Public previewmicrosoft-foundryDay 2
Why it matters: Most agent calls are routine and don't need a frontier model, so routing is one of the easiest cost levers available. The real value is that you can now check that it works. Run an evaluation of the…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 2
Why it matters: SVG output means icons and illustrations can go straight into a design system without being redrawn, and markup editing cuts down on rerunning a whole prompt to fix one detail. Since both models cost…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 2
Why it matters: Turn-by-turn traces show whether a slow answer came from speech recognition, the model, a tool call or speech synthesis, which is what makes a voice agent supportable in a contact centre. Realtime Fl…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 1
Why it matters: The Agent 365 integration links the agent inventory and policies to the gateway that actually sits in the request path, so restrictions on which MCP tools an agent may call are enforced at runtime ra…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 1
Why it matters: On-behalf-of authentication is the security point: each tool call runs as the signed-in user, so an agent sees only what that person can open and audit trails show the real user. That fixes the share…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 1
Why it matters: This is the network-level control point for multi-model AI: Entra-based access, content safety, cost limits that actually stop requests, and a record of every model and tool call. Provider keys never…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 1
Why it matters: Switching or adding a model provider becomes a gateway change rather than an application rewrite, which lowers the cost of not committing to a single vendor. Combined with API Management's backend ro…
Microsoft FoundryGenerally availablemicrosoft-foundryDay 1
Why it matters: The governance angle is calibration: the model returns confidence scores and can abstain, so developers set thresholds for when to act, flag or escalate. That makes it a candidate control point insid…
Microsoft FoundryAnnouncedmicrosoft-foundryDay 1
Why it matters: Agents that change themselves over time are a governance question as much as a quality one. Microsoft’s answer is human approval: nothing changes in production until the team applies it, and skills a…
Microsoft FoundryPublic previewmicrosoft-foundryDay 1
Why it matters: Microsoft now has its own competitive image models in Foundry rather than relying only on partner models, and it is competing on price: Microsoft says 2.7 tops an independent image-editing ranking at…
Microsoft Foundry