AssemblyAI ships multi-region streaming and dynamic prompts (Full Transcript)

[00:00:00] Speaker 1: This month at Assembly AI, we made streaming audio transcription faster, smarter, and accessible everywhere. Here's our full monthly recap for everything we shipped in November. So first up, we released multi-region streaming. We now have our models hosted in US East, which provides regional redundancy and expanded compute capacity for production workload. The infrastructure update reduces single-region dependency, which maintains your service continuity in the event of any regional incident. Your traffic is already being routed automatically based on latency and capacity, so there's nothing you need to do. If you want instead to use EU streaming, you can replace this base URL with our EU1 in your connection configuration right here. We've also released updates and improvements to our self-hosted streaming offering. If you're looking to self-host Assembly AI's streaming model, please get in touch with our team. We've also updated Key Term Prompting to support Dynamic Key Terms Prompt. So you can send an update configuration message with your new key terms, and that way the model will start to buy us for your terms mid-session. We've also added language detection to our streaming utterances, so you'll notice a language code being returned with your terms in our term response. And next, we also released Gemini 3 Pro on LLM Gateway, which is a great addition for any agentic workflows that you're building on your transcriptions. So that was all we had this month. Any relevant links to what we mentioned today are in the description. And I hope to see you again next month. Bye.