Docket strives for the highest standard of precision and accuracy in all our responses. Our AI model is tailored for high precision, targeting nearly 95% accuracy and a maximum hallucination of 2.7%.
Docket's multiple-model Galaxy Architecture employs dedicated AI models for retrieving, analyzing, and selecting answers. Galaxy excels in fetching answers from optimal sources, scrutinizing various knowledge repositories like Slack, Drive, Highspot, etc., and ultimately determining the most fitting response to deliver.
Docket strives for the highest standard of precision and accuracy in all our responses. Our AI model is tailored for high precision, targeting nearly 95% accuracy and a maximum hallucination of 2.7%.
We ask for the most trusted sellers in your company, and everytime they have answered a question across channels, we weigh those answers higher. Thus making Docket a clone of your best sellers and becoming a launchpad to ramp up new sellers.
Implicit feedback is based on users actions. If users copy Docket’s response, it indicates a high value of the response. We gather implicit feedback by watching user interactions with the content, offering clues about its value and relevance.
Explicit feedback for responses comes in from users checking the system's answers for accuracy, making changes to them, or directly showing approval or disapproval, like giving a thumbs up or down, as a clear signal of value.
If a human verifies a Docket response as it is or with minor edits, we save it as human-verified. This exact response will then be used for similar future questions, rather than creating a new one. It’s a great way to democratize sales scripts.
