Skip to content

Commit a0d0198

Browse files
brown9804brown9804
authored
Create overview for Foundry IQ and Control Plane
Added an overview of MSFT Foundry IQ and Control Plane, detailing their functionalities, purposes, and capabilities.
1 parent 27e5574 commit a0d0198

File tree

1 file changed

+90
-0
lines changed

1 file changed

+90
-0
lines changed

‎0_Azure/3_AzureAI/AIFoundry/demos/11_FoundryIQ_ControlPlane.md‎

Lines changed: 90 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,90 @@
1+
# MSFT Foundry IQ \& Control Plane - Overview
2+
3+
Costa Rica
4+
5+
[![GitHub](https://img.shields.io/badge/--181717?logo=github&logoColor=ffffff)](https://github.com/)
6+
[brown9804](https://github.com/brown9804)
7+
8+
Last updated: 2025-07-17
9+
10+
------------------------------------------
11+
12+
<details>
13+
<summary><b>List of References</b> (Click to expand)</summary>
14+
15+
- [What is the Microsoft Foundry Control Plane?](https://learn.microsoft.com/en-us/azure/ai-foundry/control-plane/overview?view=foundry)
16+
- [MSFT Docs - What is the Microsoft Foundry Control Plane?](https://github.com/MicrosoftDocs/azure-ai-docs/blob/main/articles/ai-foundry/default/control-plane/overview.md) - GitHub repo
17+
18+
</details>
19+
20+
<details>
21+
<summary><b>Table of Content</b> (Click to expand)</summary>
22+
23+
- [Control Plane](#control-plane)
24+
- [Foundry IQ](#foundry-iq)
25+
26+
</details>
27+
28+
## Control Plane
29+
30+
> The Control Plane is the central management layer that governs how your AI agents operate across an enterprise. It does not execute model inference (that’s the data plane); instead, it provides visibility, governance, and orchestration for everything running in the data plane.
31+
> `Think of it as mission control for your AI ecosystem, where you set policies, monitor performance, and enforce compliance across thousands of agents.` Click here to read more about [Azure control plane and data plane](https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/control-plane-and-data-plane), and [Microsoft Foundry architecture
32+
](https://learn.microsoft.com/en-us/azure/ai-foundry/concepts/architecture?view=foundry-classic)
33+
34+
<img width="1007" height="381" alt="image" src="https://github.com/user-attachments/assets/bd7f3cc6-78c1-414c-8952-498c2e12c29e" />
35+
36+
Based on [What is the Microsoft Foundry Control Plane?](https://learn.microsoft.com/en-us/azure/ai-foundry/control-plane/overview?view=foundry)
37+
38+
| **Function** | **Description** | **Example Use Case** |
39+
| ------------------------- | -------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- |
40+
| **Fleet-wide Operations** | Centralized management of all AI agents and services across environments. | Update pricing logic for 500 retail agents in one click; scale agents during peak demand. |
41+
| **Observability** | Real-time monitoring of performance, cost, latency, and anomalies. | Detect an agent making ungrounded recommendations and fix before customer impact. |
42+
| **Governance & Control** | Apply policies, guardrails, and compliance rules across the fleet. | Restrict finance agents to approved datasets; block marketing agents from accessing PII. |
43+
| **Security** | Identity, access control, and threat protection integrated with Microsoft tools. | Revoke compromised agent credentials instantly using Entra Agent ID and Defender. |
44+
45+
> [!NOTE]
46+
> Control Plane ~ Management Layer: `E.g: Air Traffic Control, sets flight paths, monitors planes, enforces rules.`
47+
> - Purpose: Governs, orchestrates, and monitors operations across your AI ecosystem.
48+
> - Activities: Policy enforcement, lifecycle management, observability, security, compliance. <br/>
49+
>
50+
> Data Plane ~ Execution Layer: `E.g: The Airplanes, they fly and carry passengers (data/tasks).`
51+
> - Purpose: Runs the actual workloads (model inference, data processing, agent actions).
52+
> - Activities: Executes prompts, retrieves data, performs reasoning tasks.
53+
54+
<img width="1904" height="830" alt="image" src="https://github.com/user-attachments/assets/29dcf372-c5ef-41d1-8d83-1ff44a4c723f" />
55+
56+
57+
## Foundry IQ
58+
59+
`Foundry IQ ~ embedded RAG (Retrieval-Augmented Generation) layer for enterprise AI agents.`
60+
61+
> **Foundry IQ** is Microsoft’s **enterprise knowledge layer** designed to power Retrieval-Augmented Generation (RAG) for AI agents. It connects agents to organizational data securely and efficiently, enabling them to retrieve, reason, and respond with grounded context.
62+
> `Think of it as the “knowledge engine” for your AI ecosystem, where agents can access structured and unstructured data across multiple sources without manual integration.` Click here to read more about [Foundry IQ: Unlocking ubiquitous knowledge for agents](https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/foundry-iq-unlocking-ubiquitous-knowledge-for-agents/4470812)
63+
64+
<img width="1536" height="1024" alt="image" src="https://github.com/user-attachments/assets/12d582fe-bd33-4444-8c99-f08194210424" />
65+
66+
> [!NOTE]
67+
> **Foundry IQ ~ Knowledge Layer**: `E.g: A smart library system, agents ask questions, IQ finds the right books and pages instantly.`
68+
> - **Purpose**: Provides unified, secure access to enterprise data for AI agents.
69+
> - **Activities**: Indexing, vectorization, query planning, synthesis, multimodal retrieval. <br/>
70+
>
71+
> **Data Plane Role**:
72+
> - Foundry IQ operates in the **data plane**, serving as the retrieval and grounding engine for agents.
73+
> - It works under governance from the **Control Plane**, which enforces compliance and security.
74+
75+
> Core Capabilities of Foundry IQ:
76+
77+
| **Capability** | **Description** | **Example Use Case** |
78+
| -------------------------------------- | ---------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
79+
| **Unified Data Access** | Connects to OneLake, SharePoint, Fabric IQ, Snowflake, S3, and more. | An agent retrieves product specs from SharePoint and pricing from Snowflake in one query. |
80+
| **Automated Indexing & Vectorization** | Converts enterprise data into searchable embeddings automatically. | HR documents and product catalogs indexed for semantic search without manual setup. |
81+
| **Advanced Query Planning** | Breaks down complex queries into multi-step retrieval tasks. | A compliance agent queries “latest GDPR policy impact on EU marketing” across multiple sources. |
82+
| **Secure Retrieval** | Enforces Purview sensitivity labels and access policies at query time. | Finance agent blocked from accessing confidential HR data during retrieval. |
83+
| **Multimodal RAG** | Supports text, tables, and structured data for richer grounding. | AI assistant combines text from policy docs and tables from Excel for a complete answer. |
84+
85+
<!-- START BADGE -->
86+
<div align="center">
87+
<img src="https://img.shields.io/badge/Total%20views-1633-limegreen" alt="Total views">
88+
<p>Refresh Date: 2025-12-03</p>
89+
</div>
90+
<!-- END BADGE -->

0 commit comments

Comments
 (0)