AI & Voice Disclosures

Corvyn Labs provides automation and AI-assisted capabilities, including LLM-based components and voice workflows. This page describes important disclosures and risk factors so businesses can make informed decisions.

These disclosures are not legal advice. They are intended to help you understand how AI-assisted and automated features may behave and what responsibilities remain with you.

1. Deterministic and non-deterministic behavior

Some parts of the Services use deterministic logic (rules, routing, validation) and some parts may use non-deterministic systems (for example, large language models) that can produce variable outputs.

In practice, features may combine both, such as deterministic eligibility checks with AI-assisted drafting.

2. What AI may be used for

Depending on configuration and enabled features, AI may be used to help:

• Draft suggested SMS/email replies to End Users.
• Summarize conversation history and extract key details.
• Classify leads, intents, or topics.
• Suggest next steps or routing outcomes.
• Support voice workflows through transcription or response generation.

3. LLM limitations

• LLMs can generate inaccurate, incomplete, or misleading responses.
• LLMs may hallucinate facts or present speculation as certainty.
• LLMs can be influenced by user-provided inputs, including malicious prompt injection attempts.
• LLMs may produce content that is inappropriate or non-compliant if not properly constrained.

Because of these limitations, you should treat AI Output as a suggestion that may require review, editing, and validation before being used.

4. Human review and business responsibility

You are responsible for configuring the Services appropriately for your use case and for implementing meaningful human review where needed. Corvyn Labs does not provide legal, medical, financial, or safety advice.

You remain responsible for:

• The content of communications sent to End Users.
• Any offers, quotes, scheduling commitments, or representations made to End Users.
• Compliance with consent, opt-out, and content requirements.
• Determining whether the Services are appropriate for your workflows and risk profile.

5. Voice and telephony considerations

• You are responsible for complying with telemarketing, anti-spam, consumer protection, and consent requirements (including SMS/voice consent and opt-out).
• Call recording laws vary by jurisdiction. If you record calls, you must provide required notices and obtain required consent.
• Carrier filtering, delivery failures, and phone number reassignment can occur. You are responsible for handling delivery issues and maintaining accurate contact data.

Voice systems can also be affected by background noise, accents, transcription errors, and ambiguous intent. For important details, implement confirmation steps and human escalation paths.

6. Safety and security

We design the Services to be safe by default, but no system is perfectly secure. Threats may include account compromise, misconfiguration, malicious prompts, and downstream third-party failures.

You should:

• Limit who can configure automation rules and AI prompts.
• Use least-privilege access and strong authentication.
• Review logs and performance metrics for anomalies.
• Test workflows before enabling them broadly.

7. Prohibited high-risk use

Unless expressly agreed in writing, you must not use the Services for high-risk activities where errors could reasonably be expected to cause harm, including emergency response, critical infrastructure control, life-safety systems, or medical diagnosis.

8. Incident response

If you believe there is a security incident, misuse, or a safety issue related to the Services, contact us using the contact method listed on the Site.