Aeterna AI Architects

Security

How a delivered Voice Qualifier instance is built to fail closed, not open.

API-key authentication

Every control endpoint (trigger a call, update the script, read results) requires an X-API-Key header. No key configured means every request is rejected — the system fails closed, never open.

Encryption in transit

All traffic to and from a delivered instance runs over TLS, including the webhook exchange with Twilio and the classification call to your chosen LLM provider.

Your data stays yours

Call outcomes and transcripts are delivered directly to the destination you configure — your CRM, your webhook. We don't retain a copy or route your data through a third-party store we control.

Bring your own credentials

Twilio and LLM-provider keys the build uses are yours — you control the account, the spend, and can revoke access at any time. We don't hold a standing copy after delivery.

Never gets stuck, never fails silently

If the LLM classification call fails mid-conversation, the instance falls back to keyword classification rather than dropping the call or the result — every call resolves to a definite outcome.

Credential handling during support

If a support task requires access to your environment, access is scoped to that specific task and time-boxed. We do not request or hold standing admin credentials as a condition of delivery.

Incident response

If a security issue affecting a delivered build is identified, we notify the affected party without undue delay, describe the exposure plainly, and ship a fix on a priority track.

Responsible disclosure

If you believe you've found a vulnerability in a delivered build or on this site, contact us through the channel you used to place your order. Describe the issue and, if possible, steps to reproduce it. We take reports seriously, do not pursue legal action against good-faith researchers who report responsibly, and will confirm receipt and a fix timeline directly.