1. When AI is used
Ravensford products may use AI to respond to enquiries, classify leads, summarise conversations, draft follow-ups, recommend booking actions and produce performance insights. Some content may be generated by third-party models.
2. Disclosure of AI interaction
Customers should configure agents to identify themselves as automated assistants where required by law or reasonably necessary to prevent deception. Voice cloning, impersonation and false claims of human identity must not be enabled.
3. Human control and escalation
Every deployment should define approved topics, restricted actions, transfer triggers, emergency routing, hours and escalation contacts. An AI agent must not independently commit the customer to financial, legal or contractual obligations unless expressly authorised with suitable controls.
4. Accuracy and limitations
Generated outputs can contain mistakes, omissions, bias or inappropriate recommendations. Knowledge bases should be maintained, performance reviewed, and customer-facing commitments validated. Analytics are indicative unless verified against reliable source systems.
5. Training and data handling
The production service must disclose which vendors process model inputs, whether data may be retained or used for training, and what enterprise controls are available. Customers should avoid entering unnecessary sensitive information.
6. Monitoring and improvement
Performance Engine may identify containment rates, conversion trends, handoff failures and response quality. Metrics must be labelled accurately; simulated demo values must never be represented as actual commercial performance.
7. Complaints and corrections
Customers should provide accessible human escalation and a process to review AI-caused errors, objection requests and harmful interactions.
Questions about this policy?
Use the Ravensford contact page to submit an enquiry. Production legal, privacy and security contact addresses must be confirmed before launch.
