Kenect — agentic threat model

Not certain from the listing — likely utilizes commercial LLMs or fine-tuned conversational models for SMS and voice. Primary threats include prompt injection attacks that could trick the model into bypassing payment verification or sending unauthorized messages to customers.

Not certain from the listing — relies on integrations with Dealership Management Systems (DMS) and CRMs to access customer records and inventory. Threats include data exfiltration of customer PII and database poisoning via malicious customer input fields.

Not certain from the listing — orchestrates multi-step workflows for scheduling, review gathering, and payment collection. Threats include tool misuse where the agent is manipulated into triggering unauthorized payment requests or scheduling conflicts.

Not certain from the listing — likely hosted on cloud infrastructure with integrations to telephony/SMS gateways (e.g., Twilio) and payment processors. Threats include API key theft, which could allow attackers to hijack SMS channels or intercept payment flows.

Not certain from the listing — requires strict conversational guardrails and transaction monitoring to prevent reputational damage or fraudulent operations. Gaps in observability could lead to undetected social engineering of customers by the AI.

Not certain from the listing — must adhere to PCI-DSS for payment processing and TCPA regulations for automated SMS outreach. Lack of robust access controls or audit logs for AI-driven actions poses severe compliance and legal risks.

Not certain from the listing — operates within a horizontal ecosystem connecting dealerships, customers, and third-party financial/scheduling APIs. Vulnerabilities in these external integrations could lead to cascading failures or unauthorized data access across platforms.