Evrcad Insurance Suite — Evrcad LLC | Version ai-disclosure-2026-07-v2 | Last updated: [DATE]

AI Features & Disclosure Notice

01Purpose

This Notice explains how the Evrcad Insurance Suite (the "Service") uses artificial intelligence ("AI"), what AI does and does not do, and the disclosures Evrcad LLC provides in connection with AI features. It supplements the Terms of Service and Privacy Policy.

02Where AI Is Used

The Service uses generative AI, provided through Amazon Web Services (AWS) Bedrock under Evrcad’s executed AWS Business Associate Agreement, for assistive functions such as drafting suggested text and summarizing prior interactions for an agent’s review. Additional AI features (including AI-assisted messaging and, in the future, an AI voice assistant) are in staged development and are described in the Terms of Service. Features not yet deployed are not available to users until Evrcad completes required legal and regulatory review.

03AI Is Assistive Only — Human Review Required

AI outputs are drafts and suggestions. A licensed agent must review, edit, and approve all AI-generated content before it is sent to any client or beneficiary. AI does not make decisions, does not bind Evrcad or any agent, and does not replace the professional judgment of a licensed insurance agent.

04Scope Limitations for Medicare

Consistent with CMS Medicare Communications and Marketing Guidelines, AI features are limited to administrative and informational functions. AI features are not permitted to recommend specific Medicare plans or products, quote or compare premiums or out-of-pocket costs, discuss specific plan benefits or formularies, enroll beneficiaries, or complete applications. These functions require a licensed agent.

05AI Behavioral Limitations and Human Hand-Off (Medicare)

In addition to the scope limitations in Section 4, the Service’s AI features are technically configured to keep AI interactions within permitted administrative and informational functions. The AI voice assistant, in particular, is designed with an automatic "hard hand-off": when a conversation moves toward regulated Medicare marketing or sales activity, the AI is programmed to stop that line of discussion and route the consumer to a licensed human agent, take a message, or end the topic.

The AI voice assistant is designed so that it will not: (i) recommend, endorse, rank, or steer a consumer toward any specific Medicare plan, carrier, or product; (ii) quote, estimate, or compare premiums, cost-sharing, or out-of-pocket amounts; (iii) describe specific plan benefits, coverage, provider networks, or drug formularies; (iv) enroll a beneficiary or complete an application; or (v) collect a Scope of Appointment or other sales authorization. When a consumer asks for any of these, the AI is designed to decline and hand the consumer off to a licensed agent.

These behaviors are platform guardrails configured by Evrcad. Agents and agencies cannot disable, weaken, or reconfigure the marketing hard hand-off. Any Third-Party Marketing Organization (TPMO) disclaimer, Scope of Appointment, and related Medicare marketing obligations remain the responsibility of the licensed agent and are not performed by the AI. This is a disclosed limitation of the AI and is not a substitute for the agent’s own compliance.

06AI Self-Identification (Utah Artificial Intelligence Policy Act)

Where a consumer interacts directly with a generative AI feature of the Service, the AI is designed to clearly and conspicuously identify itself as artificial intelligence, and not a human, at the outset of and throughout the interaction. For high-risk interactions involving sensitive personal information or personalized professional advice, this disclosure is provided prominently and proactively. If a consumer asks whether they are interacting with AI, the Service will clearly disclose that they are.

07AI Voice Assistant — Mandatory Call-Start Disclosure

When the AI voice assistant is enabled for an account and answers or handles a call, the Service plays a clear, audible disclosure at the very beginning of the call — before the AI engages in any substantive conversation — identifying the consumer’s counterpart as artificial intelligence and stating that the line is recorded.

This call-start disclosure is generated and controlled by the Service. It is mandatory and non-configurable: agents and agencies cannot remove, suppress, shorten, re-record, or disable it, and cannot configure the AI voice assistant to answer or handle a call without it. It plays automatically on every AI-handled call, whether the AI answers a call placed to a Service-provided number or handles a call forwarded from the agent’s own number.

If a consumer asks at any point whether they are speaking with a person or a machine, the AI is designed to confirm clearly that it is an AI assistant, consistent with Section 6. A consumer may ask to speak with a human, decline to continue, or end the call at any time, and the AI is designed to route the consumer to the agent or take a message on request.

08Recording and Transcription of AI Voice Calls; Consent

AI voice calls are recorded and transcribed so that the AI voice assistant can function, generate summaries and messages for the agent’s review, and support the agent’s recordkeeping and compliance obligations. Recordings and transcripts are processed and stored within Evrcad’s AWS environment under Evrcad’s executed AWS Business Associate Agreement, and transcription is performed by AWS services under that agreement. Telephony transport is provided by a telecommunications conduit that carries the call but is not used to store or semantically process call content beyond transport; call recordings are delivered directly to Evrcad’s AWS storage.

The mandatory call-start disclosure in Section 7 informs the consumer at the outset that the line is recorded. A consumer who does not wish to be recorded may decline to continue and end the call at that time; by continuing the call after the disclosure, the consumer proceeds on clear notice that the call is recorded and transcribed.

Recording, wiretap, and consent laws vary by state, and some states require the consent of all parties to a call. Because a consumer may be located in any state, the licensed agent is responsible for ensuring that recording of the agent’s calls complies with the laws applicable to the agent and to the consumer, including any all-party-consent requirements. The Service provides the call-start disclosure to support notice of recording; the licensed agent remains responsible for lawful recording of the agent’s calls.

For Medicare marketing, sales, and enrollment calls, CMS requires that call recordings be retained for defined periods that vary by call type. As between Evrcad and the agent, these recordkeeping and retention obligations belong to the licensed agent or agency as the Third-Party Marketing Organization. Evrcad provides the platform and storage as a software vendor and does not assume the agent’s independent regulatory retention duty. On platform shutdown or agent offboarding, Evrcad will make the agent’s records available for the agent to retain, after which the ongoing retention obligation rests with the agent or agency.

09Responsibility for AI

Evrcad and its agent-users remain responsible for the outputs of AI features used through the Service. The use of AI is not a defense to any violation of consumer protection, HIPAA, CMS, or other applicable law. Agents are responsible for all communications they send, whether or not drafted with AI assistance. This responsibility extends to AI-handled and recorded voice calls and to satisfying the recording-consent, TPMO, Scope of Appointment, and other Medicare disclosures required for those calls.

10AI and Your Data

AI features are implemented using AWS Bedrock. Under AWS policy, prompts and outputs processed through Bedrock are not used to train or improve foundation models used by other customers and remain logically isolated to Evrcad’s AWS account. Evrcad does not use PHI or personal information to train shared AI models. PHI included in AI prompts is limited to the minimum necessary for the feature. Audio recordings and transcripts generated by the AI voice assistant are processed and stored as described in Section 8 and are subject to the same AWS account isolation and no-shared-model-training protections described in this Section.

11Changes

Evrcad may update this Notice as AI features change or as law requires. Material changes affecting how AI processes personal information or PHI will be communicated with reasonable notice, and, where the change alters Evrcad’s HIPAA posture (for example, AI processing of call content), the Terms of Service and Business Associate Agreement will be updated accordingly.

12Contact

Questions about AI features: support@evrcad.com. Legal inquiries: legal@evrcad.com.