Overview
Mnemix uses managed infrastructure providers and enrichment providers to deliver the API. Some providers are required for the core service; others are optional depending on customer plan, feature flag, or production secret configuration.
Current Providers
| Provider | Purpose | Region | Status |
|---|---|---|---|
| Cloudflare | Edge compute, routing, Worker secrets, Hyperdrive connection pooling | United States / global edge network | Required |
| Supabase | Postgres database and dashboard authentication | United States | Required |
| Upstash | Redis cache, rate limiting, QStash background queues | United States / managed regions | Required |
| Trestle | Phone and caller enrichment | United States | Enabled for enrichment |
| Twilio Lookup | Phone-line intelligence such as carrier, line type, and SIM swap signals | United States / managed regions | Optional, if configured |
| Anthropic | Optional post-call summary generation and memory extraction | United States / managed regions | Optional, if configured |
| Qdrant Cloud | Managed vector index for memory similarity search; Mnemix syncs each memory's embedding vector (a numerical representation derived from memory text) together with memory metadata — tenant and entity identifiers plus classification fields (kind, confidence, source trust, privacy class, validity window, ingestion time) — and never raw message content, transcripts, summaries, or other plaintext PII | Provider-managed regions | Enabled when vector sync is configured |
| Voyage AI | Embeddings for memory similarity search; Mnemix redacts detected phone numbers, SSNs, email addresses, and payment card numbers from memory text before embedding requests | Provider-managed regions | Required for production vector search when configured |
Changes
Mnemix may update this list as the product and provider set changes. Customers with subprocessor notification requirements should contact hello@mnemix.ai before using Mnemix for regulated production workloads.