This support article announces an enhanced training flow for Conversation AI that improves how the system analyzes user queries. Previously, the system relied on suggestion flows and LLM tools that sometimes produced inaccurate results.
The new system examines the last 5 messages of conversation on each AI request to determine whether contacts need assistance or have business questions about pricing and services. It automatically pulls queries from discussions and generates analysis scores.
The three-step process involves automated conversation review to grasp context, generating a query analysis score showing whether support is needed, and applying the most pertinent training information from a sub-account's bot to deliver customized responses.
The article emphasizes that no action is required from users since these improvements function automatically behind the scenes to boost response precision and relevance.
The update reflects customer feedback and represents an internal enhancement requiring no configuration changes from end users.