Hello everyone! I have a question. Do you know how to make it so that when I use a tool, the message my assistant says isn't always the same? In the API documentation they say that supposedly by doing a patch and sending "contents" in plural, and the message objects should theoretically work, like: "messages": [ { "type": "request-start", "contents": [ { "type": "text", "text": "Dame un segundo, lo busco", "language": "es" }, { "type": "text", "text": "Un momento, verifico eso", "language": "es" }, { "type": "text", "text": "Déjame revisar ese dato", "language": "es" }, { "type": "text", "text": "Ya lo reviso", "language": "es" }, { "type": "text", "text": "En eso ahora", "language": "es" }, { "type": "text", "text": "Confirmo eso ahora", "language": "es" }, { "type": "text", "text": "Buscando ese dato", "language": "es" }, { "type": "text", "text": "Revisando la información", "language": "es" }, { "type": "text", "text": "Let me look that up", "language": "en" }, { "type": "text", "text": "One sec, checking", "language": "en" }, { "type": "text", "text": "Give me a moment", "language": "en" }, { "type": "text", "text": "Checking that now", "language": "en" }, { "type": "text", "text": "Let me confirm", "language": "en" }, { "type": "text", "text": "On it now", "language": "en" } ], "blocking": false }] But in my case, it always picks the first one, it's like the assistant is "lazy" or sluggish, and it even says the message in the wrong language (e.g., we're talking in Spanish and it tells me "Let me look that up"). Has anyone experienced this before and/or knows how to solve this?