Best practices for generating guides using transcripts

Genesys Cloud Resource Center provides best practices for generating guides using AI-generated transcripts to optimize results. Key requirements include enabling AI Insights at the program level for the selected queue and ensuring the queue remains AI Insights-enabled during the chosen date range. The system requires sufficient data volume, with each reason-for-contact group containing a minimum of 30 conversations; queues should realistically contain several hundred conversations during the selected period to prevent transcript filtering and guide generation failures. Filters must accurately match queue parameters, including date range, channel, and language, as mismatches result in zero conversations being identified. A sufficiently wide date range is recommended, particularly for low-volume queues that should look back further (such as 30 days) to enable reason-for-contact groups to reach the required threshold. Additionally, selecting queues with resolved or partially resolved interactions is preferable, as conversations without resolution outcomes do not contribute to clustering and guide generation. Well-crafted prompts should define AI agent behavior, tone, and constraints while describing the guide's goal and intent with streamlined, sequential instructions that specify tools, data actions, and variables as placeholders.