Browse Artificial Intelligence Community (130)

Steven Truitt shares a hands-on walkthrough of using Microsoft Discovery as AI assistance to take a small chip-design project from a rough spec through C and HLS, into Verilog RTL, and finally to physical layout and basic timing/power optimization using open-source EDA tooling.

willchap describes a practical framework for evaluating scientific AI systems—execution, optimization, and exploration—using three RF engineering design challenges. The post shows how Microsoft Discovery orchestrates agentic workflows that combine analytical reasoning with HFSS simulations, while surfacing assumptions, trade-offs, and decision points for human guidance.

MinseokSong walks through a reference implementation that uses Microsoft Foundry and an Azure Function to classify incoming support emails, then uses Power Automate to validate the AI recommendation against a SharePoint allow-list, require human approval, and create an Outlook draft (without sending).

christinchen announces the general availability of Azure Copilot’s Troubleshooting Agent, a built-in experience in the Azure portal that helps teams investigate and resolve operational issues faster by combining resource context, supported diagnostics, and recommended actions—while keeping customers in control through identity and RBAC.

bharadwajr explains how the Azure team built GEM (Great Experiences Matter), an AI-enabled “feedback-to-fix” loop that aggregates customer feedback at scale, turns it into prioritized insights, and creates actionable Azure DevOps work items with an auditable evidence trail and human review.

kshatzkamer outlines a reference architecture for “orchestrated media intelligence”: an orchestration layer that connects trusted data, specialized AI models/agents, cloud infrastructure, and governance across end-to-end media workflows. The post uses Azure OpenAI, Azure AI Speech, and Azure AI Foundry examples to show how to keep humans accountable while scaling automation.

newmancheng explains how Microsoft Discovery combined agentic AI, computational predictors, and wet-lab feedback to design and validate a novel organic negolyte for redox flow batteries, with a focus on preserving negative results as durable knowledge that improves subsequent design rounds.

carlottacaste introduces “Inside Microsoft Foundry: Quickstart”, a video series focused on the practical work of building and shipping AI agents with Microsoft Foundry and the Microsoft Agent Framework, including model selection, grounding in your own data, tracing, evaluation, and moving from prototype to production.

rgarofalo walks through a hands-on build that locks down Azure AI Foundry both inbound (Private Endpoint) and outbound (agent egress via network injection), including the DNS zones you must create, how to prove public access is actually blocked, and how to troubleshoot opaque agent tool-call failures like 424 errors.

Christine Caggiano, Joshua Bradley, Steven Truitt, and William Chappell describe how Microsoft Discovery’s Discovery Engine adds CLIO (a self-adaptive “cognitive loop”) to improve agentic scientific problem-solving, and share benchmark results from Agent’s Last Exam showing higher scores and better consistency through multi-model, evidence-driven exploration.

pranabpaul walks through the first part of building a retail assistant demo using Microsoft Fabric and Azure AI Foundry, focusing on setting up the Azure environment, provisioning Fabric capacity, loading sample data into a Lakehouse, creating an ontology, and wiring a Fabric Data Agent to query that data.

pranabpaul continues the 3IQ Retail Assistant demo by wiring unstructured PDF content into an Azure AI Foundry agent using Azure Blob Storage and Azure AI Search, including the RBAC permissions needed for managed identities and the networking trade-offs involved when publishing the agent to Microsoft 365 Copilot.

Mohammed Barqawi introduces Logic App Storage Inspector, a read-only Kudu site extension that helps troubleshoot Azure Logic Apps Standard by inspecting the underlying storage for workflow history, versions, and queue health, with guidance on installation, configuration settings, and managed identity permissions.

Dalibor Kovacevic explains how to use Azure SRE Agent’s advanced ServiceNow filtering to route the right incidents to the right response plan, then control how much autonomy the responder has (Autonomous vs Review) while you preview and tune matching before enabling it.

yairgil explains how the Azure Copilot Observability Agent can use Azure Monitor Log Analytics Basic and Auxiliary table plans as investigation evidence, helping teams reduce log costs while still correlating alerts, metrics, traces, topology, and recent changes during incident response—especially in high-volume AKS environments.

kinfey explains why “AI agent infrastructure” needs stronger security boundaries than traditional model hosting, using kars (an open-source Agent Reference Stack for Kubernetes) to show how to isolate agent runtimes, broker identity, govern tools/MCP, control egress, enforce token budgets, and produce auditable evidence for incident response.

reshmarahim announces the public preview of Radius Canvas for the GitHub Copilot app, a Canvas extension that builds a versioned application model from a repository so developers can visualize architecture, review PR impact via an application-level diff, and deploy across environments using generated GitHub Actions workflows with OIDC-based credentials.

budzynski summarizes the August 2026 updates to the AI Gateway tier (preview) of Azure API Management, focusing on improved OpenTelemetry-based observability for AI workloads and new controls for monitoring and enforcing model spend via budgets and cost-limit policies.

Amit Ganguli announces the public preview of the Resiliency Agent in Azure Copilot, which helps teams assess zonal resiliency across subscriptions, prioritize fixes with cost-aware indicators, and generate deployment-ready Bicep/Terraform templates and scripts to remediate gaps.

junjieli’s August 2026 roundup for Foundry Toolkit for VS Code covers new end-to-end agent-building workflows: improved “Create Agent” entry points, centrally managed MCP toolboxes, preview Agent-to-Agent connections, user-scoped hosted sessions, and deeper run inspection with latency and token breakdowns.

AbhinavPremsekhar explains how to run authenticated browser workflows for Azure AI Foundry agents without exposing credentials to the model, using Browser Automation Tool (BAT) with Playwright Workspaces and Azure Key Vault. The post focuses on secret storage, RBAC access, just-in-time retrieval, and practical security guardrails.

demiajayi announces new Azure Cost Management capabilities in the Azure Resource Manager (ARM) MCP server, including default cost and pricing tools plus an optional CostManagement toolset. The post shows how to enable the toolset for GitHub Copilot Chat in VS Code and Copilot CLI, and what scenarios the tools support.

Sherry Xu, Prashant Ranjan, and Torsten Hoefler explain how Azure Maia 200 targets efficient, predictable large-scale AI inference by making data movement explicit (SDLA) and extending that approach across an all-Ethernet scale-up network, with performance discussion across realistic matmul and collective-communication workloads.

yashnisar introduces Live Reports in Azure SRE Agent (public preview), a way for ops teams to describe an operational dashboard in chat and have the agent generate a deterministic page that refreshes live data on each open. The post explains token-cost behavior, connector/tool-call mechanics, and the guardrails used to keep report execution constrained.

sanchitmehta (with co-author visagarwal) explains how Azure SRE Agent was redesigned to make autonomous operations safer by moving enforcement outside the agent’s reach—using microVM sandboxing, secretless authentication, output scrubbing, and risk-based approval so the environment, not the prompt, becomes the policy boundary.

WSilveira explains how the MCP Connectors canvas extension for the GitHub Copilot app lets agents use hosted MCP servers from Azure Connector Namespace without manually wiring endpoints, headers, or local proxies. It also covers how the user-scoped MCP config works across the Copilot app and Copilot CLI, plus key security details.

ronitauber explains how to turn Azure Monitor issues created by the Azure Copilot Observability Agent into downstream operational workflows using Action Groups. The post shows patterns for pushing rich issue context into Microsoft Teams, Azure DevOps, and custom automation via Logic Apps, Azure Functions, and event streaming.

Pamela Fox shows how to build a browser-using QA agent by combining Pydantic AI with Playwright, backed by Microsoft Foundry models. The post covers keyless Entra authentication, safe browsing constraints, and OpenTelemetry tracing with optional export to Azure Application Insights.

Lee Stott wraps up the FibreOps series with a practical look at running autonomous agent systems in production: adding hands-free spoken updates via Azure AI Voice Live, and building an observability stack with OpenTelemetry and Application Insights so agent decisions can be traced, evaluated, and audited.

Dirk Brinkmann explains a practical FinOps pattern for attributing AI spend (tokens, Copilot, and agent meters) when cost data is scattered across inconsistent meter categories and free-text meter names. The approach builds a customer-owned “meter map” from the Azure price sheet and joins it to FOCUS cost rows to enable reliable reporting.

supriyas breaks down how to choose between Prompt Agents and Hosted Agents in Microsoft Foundry Agent Service, focusing on the practical architecture question: where orchestration, state, and operations should live. The post outlines decision criteria, warning signs, and enterprise runtime concerns like identity, content safety, tracing, and cost controls.

Chris Noring explains how to govern GitHub Copilot spend in an enterprise by separating seat assignment, cost-center attribution, included AI credit boundaries, paid-usage budgets, and per-user limits. The article walks through a practical end-to-end model with two cost centers (Business and Developers) and shows how to avoid common misunderstandings.

artigulwadi announces an August 2026 Azure Copilot update that lets customers invoke specialized Azure Copilot agents directly (instead of starting from a single general chat), with new admin controls for enabling agents per tenant and clarified pricing for the agent portfolio.

lily-ma explains how to extend Azure SRE Agent with MCP servers by hosting them in Azure Connector Namespace, then wiring them into the agent’s connector experience. The post includes an end-to-end example using Azure Developer CLI (azd), managed identity auth, and access policies to connect a hosted SQL MCP server.

Abhinav Premsekhar explains why cloud-hosted browsers are becoming a core platform capability for running browser automation and AI agents at scale, with a focus on the operational requirements enterprises hit in production: isolation, governance, observability, reproducibility, and geographic execution.

Harel Broitman explains how to connect Azure Monitor issues with ServiceNow ITOM so incidents and alerts stay linked and status changes sync both ways during an active response, reducing manual copy/paste and keeping technical investigation context aligned with ITSM workflows.

Lee Stott explains how to take agents from development to production in Microsoft Foundry Agent Service, covering Hosted Agents (container + manifest deployment), the Agent Optimizer, deterministic Routines, Memory, Toolboxes, and end-to-end tracing with OpenTelemetry and Application Insights.

Arturo Quiroga explains how real-world usage reshaped the open-source Azure Architecture Diagram Builder into a guided workflow (Create → Refine → Validate & Improve → Share or Build), with concrete changes around preserving human edits, improving diagram integrity, iterating Well-Architected validation, and generating export/IaC artifacts with clear limitations.

ceciliaxia outlines how Microsoft is co-designing Azure AI infrastructure end-to-end—from accelerators and CPUs to rack power, cooling, and networking—to improve performance per dollar and per watt for large-scale inference and agentic workloads, and shares related open standards work with the Open Compute Project ecosystem.

Pamela Fox walks through practical MCP server designs for giving coding agents safe access to a PostgreSQL database, from free-form SQL to fully templated tools. She highlights where flexibility creates risk, and shows concrete guardrails like schema discovery patterns, read-only enforcement, and confirmation flows for destructive actions.