Responsible AI Policy

1. Our Commitment

ASSIST™ deploys AI agents that act in your name. We take that responsibility seriously. This policy describes the principles that govern how ASSIST™ AI operates, and the commitments we make to our customers and to the people their agents interact with.

2. Human Oversight Is Non-Negotiable

AI handles execution; humans handle judgment. This is not a disclaimer — it is the design of the system.

• No ASSIST™ agent activates without customer-defined configuration that a human has reviewed and approved.
• When a situation falls outside configured parameters, agents escalate to humans. They do not improvise.
• Customers remain responsible for reviewing agent performance, approving templates before activation, and acting on escalations.
• ASSIST™ is designed to log automated agent actions for auditability. Customers may access their audit trail as part of the service.

3. Transparency

ASSIST™ will not affirmatively claim its agents are human when directly and sincerely asked by the person they are communicating with. Customers who are required by law to disclose that communications are AI-generated are responsible for ensuring that disclosure is made. ASSIST™ will inform Enterprise Customers in advance of material changes to agent behavior where practicable. This policy, our AI Disclaimer, and our Acceptable Use Policy are publicly available at poweredbyassist.com/legal.

4. Escalation by Design

ASSIST™ agents are designed to escalate — not improvise — when they encounter situations outside their configured scope. Escalation routing is a first-class feature of the platform, not a fallback. Customers who fail to act on escalations accept responsibility for the consequences.

5. No High-Risk Autonomous Decisions

ASSIST™ agents are designed for front-office business automation. They are not designed to make autonomous decisions in high-stakes contexts where errors could cause serious harm. Specifically:

• Agents will not autonomously execute contracts, agreements, or financial commitments.
• Agents will not make autonomous decisions about hiring, promotion, or termination.
• Agents will not provide legal, medical, or financial advice.
• Agents will not operate in clinical triage, emergency response, or life-safety contexts.
• Agents will not operate in regulated industries without the required compliance arrangements in place.

6. Customer Responsibility

The most important safety control for ASSIST™ is the customer. By deploying ASSIST™, customers accept responsibility for: the accuracy and lawfulness of the data they provide; the content of the templates and sequences they approve; compliance with all applicable laws governing AI-assisted communications; maintaining human oversight of deployed agents; and acting on escalations promptly.

7. Fairness and Non-Discrimination

ASSIST™ customers may not configure agents to discriminate against individuals on the basis of protected characteristics. Customers may not use AI-generated outputs to make decisions that violate applicable anti-discrimination law. ASSIST™ will not knowingly facilitate discriminatory use of its platform.

8. Auditability

ASSIST™ is designed to log automated agent actions for auditability, making agent activity traceable and available for Customer review. Customers may request an explanation of any automated decision or escalation. ASSIST™ maintains logs for a period consistent with operational, contractual, and legal obligations.

9. Regulated-Use Restrictions

The following deployments require additional arrangements before ASSIST™ may be used: Healthcare — Business Associate Agreement (BAA) required before any PHI is processed; Financial services — applicable regulatory compliance review required; Legal services — use limited to administrative functions, not functions constituting the practice of law; Employment decisions — ASSIST™ may not make autonomous hiring, promotion, or termination decisions.

10. Model and Provider Changes

ASSIST™ uses AI models provided by third-party providers as components of the platform. ASSIST™ will notify Enterprise Customers in advance of material changes to AI models or providers used in their deployment, where practicable. ASSIST™ evaluates AI model providers for reasonable data handling practices and executes data processing terms with those providers where applicable. Provider changes will not be used to circumvent Customer Data protections.