Appointment Setter Voice — agentic threat model

Not certain from the listing — likely relies on a combination of speech-to-text, LLM, and text-to-speech models. Key threats include voice-based prompt injection (vishing) and adversarial audio inputs designed to hijack the underlying LLM's instructions.

Not certain from the listing — processes real-time voice data, calendar availability, and CRM records. Threats include the exfiltration of sensitive customer PII from the CRM and potential database poisoning if malicious inputs from calls are written directly to customer profiles.

Not certain from the listing — orchestrates voice dialogue, calendar scheduling, and CRM updates. Insecure tool integration is a major threat, where a caller could manipulate the agent into executing unauthorized API calls to delete calendar events or modify CRM records.

Not certain from the listing — requires telephony/VoIP infrastructure and API connections to third-party platforms. Threats include insecure storage of CRM/calendar API keys and potential exploitation of the telephony layer for toll fraud or unauthorized call routing.

Not certain from the listing — no details are provided regarding call logging, transcript auditing, or real-time guardrails. This creates a risk of undetected drift, prompt injection success, or inappropriate agent behavior during live calls.

Not certain from the listing — handling voice recordings and CRM data triggers strict compliance requirements (GDPR, CCPA, and potentially HIPAA). The listing does not mention encryption standards, access controls, or compliance certifications.

Not certain from the listing — operates within a vertical ecosystem by connecting directly to external CRM and calendar APIs. Threats include cascading failures if third-party APIs experience downtime or if compromised external integrations push malicious payloads back to the agent.