Sub-processors
Sub-processors
Last updated 2026-04
These are the third parties PersonaMesh, Inc. uses to deliver the Service. We add or replace sub-processors only with 30 days' advance notice to workspace owners; Controllers on an executed DPA may object in writing.
| Sub-processor | Purpose | Data | Region |
|---|---|---|---|
| Amazon Web Services | Hosting, Postgres (RDS), Redis (ElastiCache), object storage (S3), KMS, networking | All workspace data at rest and in transit | us-east-1, us-west-2, eu-west-1, ap-southeast-1 |
| Anthropic | Default model provider — Claude family for chat, planning, summarization, classification | Persona spec excerpts, recalled memory snippets, user messages submitted to chat | US |
| Voyage AI | Default embedding provider | Memory + knowledge text submitted for embedding | US |
| Stripe | Billing and invoicing for paid plans | Workspace + billing contact, plan + metered usage events | US, EU |
| Postmark (ActiveCampaign) | Transactional email — invites, password resets, delivery receipts | Recipient email, message body for transactional templates | US |
| Cloudflare | Edge caching, DDoS protection, WAF on personamesh.ai | Request metadata only (no content) | Global edge |
| Datadog | Logs, metrics, traces (operations only — no Content) | Operational telemetry | US |
| GitHub | Source repository, container registry (GHCR) | Source code, build artifacts; no customer data | US |
Customer-selected providers (BYO)
Customers can swap default providers. When you do, that provider becomes a sub-processor of yours and is governed by your relationship with them, not ours. Examples include:
- Model: OpenAI, Bedrock, Vertex AI, self-hosted vLLM / TGI / Ollama / llama.cpp.
- Embeddings: OpenAI, Cohere, self-hosted.
- Vector store / object store / KMS: AWS, GCP, Azure, on-prem.
- Identity: any OIDC / SAML provider.
On-prem and edge deployments
When PersonaMesh runs on-prem or at the edge under a customer's control, the customer is the controller of all data; PersonaMesh, Inc. is not in the data path and the sub-processor list above does not apply.
Notification list
Subscribe to sub-processor change notifications by emailing trust@personamesh.ai.