants call providers under their own agreements; air-gapped deployments use local models and engage no model subprocessor at all.

To subscribe to change notices, email privacy@cortexrizi.com with subject "Subprocessor notices".

Questions about this document: legal@cortexrizi.com · See also the Trust Center

Subprocessor Register | Cortex AI OS