News
Microsoft Expands Azure AI and Integration with Voice, Search and Indexing Tools
Microsoft revealed a set of updates to its Azure platform during the Build 2025 conference aimed at improving dev access to AI-powered voice agents, document indexing and vector search integration.
The releases include the Voice Live API for Azure AI, new document indexing features in Azure Cosmos DB and expanded support for Azure Logic Apps connectors in Azure AI Search.
Voice Live API Enables Real-Time AI Voice Agent Applications
Microsoft has released the public preview of the Voice Live API, a new tool that allows developers to build AI voice agents capable of real-time, two-way conversations. Built on Azure AI Speech and integrated with Azure Communication Services, the API enables voice agents to speak and listen simultaneously. The company said the feature will mimic the feel of a natural phone call.
The API supports conversational features such as barge-in (where users can interrupt the agent) and latency-aware responses, which help maintain natural timing. It can work with models hosted on Azure OpenAI or other Azure endpoints.
Audio is streamed both ways in real time, making the API suitable for use in customer service bots, virtual assistants and other interactive voice-based applications.
Microsoft has also introduced a unified Azure AI Speech Resource that combines speech synthesis and transcription functions that aims to reduce deployment times. The Voice Live API uses JSON messages over WebSockets and is designed to fit into both customer service systems and embedded applications.
Document Indexer Adds Unstructured Data Processing in Cosmos DB
Microsoft also introduced the Document Indexer for Azure Cosmos DB, now in preview, allowing automatic processing of unstructured documents stored in the NoSQL database. The indexer supports PDFs, Office files and plain text, and extracts content into Azure AI Search for retrieval and summarization.
The feature is built using Azure AI Document Intelligence and leverages Azure Logic Apps for orchestration. Organizations can use it to build search and retrieval systems across document repositories without requiring external ETL pipelines.
The preview release currently supports Cosmos DB's API for NoSQL and is available in select Azure regions, with expanded support for additional APIs and areas coming soon.
Logic Apps Connectors Announced
Microsoft also announced this week it has added Azure Logic Apps connectors to Azure AI Search, allowing developers to automate ingestion and enrichment of enterprise content for AI retrieval tasks.
The connectors support integration with vector search capabilities, enabling ingestion of content from Microsoft 365, Azure Cosmos DB, Azure SQL and Blob Storage. These sources can then be refined using OpenAI, Azure AI Language and other tools before being indexed as hybrid or vector-only search entries.
Microsoft said use cases include building retrieval-augmented generation (RAG) applications that pair user queries with relevant internal knowledge using custom data and AI models. The solution supports integration with Copilot Studio, allowing organizations to route indexed content into agents or other generative AI workflows.
All new capabilities are currently in preview and accessible via the Azure portal.