Awesome Azure OpenAI & Copilot

Azure OpenAI is Microsoft's cloud service for deploying OpenAI models, and Microsoft Copilot is the family of AI assistants and platforms built into Microsoft products and developer workflows.

Contents

• Azure OpenAI & Foundry Overview
• Orchestration Frameworks
• Prompt Engineering & Tooling
• Agent Frameworks
• Model Training & Inference
• Safety, Security & LLMOps
• Data Processing & Memory
• Dev Tools, MCP & Extensions
• Copilot Product Catalog
• Microsoft Foundry & AI Services
• Azure AI Search
• Agent Development
• Microsoft 365 Agent Development
• Learning Resources & Workshops
• Microsoft Research
• Sample Applications
• Solution Accelerators
• Code Samples & Workshops
• Architecture Patterns & Use Cases

Azure OpenAI & Foundry Overview

• Abuse Monitoring - Data handling and retention policy for abuse detection in Azure OpenAI.
• Azure OpenAI Models - Catalog of foundation and reasoning models available from Azure OpenAI.
• OpenAI Models - Catalog of foundation and reasoning models available from OpenAI.
• What Is Azure OpenAI Service? - Overview of Azure-hosted OpenAI capabilities, security and integration features.

Orchestration Frameworks

• Artificial Intelligence Controller Interface (AICI) - Secure sandboxed controllers for structured real-time LLM output control.
• Azure ML Prompt Flow - Visual designer for prompt orchestration and evaluation supporting Jinja templates.
• Microsoft Agent Framework - Unified SDK (Semantic Kernel & AutoGen), Graph-based Workflows, DevUI, Built-in connectors, MCP + A2A + OpenAPI.
• Semantic Kernel - SDK for orchestrating AI services, functions, memory, and planners across languages.

Prompt Engineering & Tooling

• GenAIScript - JavaScript framework to orchestrate LLM calls, tools, and data in unified scripts.
• guidance - DSL for structured prompting and controlled generation with chain-of-thought patterns.
• LLMLingua - Prompt and KV-cache compression achieving substantial token reduction with minimal quality loss.
• LMOps - Toolkit for optimizing and evaluating prompt quality across multimodal scenarios.
• PromptBench - Unified framework for large language model prompt evaluation.
• Prompt Engine - Utilities for systematic prompt authoring and experimentation (Python variant available).
• Promptions - "Prompt" plus "options". Ephemeral UI for prompt refinement — a flexible dynamic prompt middleware.
• Prompty - Template language for prompt definition.
• SAMMO - Framework for automated prompt optimization through search and evaluation loops.
• TypeChat - Schema-driven natural language interfaces using strongly typed APIs.

Agent Frameworks

• AGDebugger - Visual tool for AutoGen-based multi-agent debugging.
• Agent Harness: Scaling Capabilities - Adds skills, shell access, CodeAct, and background agents to a harness.
• Agent Lightning - Train and optimize agents (including RL, prompt‑optimization, multi‑agent workflows).
• AIOpsLab - Research-driven AI agents for cloud incident analysis and root cause automation.
• AutoGen - Customizable multi-agent conversation and tool orchestration framework (community AG2 and Microsoft variants).
• Copilot SDK - Multi-platform SDK for integrating GitHub Copilot Agent into apps and services.
• ExACT - Adaptive retrieval and planning using interaction memory and Monte Carlo Tree Search.
• JARVIS - Bridge connecting LLMs with specialized AI models for composite task execution.
• Magentic-Marketplace - Market simulation using agents.
• Magentic-One - Generalist multi-agent system for complex multi-step automation.
• OmniParser - Vision-based GUI parsing for screen-grounded agent control.
• Qlib - Quantitative finance platform integrating supervised, temporal, and reinforcement modeling.
• RD-Agent - Open-source R&D automation for structured experimentation workflows.
• Semantic Workbench - Development environment for intelligent multi-agent scenario design and debugging.
• TaskWeaver - Code-first natural language to executable workflow translation with rich data planning.
• TinyTroupe - Multi-agent persona simulation for scenario exploration and insight generation.
• TypeAgent - Personal-agent architecture using structured prompting, action schemas, and Structured RAG memory.
• UFO - UI-centric Windows interaction agent for system-level task automation.
• Vanilla AI Agents - Minimal agent patterns without heavy framework abstractions.
• Webwright - Terminal-driven browser agent framework with Playwright and reusable scripts.
• Windows Agent Arena - Benchmarking and evaluation platform for multimodal Windows desktop agents.

Model Training & Inference

• BitNet - Inference framework for ultra-low precision (1-bit) large language models.
• DeepSpeed - Distributed training and inference optimization library featuring ZeRO optimizations.
• FLAML - Lightweight automation for model selection, tuning, and economical experimentation.
• PixelCraft - High-fidelity visual reasoning on structured images.
• Tutel - Optimized Mixture-of-Experts (MoE) library with FP8/NVFP4/MXFP4 support.
• UniLM (Foundation Models) - Collection of large-scale pretrained architectures for multimodal and language tasks.

Safety, Security & LLMOps

• Agent Governance Toolkit - Policy enforcement, identity, and sandboxing for secure enterprise agents.
• AI Central - Control plane for authenticated, resilient access to multiple OpenAI endpoints.
• Azure AI Content Safety - Managed text and image moderation service with APIs for harmful-content analysis, groundedness detection, protected material, and custom categories.
• Azure AI Evaluation SDK - Evaluation tooling for generative AI quality and safety metrics.
• EvalsforAgentsInterop - Evaluates LLM-powered productivity agents through realistic scenario simulations.
• Foundry Control Plane and Agent 365 - Explains developer and IT governance control planes: Agent 365 → Purview → Defender → Foundry Guardrails → APIM → Content Safety → models.
• MCP Enterprise Authorization - Explains Entra-governed MCP authorization and the EMA ID-JAG protocol.
• MDASH - Multi-model agentic security harness for vulnerability discovery and proof.
• Microsoft Entra Agent ID - Central registration and governance for organizational AI agents.
• Not All Model Upgrades Are Upgrades - Shows why model changes need workload-specific cost and quality evaluation.
• Presidio - Context-aware, pluggable PII detection and de-identification for text and images.
• Presidio Research - Data science tools for developing and evaluating Presidio PII recognizers.
• Prompt Shields - Azure AI Content Safety protection for direct user-prompt attacks and indirect prompt injection in third-party documents.
• PyRIT - Risk identification toolkit targeting robustness, safety, and adversarial prompt detection.
• Securing MCP: A Control Plane for Agent Tool Execution - Introduces Agent Governance Toolkit for MCP tool definition scanning, policy enforcement, response inspection, identity, and audit logs.

Data Processing & Memory

• Azure HorizonDB - New PostgreSQL service with advanced DiskANN vector indexing and large-scale replication.
• Data Formulator - Iterative AI-assisted creation and transformation of data visualizations.
• DiskANN - Graph-based indices for scalable and fast approximate nearest neighbor search.
• Kernel Memory - Service and plugins for scalable ingestion, hybrid embedding, and memory pipelines.
• MarkItDown - Conversion utility for office and structured documents to Markdown.
• Memora - Harmonic agent-memory representation balancing abstraction with specificity.
• Microsoft Fabric - Unified platform combining data integration, engineering, warehousing, and BI.
• NLWeb - Conversational interfaces for web data with MCP, Schema.org, and RSS support.

Dev Tools, MCP & Extensions

• ACA Secure MCP Server OAuth 2.1 - Secures MCP servers on Azure Container Apps with Entra ID.
• Agent 365 DevTools CLI - CLI for developing, deploying, and managing Microsoft Agent 365 applications.
• Agent Harnesses in VS Code - VS Code blog on agent harnesses for GitHub Copilot.
• Agent Skills - 131 skills, MCP configs, and agents for Azure SDK and AI Foundry development.
• AIShell - Interactive PowerShell shell integrating AI-assisted command workflows.
• AI Toolkit for VS Code - VS Code extension integrating local and cloud AI workflows.
• APM - Agent Package Manager - Dependency manager for AI agents — declare skills, prompts, and tools in apm.yml.
• Awesome GitHub Copilot - Community catalog of Copilot agents, instructions, skills, hooks, workflows, plugins, tools, and learning resources.
• Azure Communication Services WebChat Adapter - Bot Framework WebChat adapter for Azure Communication Services Chat.
• Azure Functions MCP Extension - Azure Functions extension for hosting MCP server endpoints.
• Azure Icons - Searchable Azure architecture icon catalog sourced from Azure icon set.
• Azure Skills Plugin - Official agent plugin with Azure skills and MCP server configurations.
• Chat Customizations Evaluations - VS Code extension for LLM-powered analysis of prompt, agent, skill, and instruction files.
• Claude Code + MS Foundry - Claude Code + Microsoft Foundry enterprise setup on Azure.
• Copilot Studio Extension for Visual Studio Code - VS Code extension for authoring and syncing Copilot Studio agents.
• Cosmos DB Agent Kit - A collection of skills for AI coding agents working with Azure Cosmos DB.
• ExcelMcp - MCP Server for Microsoft Excel - MCP server that lets AI assistants (e.g. via GitHub Copilot or other LLM-based agents) control Microsoft Excel — automate Power Query, VBA, DAX, formatting, and more.
• Foundry-Local - Local-first dev setup for Microsoft Foundry projects.
• GitHub App - GitHub's MCP-enabled application surface for agent integrations.
• GitHub Copilot Chat for VS Code - In-editor contextual multi-turn AI assistance.
• GitHub Copilot CLI - AI-powered coding assistance in your terminal.
• GitHub Copilot Office Add-in - GitHub Copilot add-in for Word, Excel, and PowerPoint.
• Hypervelocity Engineering (HVE) Core - Hypervelocity Engineering assets: agents, instructions, prompts, and Copilot workflows.
• Intelligent Terminal - AI-enhanced terminal experience from Microsoft.
• Introducing Skills in Azure API Center - Govern reusable AI skills alongside APIs and MCP servers.
• MCP Gateway - Reverse proxy and management layer for MCP servers with Kubernetes lifecycle support.
• Microsoft App Icons - Searchable Microsoft app icon catalog.
• Microsoft MCP - Catalog of official Microsoft MCP (Model Context Protocol) server implementations for AI data access and tool integration.
• .NET Agent Skills - Skill packs and templates for .NET-focused AI coding and agent workflows.
• OfficeCLI - Agent-ready CLI for reading and automating Office documents.
• Playwright CLI for AI Agents - Browser automation CLI with lower token cost than MCP browser sessions.
• Power BI Modeling MCP Server - Power BI semantic modeling capabilities to your AI agents through a local MCP server.
• Power Platform Skills - Skill marketplace for Claude Code, Copilot, Power Apps, and Power Pages.
• Rayfin - AI-first SDK/CLI for code-defined Fabric backends with data, identity, APIs, and governance.
• Stateless MCP on App Service - Explains MCP spec changes for scalable App Service hosting.
• SWE-bench-Live - Evaluating your AI system on latest software engineering tasks. (NeurIPS 2025 D&B).
• Wassette - Runs WebAssembly Components as MCP tools with Wasmtime sandbox isolation.
• Waza - Go CLI for creating, testing, measuring, and improving agent skills and custom agents.
• WorkIQ CLI and MCP - MCP Server and CLI for accessing Work IQ.

Copilot Product Catalog

• A Closer Look at Work IQ - Microsoft 365 Copilot intelligence layer combining tenant data, semantic index, memory, skills, tools, and business context.
• Agent 365 - Control plane for AI agents (registry, access control, visualization, interoperability, security).
• Agent 365 Governance Maturity Heatmap - Maps Agent 365 governance maturity across Entra, Defender, and Purview.
• Agent 365: The Identity-First Control Plane - Identity-first control plane for governing enterprise AI agents.
• Azure Copilot Migration Agent - Azure Migrate agent for discovery, 6R recommendations, landing-zone automation, wave planning, and GitHub Copilot modernization handoff.
• Azure SRE Agent - A cloud agent service. Site reliability engineering (SRE) covers Usage, Incident, and Performance Monitoring.
• Azure SRE Agent GitHub Repository - Official sample of Azure SRE Agent.
• Build 2026 Keynote - Microsoft Build keynote announcements across agents, Copilot, and cloud.
• Clarity Copilot - Analytics insights assistant for session telemetry.
• Copilot Checkout and Brand Agents - In-chat purchases and automated brand interactions.
• Copilot Health - Health-focused Copilot combining records, wearables, and proactive guidance.
• Copilot in Azure Quantum - Quantum workload guidance with Copilot integration.
• Copilot in Windows - System-level AI integration in Windows.
• Copilot Pages - Persistent collaborative canvas in Copilot chat.
• Copilot+ PCs - Windows PCs with integrated NPUs for AI workloads.
• Copilot Pro - Premium Copilot features, model priority, and M365 integration.
• Copilot Scenario Library - Examples of Copilot business use cases.
• Copilot Tasks - Copilot executes delegated tasks like booking, research, and purchasing.
• Copilot Usage Report 2025 - Health queries dominate mobile use, programming peaks on weekdays and gaming on weekends, and February sees a Valentine's Day spike in relationship and personal growth conversations.
• Copilot Vision - Edge experience for browsing with on-page visual grounding.
• Cowork Investment Assessment - Estimates Microsoft 365 Copilot Cowork value from Work IQ signals. Guide
• Dynamics 365 Copilot - Embedded generative assistance across Dynamics workloads.
• Fabric Copilot - Copilot features for Fabric.
• GitHub Copilot - AI pair programming for code completion and chat.
• How Many Products Does Microsoft Have Named 'Copilot'? - Maps 81 Microsoft Copilot-branded products and features.
• Introducing Multi-Model Intelligence in Researcher - Researcher adds Critique and Council workflows that combine generation, review, and model comparison for higher-quality deep research.
• Microsoft 365 Copilot - Cross-suite productivity AI assistance.
• Microsoft 365 Copilot Chat - Chat experience (free tier) and pay-as-you-go agents.
• Microsoft 365 Copilot ROI Calculator (honest defaults) - Interactive, skeptical ROI model for Microsoft 365 Copilot with adoption-rate and time-savings sensitivity analysis.
• Microsoft Already Has 80 Different Products Called Copilot - Overview of Microsoft's expanding Copilot product naming footprint.
• Microsoft Copilot - Web-based AI assistant (formerly Bing Chat Enterprise).
• Microsoft Copilot for Azure - AI-assisted cloud operations for management and optimization.
• Microsoft Scout - Microsoft 365's first Autopilot: an always-on personal agent that works across Teams, Outlook, OneDrive, SharePoint, desktop, web, and MCP servers under enterprise controls.
• Nuance DAX Copilot - Clinical documentation automation.
• Power Apps AI Copilot - Copilot feature in Power Apps AI.
• Power Automate Copilot - Copilot feature in Power Automate cloud flows.
• Power BI Copilot - Copilot features in Power BI.
• Process Mining Ingestion - Copilot feature for process mining ingestion in Power Automate.
• Security Copilot - Security operations copiloting for defenders.
• Team Copilot - Meeting, collaboration, and project facilitation.
• Viva Copilot - Employee experience and engagement assistance.
• Web IQ - Bing announces Web IQ for grounding Copilot and agents.
• Windows AI Foundry - Platform for local and hybrid AI app development (evolved from Windows Copilot Runtime).
• Windows Copilot Runtime - The set of APIs powered by the 40+ on-device models.
• Word, Excel, and PowerPoint Agents in Microsoft 365 - Dedicated agents that help users create documents, spreadsheets, and presentations.
• Work IQ - Enterprise intelligence layer behind Microsoft 365 Copilot agents (Fabric IQ, Foundry IQ).

Microsoft Foundry & AI Services

• Artificial Intelligence Architecture Design - Overview of Azure AI services and solution patterns.
• Assistants API - Code Interpreter, Function calling, Knowledge retrieval tool, and Threads.
• Azure AI Agent Service - Managed agent service in Azure AI Foundry.
• Azure AI Content Understanding - Transforming Multimodal Data into Insights.
• Azure AI Document Intelligence - OCR and document-understanding service for extracting text, tables, structure, and key/value pairs with prebuilt and custom models for intelligent document processing.
• Azure AI Foundry - Portal for building, evaluating, and managing AI applications and agents.
• Azure AI Foundry Agent Service - Managed service for creating enterprise-grade, multi-agent systems.
• Azure AI Foundry Fine-Tuning - Reinforcement Fine-Tuning (RFT) with o4-mini.
• Azure AI Foundry SDK - Unified SDK with pre-built app templates.
• Azure AI Foundry vs Azure OpenAI: 2026 Decision - Practitioner decision guide for choosing between Azure AI Foundry and Azure OpenAI.
• Azure Blob Storage and Run:ai Model Streamer - Streams model weights from Blob Storage to cut cold starts.
• Azure OpenAI PTU vs PAYG Break-even - Cost comparison for Azure OpenAI provisioned throughput and pay-as-you-go.
• Choose the Right Speech and Voice Model - Decision guide for Foundry speech, voice, transcription, and realtime models.
• Foundry IQ Knowledge Bases - Improve enterprise recall for agents.
• Introducing Deep Research in Azure AI Foundry Agent Service - Deep Research capabilities in Azure AI Foundry Agent Service.
• Introducing Toolboxes in Foundry - Centrally curates reusable agent tools behind one MCP endpoint.
• Microsoft Foundry - Azure AI Foundry rebranded as Microsoft Foundry. Blog with recent updates.
• Responses API - Chat Completions API with the advanced tool-calling capabilities.
• Weights & Biases with Azure - Fine-Tuning with Weights & Biases and Azure OpenAI.

Azure AI Search

• Agentic Retrieval in Azure AI Search - An automated query engine that creates and runs its own retrieval plan for relevant results to complex questions.
• Azure AI Search Blog - News, best practices, and updates for Azure AI Search.
• Azure AI Search Demos using Jupyter Notebook - Jupyter notebooks that demonstrate vector search, hybrid search, image search, RAG, and evaluation.
• Azure AI Search Documentation - Official product documentation and tutorials for Azure AI Search.
• Azure AI Search Multimodal Sample - Demonstrates multimodal ingestion and retrieval patterns.
• Azure AI Search OpenAI Purview Data Security Demo - Demo integrating Azure Search, OpenAI, and Purview data security.
• Azure AI Search: Outperforming Vector Search With Hybrid Retrieval and Reranking - Vector search with hybrid retrieval and reranking.
• Azure AI Search Power Skills - Custom skills for Azure AI Search.
• Azure AI Search Python Playground - A collection of Jupyter notebooks designed to explore the various capabilities of Azure AI Search.
• Azure AI Search Samples - Official sample code for Azure AI Search.
• Document-Level Access Control - Document-level access helps restrict content visibility to authorized users, based on predefined access rules.
• Integrated Vectorization - Automatically splits documents into chunks, creates embeddings with Azure OpenAI, maps them to an Azure AI Search index, and automates query vectorization.
• Python Samples for Azure AI Search - Python samples used in Azure AI Search Documentation.
• RAG-Knowledge - Sample code for Data Preprocessing, Chunking, Index Design, Query Optimization.
• Relevance Scoring in Hybrid Search - Hybrid search with Reciprocal Rank Fusion (RRF).
• REST Examples for Azure AI Search - REST sample requests for search features.
• SPANN - Memory-disk hybrid ANN index outperforming DiskANN on billion-scale datasets.
• SPTAG - Vector search library behind SPANN and billion-scale ANN retrieval.
• Vector Search Sample Code - Sample code demonstrating Azure AI Search vector capabilities.

Agent Development

• Adding Aspire to a Python RAG Application - Use Aspire for local orchestration and observability in Python RAG apps.
• AgenticOps with AKS-Lab-GitHubCopilot - Field guide for scoped Copilot custom coding agents.
• Agentic Patterns Playbook - Playbook for Microsoft-oriented agent design patterns.
• Agentic Platform Engineering - Platform engineering patterns using GitHub Copilot agents and golden paths.
• AI BAST Agents Library - AI BAST Specialist Team agents library. Rapid Agent Prototype Pattern (RAPP).
• Aspire Azure Enterprise Networking - Enterprise networking guidance for Aspire-hosted Azure applications.
• Building a Real-Time Multi-Agent UI with AG-UI and Microsoft Agent Framework Workflows - MAF Python workflow sample streaming handoffs, interrupts, and human approvals to a React UI via AG-UI.
• Build Long-Running AI Agents on Azure App Service with Microsoft Agent Framework - Multi-Agent Workflow Travel Planner With WebJob.
• Deploy MAF agents with Foundry Hosted Agents - Production deployment path for Microsoft Agent Framework agents.
• Foundry Agents and Custom Engine Agents through the Corporate Firewall - APIM/YARP pattern for exposing private Foundry and custom agents to Teams and Copilot with Bot JWT validation.
• FoundryIQ-and-Agent-Framework-demo - Multi-agent demo using Microsoft Agent Framework SDK and Azure AI Foundry with FoundryIQ.
• Hybrid AI Agents in Python - Routes private or simple prompts locally and complex prompts to Foundry.
• Microsoft Agent Factory - Program for building agents with Work IQ, Fabric IQ, Foundry IQ via Microsoft Foundry & Copilot Studio.
• Microsoft Discovery - Graph-based scientific co-reasoning. Specialized discovery agents for conducting research.
• Microsoft Discovery Community Repository - Community repository for Microsoft Discovery resources and collaboration.
• Microsoft Global Black Belt GitHub - GitHub Organization for Azure AI patterns, accelerators, and field engineering samples.
• Project-Nighthawk - AI-powered deep technical research pipeline for AKS and Azure Red Hat OpenShift.
• ZavaShop on AKS + ACA — built by GitHub Copilo - ZavaShop multi-agent retail lab using MAF, Copilot SDK, AKS, and ACA.

Microsoft 365 Agent Development

• Agent 365 SDK - Microsoft Agent 365 extensions for .NET, Node.js/TypeScript, and Python.
• Agent 365 SDK Samples and Prompts - Sample scenarios for Microsoft Agent 365.
• Agent 365 Skills - Coding skills for Agent 365, Work IQ, and OpenTelemetry.
• Agent Builder & Plan Designer in Power Apps - Design and configure intelligent solutions in Power Apps.
• Agent Evaluation in Microsoft Copilot Studio - Structured, automated testing solution directly in Copilot Studio.
• Agents in Power Platform - AI first no-code development with Copilot agents in Power Platform.
• Combine Copilot Agents with Office Add-ins - Enhance Copilot with Office add-in actions for natural language interaction with Office documents.
• Community-Contributed Samples for Microsoft 365 Copilot - Samples for extending Microsoft 365 Copilot features.
• Copilot Prompts - Prompt examples for Microsoft Copilot.
• Copilot Studio March 2025 Update - Added autonomous agents, deep reasoning, and MCP integration.
• Copilot Studio Monthly Updates - Copilot Studio news and features for the last month.
• Copilot Tuning - Low-code model tuning and grounded agent creation.
• Deploying Agentic Services to Microsoft 365 Copilot - Gateway and OBO pattern for existing agentic services.
• Introducing Copilot Actions, new agents, Copilot Control System - New agent capabilities and management tools for IT teams.
• Microsoft 365 Agents SDK - The evolution of Bot Framework for building AI agents, with components for .NET, Node.js/TypeScript, and Python.
• Microsoft 365 Copilot APIs Client Libraries - Client libraries for Microsoft 365 Copilot APIs.
• Microsoft 365 Copilot App Builder and Workflows - Build working apps & workflows using natural language.
• Microsoft Copilot Studio - Customization and orchestration environment (formerly Power Virtual Agents).
• Power Apps MCP closed-loop learning - Agent corrections become organization-scoped memory and optimized instructions.
• Power CAT - Microsoft Power CAT tools for Power Platform and Copilot Studio adoption.
• PowerPlatform-DataverseClient-Python - Python client for Microsoft Dataverse — CRUD, SQL queries, and file uploads.
• Power Platform Prompts - Prompt examples for GPT-powered Power Platform workflows.
• SharePoint Agent - Turn SharePoint sites and documents into interactive agents.
• SharePoint Copilot Apps - Builds interactive SPFx components directly into the Copilot canvas.
• Teams SDK - Both Teams SDK and Microsoft 365 Agents SDK originate from the legacy, multi-channel Bot Framework. Teams SDK succeeds the deprecated Teams AI Library v1 and focuses on Teams extensibility—Adaptive Cards, tabs, dialogs, message extensions, meetings, and SSO—whereas Microsoft 365 Agents SDK is the multi-channel evolution of Bot Framework.

Learning Resources & Workshops

• AI Agents for Beginners - A Course - Ten-lesson introduction to building AI agents for Beginners.
• AI Developer - Challenges for Learning Azure AI Foundry and Semantic Kernel Fundamentals.
• AI Fluency - Learning resources and challenges for AI fluency.
• AI Playbook - Enterprise AI solutions, capabilities, and code for real-world problems.
• Automated Compliance Agents Hackathon - Hands-on hackathon challenges for multi-agent financial compliance workflows.
• Azure AI Foundry & Agents Workshop - Workshop materials for building agents with Azure AI Foundry.
• Build AI Solutions with Azure AI Foundry - Sample code and materials for building generative-AI solutions with Azure AI Foundry / Azure AI Studio, including prompt-engineering examples and workflows.
• Copilot Developer Camp - Hands-on labs for extending Microsoft 365 Copilot and building custom agents.
• Copilot Studio Labs - Hands-on labs for building AI agents with Microsoft Copilot Studio.
• Copilot System Overview (video) - Copilot Architecture explanation video.
• EdgeAI for Beginners - Small Language Models (SLMs) optimized samples for On-device inference.
• Forgebook - Community cookbook of practical Microsoft Foundry notebooks.
• LinkedIn Learning - Official exercise-file repositories for LinkedIn Learning courses.
• Microsoft AI Tour - A session and workshop how to best leverage AI.
• Model Context Protocol Curriculum for Beginners - Introductory MCP course.
• Model Mondays - Weekly livestream and office hours on generative AI models and model selection.
• Partner Resources: AI & ML Academy - Comprehensive learning resources for partners and developers.
• Semantic Kernel Cookbook - Example recipes and tutorials for working with Semantic Kernel.

Microsoft Research

• Arbor - Research agent project from RUC-NLPIR.
• AsyncThink - Asynchronous thinking protocol with organizer/worker fork–join actions extending sequential and parallel paradigms.
• AutoGen Studio - A No-Code Developer Tool for Building and Debugging Multi-Agent Systems.
• Azure AI Foundry Labs - Research driven tools and experimental projects.
• Be My Eyes - A modular, multi-agent framework that extends LLMs to multimodal reasoning by coordinating efficient VLM perceivers (Qwen2.5-VL-7B) and powerful LLM reasoners (text-only DeepSeek-R1) through conversational collaboration.
• Claimify - Extracting high-quality claims from language model outputs.
• Data Formulator - AI-powered Concept-driven Visualization Authoring.
• Data Formulator 2 - Iterative Creation of Data Visualizations, with AI Transforming Data Along the Way.
• Data Science Toolkit - Open-source collection of proven ML and AI implementation accelerators.
• E5 - Embedding model for Multilingual and Mutitask. Contrastive Pre-training (CCPairs dataset — millions of high-quality pairs of related texts (e.g., paraphrases or semantically similar sentences).
• Fara-7B - An Efficient Agentic Model for Computer Use.
• Flint - Visualization language for AI-generated chart specifications.
• GeoMind - A multi-agent system that lets non-experts query complex geospatial data in natural language.
• GraphRAG (by Microsoft) - RAG with a graph-based approach to efficiently answer both specific and broad questions over large text.
• Harrier - Microsoft open embedding model for retrieval and search.
• Knowledge-Augmented Large Language Models for Personalized Contextual Query Suggestion - K-LaMP. Contextual memory layer for LLM application.
• LIDA - LLM-powered data visualization and infographic generation from structured data.
• Magentic-One - A Generalist Multi-Agent System for Solving Complex Tasks.
• Magentic-UI - Experimental human-centered web agent that can browse the web, run code, and process files.
• MAI-Thinking-1, MAI-Code-1-Flash - Microsoft AI model family notes for reasoning, coding, and research models.
• MAI-Transcribe-1 - Microsoft speech-to-text model with strong multilingual WER, noisy-audio robustness, and Microsoft Foundry availability.
• MarS - A Financial Market Simulation Engine Powered by Generative Foundation Model - Financial market simulation using generative AI.
• MatterGen - Generative model for inorganic materials design.
• Memento - Extends effective output length by summarizing reasoning blocks and evicting old blocks from the KV cache.
• Microsoft AI Diagnostic Orchestrator (MAI-DxO) - Multi-Agent System. Accurately diagnoses up to 85% of NEJM case proceedings, which is over four times the rate achieved by a group of experienced physicians.
• Microsoft Office Copilot: Natural Language Commanding via Program Synthesis - Semantic Interpreter, a natural language-friendly AI system for productivity software such as Microsoft Office that leverages large language models (LLMs) to execute user intent across application features.
• Mnesis - A memory framework for LLMs combining similarity search with Global Selection, using a base graph for fast retrieval and a hierarchical graph for top-down semantic traversal.
• NL2KQL - From Natural Language to Kusto Query.
• Optimizing Model Selection for Compound AI Systems - A framework that automatically optimizes model selection for compound AI systems.
• PIKE-RAG - Specialized Knowledge & Rationale Augmentation Generation.
• PromptWizard - The prompt optimization through feedback-driven self-evolving prompts.
• R&D-Agent-Quant - R&D-Agent for Quantitative Finance, in short RD-Agent(Q), the first data-centric multi-agent framework designed to automate the full-stack research and development of quantitative strategies via coordinated factor-model co-optimization.
• SkillOpt - Trains reusable natural-language skills for frozen LLM agents.
• SpreadsheetLLM - Introduces an efficient method to encode Excel sheets, outperforming previous approaches with 25 times fewer tokens.
• Synthetic Computers at Scale for Long-Horizon Productivity Simulation - Synthetic user-specific computer environments for scaling long-horizon productivity simulations and agent training.
• The New Future of Work - Microsoft Research initiative studying meaningful, productive, equitable AI work.
• TRELLIS.2 - 4B image-to-3D generation model using O-Voxel structured latents for high-fidelity textured assets.
• VeriTrail - Detecting hallucination and tracing provenance in multi-step AI workflows.
• VibeVoice - Open-Source Text-to-Speech.

Sample Applications

• AI Dev Gallery - Windows application featuring interactive samples powered by local AI models.
• An Open-Source Template Gallery - Collection of Azure templates, including AI-focused starter projects.
• ARGUS - Hybrid approach with Azure Document Intelligence combined and GPT4-Vision to get better results without any pre-training.
• Azure AI CLI - Interactive command-line tool for AI.
• Azure Chat Solution Accelerator - Azure OpenAI chat accelerator with managed identities and enterprise controls.
• Azure Cosmos DB + OpenAI ChatGPT - Integration sample with Cosmos DB and chat interface.
• Azure OpenAI Embeddings QnA - Embeddings-powered Q&A sample.
• Azure Video Indexer Demo - Sample showing Azure Video Indexer integrated with OpenAI.
• Call Center AI - Initiate phone calls from an AI agent via API or direct dialing.
• C# ChatGPT + Enterprise Data - C# reference app using Azure AI Search and Azure OpenAI.
• ChatGPT + Enterprise Data RAG (Retrieval-Augmented Generation) - Enterprise RAG sample integrating Azure OpenAI and Azure AI Search.
• ChatGPT Plugin Quickstart using Python and FastAPI - Sample plugin implementation using Python and FastAPI.
• Chat with Your Data Accelerator - Accelerator for grounding chat on custom data sources.
• eShopSupport - A reference .NET application using AI for a customer support ticketing system.
• GPT-Azure-Search-Engine - Integration of Azure Bot Service with LangChain.
• Miyagi - Integration demo spanning multiple LangChain libraries.
• Prompt Buddy - Share and upvote favorite AI prompts.
• Simple ChatGPT UI Application - Scalable web UI pattern for Azure OpenAI chat.
• Smart Components - Experimental, end-to-end AI features for .NET apps.
• VoiceRAG - Voice-enabled RAG using Azure AI Search and GPT-4o Realtime.

Solution Accelerators

• accelerators.ms - Microsoft catalog of accelerators, reference apps, and AI solution starters.
• Agentic Applications for Unified Data Foundation - Agentic AI accelerator combining Microsoft Fabric, Azure AI Foundry agents, and Semantic Kernel.
• AI Hub Gateway Landing Zone - Enterprise-ready solution accelerator for implementing a centralized AI API gateway.
• AI-in-a-Box - Reference implementations for common Azure AI and machine learning scenarios.
• AI Solution Accelerators - List of Azure AI Solution Accelerators for demonstration and simplified deployment.
• Azure Accelerator Repos - Collection of official Azure accelerators.
• Azure-Cognitive-Search-Azure-OpenAI-Accelerator - Proof of Concept (POC) for a Generative AI Multi-Agent Architecture using Azure Services.
• Azure Landing Zones Accelerator - Collection of Azure Landing Zones Accelerator.
• Azure Multimodal AI + LLM Processing Accelerator - Build multimodal data processing pipelines with Azure AI Services + LLMs.
• Azure-samples: Accelerator Collection - Accelerator Collection sorted by stars count.
• Azure-Samples Accelerators - Community driven sample accelerators.
• Build Your Own Copilot Solution Accelerator - Solution Accelerator to show how to build your own copilot.
• Chat with Your Data - Solution Accelerator - Azure-based RAG accelerator using Azure AI Search and Azure OpenAI.
• Conversational Azure OpenAI Accelerator - Conversational AI solution accelerator using Azure OpenAI.
• Conversation Knowledge Mining Solution Accelerator - Generative AI solution for extracting insights from conversational data using Azure AI and Azure OpenAI.
• Document Generation Solution Accelerator - AI assistant for document generation.
• Generic Accelerator for Build Your Own Copilot - Build your own copilot solution accelerator.
• Healthcare Agent Orchestrator - A multi-agent accelerator that coordinates modular specialized agents to assist multi-disciplinary healthcare workflows.
• Hypervelocity Engineering - Applies HVE practices to continuously evolve Azure AI landing zones.
• MSUS Azure Accelerators - Partner accelerator implementations.
• Multi Agent Custom Automation Engine Solution Accelerator - AI-driven multi-agent automation reference solution built on Microsoft Agent Framework, Azure Foundry, and Azure Cosmos DB.
• Unified Data Foundation with Fabric Solution Accelerator - Unified Data Foundation accelerator using Microsoft Fabric with Databricks and Purview integration options.

Code Samples & Workshops

• Agent-Framework-Samples - Sample projects demonstrating Microsoft Agent Framework usage.
• AI Hub Gateway - Reference architecture and accelerator for a central enterprise AI gateway.
• AI-in-a-Box - Generative AI Bot Quickstart - A template deploys a Generative AI Virtual Assistant using Azure OpenAI and Bot Framework.
• AI-in-One-Dashboard - Power BI template for comprehensive AI usage analytics.
• AI Samples for .NET - Official .NET AI samples (evaluation, integration).
• APIM-Unified-AI-Gateway-Sample - Single APIM endpoint for Phi-4, GPT-4.1, and Gemini with centralized governance.
• Azure AI Foundry Baseline - Baseline setup for Azure AI Foundry projects.
• Azure Container Apps Dynamic Sessions Samples - On-demand, short-lived container sessions using Azure Container Apps Dynamic Sessions.
• Azure Functions OpenAI Extension - OpenAI bindings for Azure Functions (C#).
• Azure LLM Fine-tuning - SLM/LLM Fine-tuning on Azure.
• Azure Machine Learning Examples - Official community-driven Azure Machine Learning examples.
• Azure OpenAI Cookbook - The repository for all Azure OpenAI Samples complementing the OpenAI cookbook.
• Azure OpenAI Design Patterns - A set of design patterns using the Azure OpenAI service.
• Azure OpenAI Network Latency Test Script - Python script for measuring Azure OpenAI network latency.
• Azure OpenAI RAFT - RAFT Fine-Tuning for RAG Using Azure OpenAI.
• Azure OpenAI RAG Workshop - RAG using LangChain.js and OpenAI. Hosted on Azure Static Web Apps and Azure Container Apps, with Azure AI Search as the vector database.
• Azure OpenAI Using PTUs/TPMs With API Management - Scaling Azure OpenAI using APIM.
• Azure OpenAI with AKS by Terraform (simple version) - Terraform sample for deploying Azure OpenAI with AKS.
• Azure OpenAI Workshop - Workshop materials to build intelligent solutions on Open AI.
• Azure-Samples - GitHub organization for official Azure sample projects.
• Azure SQL DB + AOAI - AI applications built on data from an Azure SQL Database.
• Build Your Own Copilot - RAG demo using Cosmos DB, AKS, Semantic Kernel.
• Create an Azure OpenAI, LangChain, ChromaDB, and Chainlit ChatGPT-like application in Azure Container Apps using Terraform - ChatGPT-like application in Azure Container Apps using Terraform.
• Design and Evaluation of RAG Solutions - Resources for designing and evaluating RAG solutions.
• Document Intelligence-in-a-Box - Uses Azure AI Document Intelligence to extract data from PDF forms and store results in Azure Cosmos DB — part of the "AI-in-a-Box" accelerator.
• Document Processing with Azure AI Samples - Document Intelligence plus generative enrichment examples.
• Domain-Specific Languages Copilot - Domain-Specific Languages specialized Copilot demo & Fine Tuning Pipeline.
• Enterprise Logging - Comprehensive logging of Azure OpenAI model execution.
• Enterprise RAG Avatar - GPT-RAG Avatar integrates Generative AI responses, real-time streaming into a dynamic avatar interface.
• Evaluating a RAG Chat App - Tools for evaluation of RAG Chat Apps using Azure AI Evaluate SDK.
• Fabric Notebooks for Analyzing Chat History Stored in Cosmos DB - Fabric notebooks for analyzing chat history.
• Generate Synthetic QnAs from Real-World Data - Demonstrating how to create/augment a QnA dataset from complex unstructured data.
• GPT-RAG - Enterprise RAG solution accelerator.
• GPT RAG Ingestion - GPT-based RAG ingestion utilities.
• Legal Research Copilot Example - GraphRAG on Azure Database for PostgreSQL. Use U.S. Case Law dataset of 0.5 million legal cases.
• Microsoft Copilot Studio Samples - Samples and artifacts for Microsoft Copilot Studio.
• Microsoft.Extensions.AI - Unified C# abstractions for models, embeddings, and middleware.
• Microsoft Frontier Company - Operating model for trusted, outcome-driven enterprise AI transformation.
• OpenAI Chat Application with Microsoft Entra Authentication - Authenticated chat starter with Microsoft Entra ID.
• RAG for Azure Data - Retrieval Augmented Generation (RAG) for Azure Data.
• Setting up Azure OpenAI with Azure API Management - Azure AI to your application developers in a secure & manageable way with Azure API Management.
• Smart Load Balancing for Azure OpenAI - Intelligent routing across multiple Azure OpenAI deployments.
• Vector Similarity Search with Azure SQL & Azure OpenAI - Samples on how to use Azure SQL database with Azure OpenAI.

Architecture Patterns & Use Cases

• AI Agent-Driven Auto Insurance Claims RAG Pipeline - Agent-driven claims processing reference.
• AI Decision Tree - Selecting the most suitable tools and platforms for your generative and nongenerative AI use cases.
• AI Feed - Azure AI news, events, and discussions.
• AI Gateway Capabilities in Azure API Management - Secure, scale, observe, and govern models, agents, MCP servers, and A2A APIs with authentication, token controls, semantic caching, load balancing, and policy enforcement.
• AI Gateway Enhancements: LLM policies, Real-Time API support, Content Safety, and more - Semantic Caching. Azure AI Content Safety. LLM policies (llm-token-limit, llm-emit-metric, llm-content-safety).
• AI-Landing-Zones - Reference architecture for secure, scalable Azure AI apps and agents.
• AI Landing Zones Guardrails for Azure Machine Learning - Recommended Azure Policy initiative for Azure Machine Learning guardrails.
• AI Landing Zones Guardrails for Bot Service - Recommended Azure Policy initiative for Bot Service guardrails.
• AI Landing Zones Guardrails for Cognitive Services - Recommended Azure Policy initiative for Cognitive Services guardrails.
• AI Landing Zones Guardrails for OpenAI - Recommended Azure Policy initiative for Azure OpenAI guardrails.
• An Introduction to LLMOps - Operationalizing and Managing Large Language Models using Azure ML.
• Authentication and Authorization in Generative AI applications with Entra ID and Azure AI Search - Authentication in AI applications with Entra ID and Azure AI Search.
• Azure AI Foundry Blog - News and updates about Azure AI Foundry and related platform capabilities.
• Azure AI Services Landing Zone - Azure AI Landing Zone with Terraform.
• Azure Command Companion - Azure CLI Command Generation to translate natural language into the appropriate Azure CLI commands.
• Azure OpenAI and Call Center Modernization - Azure OpenAI and Call use case to improve its customer service and reduce the cost and time of handling customer calls.
• Azure OpenAI Best Practices Insights from Customer Journeys - Best Practices for Azure OpenAI Resources.
• Azure OpenAI Chat Baseline Architecture - Designing your production enterprise chat applications.
• Azure OpenAI Landing Zone Reference Architecture - A reference architecture that integrates a variety of services to create a seamless infrastructure for running OpenAI workloads.
• Baseline Agentic AI Systems Architecture - Agentic AI Systems Architecture to build multiple conversable agents system.
• Baseline OpenAI End-to-End Chat Reference Architecture - Azure AI Foundry chat reference architecture.
• Build Language Model Pipelines with Memory - Baseline Azure AI Foundry chat reference architecture.
• CAIRA - Composable AI Reference Architectures (CAIRA).
• Chat with Your Azure DevOps Data - The solution leverages Azure DevOps CLI to periodically export ADO Work Items as JSON and store them in Azure Blob storage.
• Check Your Facts and Try Again: Improving Large Language Models with External Knowledge and Automated Feedback - Improving Large Language Models with External Knowledge and Automated Feedback.
• Copilot Custom Agents for .NET Developers: C# Expert & WinForms Expert - First two experimental custom agents for .NET developers: C# Expert and WinForms Expert.
• Designing and Developing a RAG Solution - Designing a RAG application flow, RAG data pipeline flow.
• Evaluations, Monitoring, Tracing in Microsoft Foundry - Built-in and custom evaluators across development, CI, and production.
• Extend Your Coding Agent with .NET Skills - How to extend coding agents using .NET skills.
• Grounding LLMs - Use-cases for Grounding & Retrieval-Augmented Generation (RAG).
• How Real-World Businesses Are Transforming with AI - Collected over 200 examples of how organizations are leveraging Microsoft AI capabilities.
• How to Develop AI Apps and Agents in Azure – A Visual Guide - A Visual map to help you decide which Azure AI service is best for your use case.
• Industry Cloud Reference Architectures - Holistic architecture design and reference implementation for industry cloud.
• Integrate Private Access to Your Azure Open AI Chatbot - Private access solutions to your Azure Open AI Chatbot.
• Microsoft-AI-Decision-Framework - Decision framework for selecting Microsoft AI technologies across Copilot and Azure AI.
• Microsoft AI / Responsible AI - Responsible AI resources from Microsoft.
• NL to SQL Architecture Alternative - LLM to dynamically generate SQL queries, execute those SQL queries.
• Optimize Azure OpenAI Applications with Semantic Caching - Semantic cache pattern for latency and cost reduction.
• Responsible AI Transparency Report - Overview of Microsoft's responsible AI tools, policies, and practices.
• Retrieval Augmented Fine Tuning - RAFT: Combining the best parts of RAG and fine-tuning (SFT).
• Revolutionize your Enterprise Data with ChatGPT - One of well-known RAG application sample utilizing Azure.
• Safeguard and Trustworthy Generative AI Applications - Announcing new tools in Azure AI to help you build more secure and trustworthy generative AI applications.
• Securing Azure AI Applications - The emerging threats and how to defend your Azure AI stack.
• Security Best Practices for GenAI Applications (OpenAI) in Azure - The best practices of security for GenAI applications in Azure.
• Smart Document Processing with ADI, AI Search and OpenAI - Reusable extraction pipeline for large semi-structured documents.
• Sovereign controls in data processing for Microsoft 365 Copilot - Microsoft 365 Data Residency and Processing commitments map.
• Using Keyless Authentication with Azure OpenAI - Entra ID token-based access to Azure OpenAI.
• Vector Search Technology Choices (Azure Architecture Guide) - Decision guide for vector search technology choices.