Official news, first rumor reports and substantive progress. Each entry keeps the information published at that time.
- Microsoft outlines how Azure services give visibility and guardrails over AI agents in Foundry
A Microsoft Foundry blog post says developers are shipping agents faster than security teams can catalog them, and walks through how Azure services can establish visibility, guardrails and cost accountability as organizations move from pilots to running dozens or hundreds of agents.
- Microsoft argues production agents need more than a container on Kubernetes
A Microsoft Foundry blog post says enterprise teams moving their first agents to production often plan to containerize them and run them on an existing Kubernetes cluster, and explains why agents need more than a place to run.
- Microsoft explains voice agents in Foundry Agent Service: architecture, trust boundaries and latency
A Microsoft Foundry blog post says voice agents in Foundry Agent Service consolidate speech services, real-time models, turn detection and tool orchestration into a versioned, managed agent, and covers identity, trust boundaries and real-audio latency.
- Microsoft details using LLMs to optimize retrieval beyond query text
A Microsoft Foundry blog post describes using LLMs to move retrieval from matching query text toward understanding user intent, since users rarely phrase requests the way documents are indexed.
- Microsoft outlines how banks such as BNY balance agent autonomy and regulatory control
A Microsoft Foundry blog post breaks down the architectural choices banks make when deploying agents, from runtime hosting to network isolation, using BNY's enterprise-scale deployment as the example.
- MAI voice models arrive in LiveKit for expressive voice agents
Microsoft AI announced that its MAI models are now available in LiveKit, enabling developers to build expressive voice agents on the real-time platform, per the company's AI account.
- Microsoft AI chief: all MAI models are available today in Microsoft Foundry
Microsoft AI CEO Mustafa Suleyman announced that all of the company's models are available today in Microsoft Foundry. Community reports ahead of the post flagged the batch as including MAI-Transcribe-2-Streaming, MAI-Voice-2.1 and MAI-Voice-2.1-Flash, per TestingCatalog.