Key Features

Data Anonymization

Protect your customer data Guard personally identifiable
information through anonymization

• Keep customers safe: Protect sensitive customer data from being shared with third-party LLMs.
• Stay private: Mask personal information with pre-built blocked fields like emails and phone numbers.
• Create custom fields: Define custom fields for removal for your specific business use case.

Response Control

Control responses Preserve your brand image by filtering for bias, toxicity and
inappropriate topics

• Be safe: Prevent dissemination of harmful responses being sent to your customers.
• Set boundaries: Ensure conversations are within acceptable boundaries and avoid conversations with sensitive topics.
• Avoid extremes: Enable LLM responses to be neutral and avoid expressing biases.

Model Visibility

Maintain model visibility Understand model performance to ensure IVAs are behaving as intended with monitoring and testing, including coherence, relevance and consistency

• Maintain quality: Align prompts with use cases by analyzing qualitative metrics.
• Check performance: Compare various model prompt performance on relevancy, cohesiveness and other metrics on accuracy.
• Monitor usage: View usage metrics such as tokens consumed, success rate and average response time to optimize performance and cost.