Privacy Policy

Mnemix, Inc. provides a memory and real-world enrichment API for AI voice agents. Our customers use Mnemix to recall prior caller context, attach caller intelligence to phone-based workflows, and write post-call memory for future agent interactions.

Mnemix is primarily a service provider to businesses. If you interact with a voice agent powered by Mnemix, the business operating that agent is usually responsible for its own notices, consents, call recording practices, and use of caller data.

• Account and tenant information, such as company name, plan, settings, webhook configuration, and dashboard authentication data.
• Caller and contact information, such as phone number, email, name, company, role, tags, custom attributes, summaries, enrichment status, and enrichment results.
• Interaction information, such as call summaries, transcripts if submitted by a customer, outcomes, sentiment, agent metadata, and timestamps.
• Operational information, such as trace identifiers, endpoint names, timing metrics, cache status, background job metadata, and delivery failure records.

Mnemix normalizes phone numbers to E.164 format in the application before storage and lookup. Audit logs do not store raw phone numbers. Instead, audit records store a phone_number_hash generated with HMAC-SHA256 using a per-tenant secret. That tenant secret is generated for each tenant and must not be logged or exposed.

• Provide, secure, operate, and debug the Mnemix API.
• Authenticate API requests and enforce tenant-scoped access.
• Recall caller memory and write interaction history.
• Run enrichment workflows through configured providers such as Trestle and, where enabled, Twilio Lookup, Baylio, or Anthropic.
• Monitor performance, reliability, spend limits, background jobs, and abuse risk.

The Mnemix dashboard uses Supabase authentication and session cookies to keep signed-in users authenticated. These cookies support login, dashboard access, and session refresh. Mnemix does not need browser cookies for server-to-server API calls.

Customer data is retained as needed to provide the service, meet operational requirements, support customer configuration, and satisfy legal or security obligations. Where configurable retention controls are not yet exposed in the product, customers may contact us to request deletion or export workflows.

Mnemix uses tenant-scoped application queries, hashed API keys, per-tenant phone audit HMACs, Cloudflare Worker secrets, sanitized operational logging, and managed infrastructure providers. No security measure is perfect, but we design the system to minimize raw sensitive data in logs and operational telemetry.

Mnemix relies on infrastructure and enrichment providers to operate the service. These may include Cloudflare, Supabase, Upstash, Trestle, Twilio Lookup, Baylio, and Anthropic depending on configuration. The current subprocessor list is available at /subprocessors.

Mnemix currently operates infrastructure primarily in the United States through managed providers. Customers with data residency or international transfer requirements should contact us before using Mnemix for regulated production workloads.

If you are a Mnemix customer, contact us at hello@mnemix.ai for access, correction, deletion, or security requests. If you are a caller whose information was processed through a customer voice agent, you may also need to contact the business operating that agent because it controls the purpose and context of the call.

We may update this Privacy Policy as the product, provider set, or legal requirements change. The date at the top of this page shows the latest revision.