MQTT
AI PlatformAI-native
Microsoft Azure AI Foundry
Microsoft's unified PaaS platform for building, deploying, and managing AI applications – agents, models, and tools – with enterprise-grade security, governance, and observability built in.
Data residencySovereign cloud
Regions
30totalRegions
SDK / Languages
3python, javascript…
Uptime SLA
99.9%
Robotics-Ready
✓
Description
Microsoft Azure AI Foundry (formerly Azure AI Studio) is a unified PaaS platform designed for AI application developers, ML engineers, and IT administrators. It consolidates Azure OpenAI Service, Azure AI Services, agents, evaluation pipelines, and MLOps tooling under a single Azure resource managed through the Microsoft.CognitiveServices provider.
Architecture and Resource Hierarchy
The platform is organized as a two-level hierarchy: the Foundry resource (governance layer – networking, security, model deployments) and projects (isolated development boundaries within the resource). Connected resources – Azure Storage, Azure Key Vault, and Azure AI Search – maintain their own governance boundaries. Azure RBAC is scoped at both the resource and project level.
Key Capabilities
The platform provides: multi-agent orchestration and workflow execution; a tool catalog of over 1,400 tools; agent memory for persistent context across interactions; Foundry IQ for grounding agent responses in enterprise or web content; publishing to Microsoft 365, Teams, and BizChat; real-time observability with built-in metrics and model tracking; and centralized AI asset management via the Operate section.
Deployment Types
Supported model deployment types include: Global Standard (cross-region, pay-per-token), Global Provisioned (reserved throughput, hourly billing), Global Batch (asynchronous, 50% discount), Data Zone Standard/Provisioned/Batch (processing within US or EU boundary), Standard and Regional Provisioned (single region). Compute infrastructure enforces per-tenant logical isolation with inline content safety guardrails.
Security and Compliance
The platform supports data encryption at rest and in transit (FIPS 140-2, 256-bit AES) with Microsoft-managed keys or customer-managed keys (CMEK via Azure Key Vault). Network integration includes Private Link, VNet injection, and managed virtual network options. Microsoft Entra ID, SAML 2.0, OIDC, and OAuth 2.0 are supported for SSO. Azure Policy enforces organizational governance policies.
MLOps LifecycleMLOps LifecycleFull model lifecycle: registry, feature store, prompt management, monitoring and human-in-the-loop.
Model Registry
Versioning — model artifact versioning
Approval workflows — approval workflow before production
Immutable artifacts — immutability of stored versions
Lineage tracking — tracking data and model relationships
3 / 4 supported · 1 unsupported hidden
Feature Store
Online serving — real-time feature serving
Offline storage — feature storage for training
Streaming ingestion — streaming ingestion (Kafka, Flink)
0 / 3 supported · 3 unsupported hidden
Prompt Management
Prompt registry — central prompt repository
Versioning — prompt versioning and history
Testing frameworks — A/B testing and prompt evaluation
3 / 3 supported · none unsupported
Monitoring
Data drift detection — input data drift detection
Concept drift detection — concept drift detection
Hallucination monitoring — LLM hallucination monitoring
Bias evaluation tools — bias evaluation tooling
3 / 4 supported · 1 unsupported hidden
Human-in-the-Loop
Labeling services — data labeling tools
RLHF workflows — reinforcement learning from human feedback
Manual override — manual override of model decisions
1 / 3 supported · 2 unsupported hidden
Data & KnowledgeData & Knowledge ManagementData connectors, vector database integration, native vector search and data management (PII, provenance, synthetic data).
ApplicationsAI ApplicationsDomains and use cases this platform is best suited for — from RAG and fine-tuning to scientific research.
SecurityEnterprise SecurityCertifications, access controls and data-protection features essential for corporate deployments and cloud privacy compliance.
Developer EcosystemDeveloper EcosystemDeveloper resources: available SDKs, supported programming languages, and infrastructure features and model-deployment methods.
SDK Languages
PyPythonJSJavaScriptTSTypeScript
API Type
REST
Community & resources
Templates library
Quickstarts
API Reference
Tutorials
Pricing & Business ModelPricing & Business ModelBilling models (usage-based, provisioned throughput), resource limits and SLA parameters (uptime, support tiers).
Pricing models
Usage-based
Provisioned throughput
Token-based
Resource quotas
Per project
Per user
Cost alerting
SLA & Support
99.9%uptime SLA
StandardEnterprise 24/7
Robotics & Humanoids ExtensionRobotics & Humanoids ExtensionSimulation engines (Isaac Sim, Gazebo, MuJoCo), communication protocols (ROS2, MQTT, Zenoh), robotics standards (URDF, OpenUSD) and edge orchestration.
No robotics support
Robotics standards
- URDF Support
- OpenUSD Interoperability
- Sim-to-Real Pipelines
Edge Orchestration
- OTA updates (over-the-air)
- Real-time kernel support
SourcesDocumentation VaultCentralized hub of links to official sources, technical guides, repositories and release notes.
SustainabilitySustainabilityCarbon footprint, renewable-energy share powering data centers, and energy-efficiency metrics (e.g. PUE).
Carbon footprint tracking
PUE global FY25 = 1.17; WUE global FY25 = 0.27 L/kWh; target 100% renewable energy matching by 2030