Responsible AI Policy

Purpose and scope

This policy applies to all AI-powered features in foretale.ai, including conversational agents, document summarization, analytics assistance, and automated workflow recommendations. It covers internal development practices and customer-facing AI behavior.

Human oversight

AI outputs are designed to assist professional judgment, not replace it. Users retain responsibility for decisions made using platform outputs. High-impact recommendations require human review before action in production workflows.

• Users can inspect AI reasoning steps and source references where available.
• Automated actions that modify customer data require explicit user authorization.
• Escalation paths to human support are available for disputed or uncertain outputs.

Fairness and bias mitigation

We evaluate models and prompts for disparate impact across common use cases. Training and inference data is scoped to customer-authorized content and platform metadata — we do not use customer data to train foundation models.

• Prompt and guardrail reviews before production deployment.
• Monitoring for anomalous or harmful output patterns.
• Regular evaluation against representative compliance and risk scenarios.

Safety controls

We use Amazon Bedrock Guardrails and application-level controls to reduce harmful, off-topic, or non-compliant outputs. Rate limiting and context budgets protect system stability and data exposure.

Transparency and accountability

We document model providers, primary use cases, and known limitations. See our AI Transparency Statement for customer-facing disclosures. Questions about this policy: contact@hexango.com.

Continuous improvement

We review this policy at least annually and after material changes to AI capabilities. Feedback from customers and internal audits informs updates.