Browse Artificial Intelligence Community (177)

SurenderSinghMalik breaks down recent Azure App Service (Linux) changes that make Python deployments faster and more reliable for AI-heavy workloads, including new compression and packaging defaults, fewer expensive file operations, and client-side improvements that reduce transient deployment failures.

EfratNauerman announces a public preview update for the Azure Copilot Observability Agent in Azure Monitor, focused on using chat-driven investigations and exploration to speed up triage and root-cause analysis across logs, metrics, traces, and alerts in distributed systems.

vikas_gautam describes an end-to-end architecture for bringing Databricks Genie into Microsoft Teams using an Azure AI Foundry agent, focusing on what breaks in private, regulated environments and how to handle networking isolation, multi-hop identity, and per-user authorization when querying governed data.

Michael Flanakin summarizes FinOps toolkit 14, including a Copilot Studio agent template for querying FinOps hub data with KQL, a new recommendations pipeline that ingests Azure Advisor and Resource Graph results, a simplified hub deployment UI, and a preview dataset for commitment discount eligibility.

RavinderGupta outlines a “self-healing” CI/CD pattern where an agent observes Azure DevOps pipeline failures, uses Azure OpenAI (via Microsoft AI Foundry) to analyze build logs, and then proposes or applies fixes—such as updating Terraform for Azure Internal Load Balancer configuration—by opening a pull request for review.

syedarshad walks through a practical workflow for testing AI agents with LangSmith, using Azure OpenAI as the target model. The guide shows how to build an evaluation dataset, run LLM-as-judge scoring (correctness and hallucination checks), and interpret per-example and aggregate results with tracing and experiment views.

NandiniMuralidharan shows how to connect browser-harness to Playwright Workspaces so an AI coding agent can drive a real, cloud-hosted Chromium browser over CDP, enabling parallel, isolated sessions for tasks like scraping and interacting with JavaScript-heavy sites.

Jingwei Wang introduces “Open in VS Code” from Azure Copilot in the Azure Portal, a guided workflow that takes AI-generated Terraform configurations into an Azure-hosted VS Code environment so teams can validate, configure state backends, and deploy to Azure with fewer handoffs.

kinfey explains why AI agents running model-generated code need stronger isolation than standard containers, then walks through deploying a GitHub Copilot SDK agent on AKS using Kata Containers (kata-vm-isolation) plus layered hardening like seccomp, NetworkPolicy egress allowlists, and deny-by-default tool permissions.

vikas_gautam introduces PII Shield, a privacy proxy that sits in front of LLM calls to detect and anonymize PII (with optional reversal) so raw identifiers don’t leak through prompts, gateways, logs, or observability pipelines.

vyomnagrani explains why Microsoft built Azure AI Foundry Agent Service on Azure Container Apps, focusing on what changes when AI agents move from prototypes to production: bursty execution, long-running workflows, secure tool execution, isolation, state persistence, and the operational requirements for running agent fleets reliably at scale.

osmancokakoglu announces the winners of the AI Dev Days Hackathon and summarizes the projects and the Microsoft stack they used, including Azure AI Foundry, Azure OpenAI models, and the Microsoft Agent Framework, plus common Azure services and DevOps practices used to ship production-grade agentic apps.

SagarPatra explains how enterprise QA teams can use GitHub Copilot to reduce the mechanical overhead of writing and maintaining automated tests, while keeping trust through human review, governance, and intentional test design that supports reliable regression cycles.

shwetayadav explains how index-based Terraform for_each keys can trigger destructive disk churn on Azure, and shows a safer migration approach using stable keys plus terraform state mv, with a reusable GitHub Copilot skill to generate deterministic state-move commands.

mscagliola shows how to use GitHub Copilot skills for spec-driven development, turning a Medallion Architecture blog post into a repeatable repo that generates Terraform for Azure platform setup and Databricks bundle files for workloads, while enforcing strict placeholder/TODO rules to avoid invented environment values.

hcamposu announces Microsoft Host Integration Server (HIS) 2028 preview, outlining the move to .NET 10 (including Linux support for non-SNA features), new REST-based connectivity for DB2 and CICS/IMS workloads, and a set of deprecations aimed at removing legacy dependencies and improving security and hybrid operations.

SagarPatra explains how their QA team used GitHub Copilot as a practical assistant for test design, automation scaffolding, and maintenance work, while keeping human review and responsible AI practices non-negotiable.

Steven Bucher announces the public preview of the Azure Resource Manager MCP Server, a remote MCP server that lets AI agents query and operate on Azure resources via Azure Resource Manager and Azure Resource Graph, including generating KQL queries from natural language and deploying ARM templates from within VS Code.

divyanshi_varshney lays out a production-oriented reference architecture for running Azure OpenAI in regulated banking environments, focusing on private networking, identity-first access, RAG guardrails, and audit-ready observability. It also calls out common failure modes like AKS-to-Private Endpoint DNS issues and gaps in telemetry privacy.

hcamposu introduces the Logic Apps Migration Agent, an open-source, AI-assisted approach for migrating BizTalk Server (and other integration platforms) to Azure Logic Apps Standard, with a structured workflow, human review checkpoints, and a code-first experience via VS Code and GitHub Copilot.

tanyabaranwal outlines an event-driven Azure pipeline for extracting structured data from contract PDFs/ZIPs using Azure AI Document Intelligence, transforming results into a canonical JSON schema, and persisting them in Cosmos DB, with practical notes on observability and security.

KimVaddi lays out a reference architecture for governing “agent sprawl” with a multi-region AI agent landing zone on Azure, using layered control planes to enforce policy, safety, evaluation, and observability across agents, models, and tools.

Valerie Cutts and Jithin Jose explain how Azure’s Fairwater AI supercomputer network is designed to keep large synchronous training jobs running through routine faults, using Multipath Reliable Connection (MRC), a two-tier multi-plane topology, and static SRv6 source routing.

mohashaikh shows how to use GitHub Copilot Spaces plus a dedicated Markdown “engineering knowledge base” repo to make Copilot answer questions and generate code in line with your team’s standards, with optional in-repo instruction files and reusable prompt-file slash commands for consistent reviews.

gurkirat explains how GitHub Copilot can speed up Azure Landing Zone work by shifting engineers from writing Terraform and pipelines by hand to prompting for a structured draft and then reviewing it, with examples spanning management groups, networking, OIDC, GitHub Actions, and policy assignments.

Shruti9162 outlines an enterprise reference approach for running Azure AI Foundry inside private network boundaries, with Responsible AI controls treated as enforceable platform guardrails. The post focuses on VNet-integrated landing zones, private endpoints, content safety, and practical constraints like IP range allowlisting and service limitations in regulated environments.

sutandan explains spec-driven development as a more reliable alternative to the “prompt → retry → guess” loop when using AI coding tools, showing how a lightweight specification (inputs, outputs, constraints, edge cases) can make generated code more consistent for APIs and refactoring tasks.

ranjsharma outlines an approach for validating Azure infrastructure consistency by comparing an Excel “source of truth” against Terraform configuration and the actual deployed resources, producing a drift report that highlights missing resources and mismatched settings like region and SKU.

Devi Priya explains how GitHub Copilot Workspace supports intent-driven, multi-file refactoring across a repository, including a practical walkthrough that modernizes an app’s authentication flow and highlights planning, review, and adoption best practices.

varghesejoji introduces the Application Resilience Framework and a companion tool that turns architecture artifacts into a measurable resilience model. The guide walks through what to import, how to prioritize workflows and failure modes, how to validate mitigations (including chaos tests), and how to map risks to health signals and governance.

Valini Sunthwal describes a practical pattern for running multi-subscription Azure AI infrastructure with drift detection and “self-healing” using Terraform, multi-repo boundaries, and a daily reconciliation pipeline that cross-checks deployment metadata against Terraform state and a central registry.

sameeraman explains how Microsoft Discovery can automate a scientific simulation workflow using a coordinated set of AI agents, reducing manual scripting and job monitoring while keeping scientific decision-making with researchers.

mosiddi explains how the Agent Governance Toolkit (AGT) “shifts left” governance for AI agents by catching security and compliance violations before runtime, using pre-commit hooks, PR gates, CI checks, and release-time controls like SBOMs, signing, and provenance attestations.

HimanshuYadav explains how to modernize brownfield Terraform codebases by refactoring legacy modules to Azure Verified Modules (AVM) with AI assistance. The post focuses on using tools like GitHub Copilot to draft changes, then relying on disciplined Terraform plan review and policy gates to keep state changes safe.

Ravindra Kumar Vishwakarma explains how GitHub Copilot CLI can run as an Agent Client Protocol (ACP) server, enabling tools, IDEs, and CI/CD systems to connect to Copilot as a backend agent with streaming, sessions, and permissioned tool execution.

Pooja Pradhan outlines an Azure-focused approach to move from drift detection to diagnosis by combining IaC signals (Terraform/Bicep), Azure Resource Graph, and Activity Logs, then using an AI model to generate a human-readable root cause analysis with impact and recommended remediation steps.

junjieli announces updates to Foundry Toolkit that bring an end-to-end image generation workflow into VS Code, including discovering and deploying GPT-Image-2 to an Azure AI Foundry project, iterating in an Image Playground, and exporting ready-to-paste API code.

Valini Sunthwal describes a multi-repo Azure platform that uses Terraform, versioned releases, and daily reconciliation to detect and recover from infrastructure drift across many subscriptions. The post breaks down repo boundaries, pipeline design, drift detection tiers, and security practices like OIDC, Key Vault, and private endpoints.

Vineela Suri explains how to configure Azure SRE Agent with Azure Monitor to cut alert fatigue: use reinvestigation cooldowns, tiered response plans, and scheduled hygiene reports to consolidate noisy alerts, improve thresholds, and keep LLM token costs under control.

Vineela Suri explains how the Azure SRE Agent plugin marketplace works: teams publish plugins (skills + MCP connectors) to a shared GitHub repo, and any SRE Agent instance can discover and install them. The post walks through an AKS incident investigation example and the marketplace.json manifest structure.