Responsible AI Practices
Dendryte Solutions believes website automation should be transparent, appropriately limited and supported by a clear path to human review.
1. Transparency
Website visitors should be told when they are interacting with an automated assistant.
2. Approved knowledge
Customer-approved information is used to configure business-specific responses.
3. Appropriate limitations
An assistant should acknowledge unsupported questions instead of inventing business information.
4. Human involvement
Important, sensitive, unusual or unauthorized requests should be routed to qualified people.
5. Data minimization
Assistants should request only information reasonably necessary for the configured purpose.
6. Sensitive information
Ordinary chat should not request complete card numbers, banking credentials, passwords, Social Security numbers, medical records or other unnecessary sensitive data.
7. Customer responsibilities
Customers must provide accurate information, review configurations, maintain their source content, respond to escalated inquiries and use the assistant lawfully.
8. Continuous review
Configurations should be reviewed as services, policies and customer needs change.
9. No guarantee of perfect output
AI-generated responses can occasionally be incomplete, outdated or incorrect. Dendryte does not represent that an assistant will produce perfect responses or eliminate the need for human judgment.
10. Reporting a concern
If you have concerns about a configuration, unexpected behavior or these practices, please let us know: