// tag
#Microsoft Foundry
19 posts
Making Agent Memory More Reliable, Transparent, and Production-Ready
Procedural memory helps agents retain successful execution patterns, while TTL, management tools, and STATE-Bench make memory easier to operate and measure.
Agent Optimizer Brings an Evaluation Loop to Foundry Agent Service
Foundry Agent Optimizer evaluates an agent, generates candidate configurations, ranks them, and promotes a winner without retraining the model.
From Building Agents to Working with Them: Enterprise Agent Distribution in Microsoft Foundry
Publishing, Autopilot agents, and Agent-to-Agent communication move Foundry agents into Microsoft 365 and Teams with identity and governance in the design.
Discovery to Execution: Scaling Agents with Toolboxes and Routines in Microsoft Foundry
How Tool Search, Skills, and Routines reduce the operational friction between an agent prototype and a production workflow.
Azure Content Understanding Makes the Document-to-Agent Path Shorter
Azure Content Understanding adds Foundry integration, GPT 5.2 extraction, native legacy Office formats, and Agent Framework, MarkItDown, LangChain, and Foundry IQ integrations.
Foundry’s Observability-to-ROI Story Is What Serious Agent Platforms Need
The latest Foundry observability announcement matters because it connects tracing, evaluation, optimization, and ROI into one operating loop for AI agents.
Foundry Managed Compute Makes Open Models a Model Deployment Problem
Foundry Managed Compute serves open-source and custom models on dedicated GPU capacity with managed runtimes, unified endpoints, monitoring, and accelerator-based billing.
Agent Harness, Hosted Agents, and CodeAct: This Is the Agent Framework Update I Would Focus On
The Build 2026 Agent Framework announcement is packed, but the most important threads are the harness model, Foundry hosted agents, and CodeAct for reducing orchestration overhead.
AI Alone Won’t Change the Business — the System Around It Will
Jay Parikh’s latest AI platform essay is worth reading because it frames the real enterprise challenge correctly: models matter, but the lasting value comes from the governed system that builds, contextualizes, runs, and improves agents over time.
The Hard Part of AI Development Is No Longer Access. It Is Operating the Right Model Well
A new Microsoft Foundry guide makes a strong case that model selection, cost control, evaluation, and lifecycle management are now the real differentiators in production AI systems.
The Agent Framework Sessions From Build 2026 Worth Following
The latest Microsoft Agent Framework Build guide is useful as a map of where the platform conversation is going: hosted agents, observability, governance, multi-agent design, and open interoperability.
Microsoft Foundry May 2026: The Updates I’d Actually Watch Closely
The latest Microsoft Foundry roundup covers a lot, but the most important threads are trace-based evaluation, new model choice, managed isolation, and the continued rise of local and production-grade agent tooling.
Claude Opus 4.8 in Foundry Is Another Sign That Model Choice Is Becoming a Platform Feature
Claude Opus 4.8 is now available in Microsoft Foundry. The important part is not just one more model launch, but the continued expansion of serious model choice inside a governed enterprise platform.
Model Router Evals Are the Step Too Many Teams Skip
The new Foundry model router evaluation repo matters because routing decisions need to be measured against quality, latency, and cost before teams treat automatic model selection as magic.
Microsoft Foundry June 2026: From Feature Drops to a Governed Agent Platform
June’s Foundry updates signal a platform transition: distribution, tooling, memory, observability, and optimization are converging into an enterprise-ready agent operations stack.
Claude GA in Foundry Is About Enterprise Plumbing, Not Model Hype
General availability matters because it resolves procurement, governance, and residency friction that blocks production AI.
OpenEnv + Foundry Push the Conversation Beyond Static Agents
The new OpenEnv and Foundry story is about much more than reinforcement learning buzzwords. It is really a push toward agent systems that can be evaluated, optimized, and improved over time against real business outcomes.
Foundry Toolboxes: One Endpoint for All Your Agent Tools
Microsoft Foundry just launched Toolboxes in public preview — a way to curate, manage, and expose AI agent tools through a single MCP-compatible endpoint without re-wiring everything per agent.
Where Should You Host Your AI Agents on Azure? A Practical Decision Guide
Azure offers six ways to host AI agents — from raw containers to fully managed Foundry Hosted Agents. Here's how to pick the right one for your .NET workload.