When a conversation is started, gathering actor and context information to pass to the Mantella server takes time. A new request called mantella_initialize has been created to force Mantella's server settings to initialize (setting up LLM, starting TTS service, loading CSVs, etc) as soon as a conversation has been triggered. This means the server can now pre-prepare for a request while the Papyrus code gathers the necessary information for the conversation to save time.
When a conversation is started, gathering actor and context information to pass to the Mantella server takes time. A new request called
mantella_initializehas been created to force Mantella's server settings to initialize (setting up LLM, starting TTS service, loading CSVs, etc) as soon as a conversation has been triggered. This means the server can now pre-prepare for a request while the Papyrus code gathers the necessary information for the conversation to save time.See here for the Mantella Software implementation of this request type: https://github.com/art-from-the-machine/Mantella/pull/428