Sowtek AI — agentic threat model

Not certain from the listing — The specific LLMs or foundation models powering the multi-language chatbots and AI recommendations are not disclosed. Potential threats include adversarial prompt injection bypassing chatbot guardrails or model reprogramming.

Not certain from the listing — While it aggregates unified customer insights and real-time KPIs, the underlying vector stores or database architectures are unspecified. Threats include data exfiltration of sensitive customer records or knowledge-base poisoning affecting AI recommendations.

Sowtek uses a low-code drag-and-drop orchestration framework for workflows, IVR routing, and multi-channel automation. Threats include insecure tool integration with SIP/E1 lines and social media APIs, leading to unauthorized call routing or mass messaging abuse.

Offers both cloud and on-premise deployment options with SIP, analog, and E1 line integrations. Threats include container/host compromise on-premise, privilege escalation, or unauthorized access to telecom infrastructure.

Provides real-time KPIs, customizable reports, and performance tracking dashboards. However, it is unclear if these tools monitor for LLM-specific anomalies, drift, or adversarial inputs, leaving potential blind spots in agent behavior.

Not certain from the listing — No specific compliance standards (e.g., SOC2, GDPR, ISO) or identity/access management controls are detailed, though on-premise deployment offers localized control. Risks include weak authorization over sensitive communication channels.

Features a multi-agent platform designed for hybrid collaboration with human agents across customer care, sales, and social media. Threats include cascading failures if a compromised agent sends rogue mass messages or corrupts shared customer insights.