Real-Time Intelligence: Building event-driven AI apps and agents | OD819
Tessa Kloster, Arindam Chatterjee, and Anshul Sharma present a Microsoft Build 2026 session on using Microsoft Fabric Real-Time Intelligence to build event-driven AI applications and autonomous agents that react to live data, combining streaming ingestion, real-time analytics, and actioning in a governed workflow.
Overview
The session focuses on Microsoft Fabric Real-Time Intelligence as a unified approach for:
- Streaming ingestion
- Real-time processing and analytics
- Taking action on insights (including agent-driven actioning)
It positions Real-Time Intelligence as a way to move from live signals to insights to action without building and maintaining complex, multi-service pipelines.
Full summary based on transcript
Introduction: Real-Time Intelligence and AI for organizations
The speakers introduce Real-Time Intelligence in the context of building AI-enabled systems that can respond to live operational data within seconds.
Microsoft Fabric as a unified data platform for AI
The session frames Microsoft Fabric as a unified data platform that brings together governed data experiences needed for AI workloads, with Real-Time Intelligence as the capability set focused on streaming and low-latency analytics.
Defining Real-Time Intelligence and core capabilities
The presenters describe Real-Time Intelligence as a set of capabilities that unify:
- Event ingestion
- Stream processing
- Real-time analytics
- Actioning (operational responses driven by insights)
Event ingestion, processing, and demo setup
Arindam introduces the real-time demo flow, centered on ingesting events, processing them, and producing insights quickly enough to drive operational decisions.
Demo: stadium operations with Eventstream and fraud detection
The demo scenario uses a stadium operations context to show how live events can be ingested and analyzed, including a fraud detection angle.
Key elements called out in the session:
- Using Eventstream for event ingestion
- Applying real-time analytics to detect suspicious patterns
- Driving operational responses based on detected signals
DeltaFlow, CDC, and Spark integration
The session discusses integrating data processing components and patterns used in real-time and near-real-time systems, including:
- DeltaFlow
- Change Data Capture (CDC)
- Apache Spark integration for processing and generating AI-relevant insights
Analyzing and acting on real-time data in Eventhouse
The presenters cover analyzing streaming/real-time data in Eventhouse and using the results to drive action.
MCP and agent-based Real-Time Intelligence with GitHub CLI and Copilot
The session connects Real-Time Intelligence to agent-based workflows, referencing:
- MCP
- GitHub CLI
- GitHub Copilot
The focus is on using these tools/protocols to help build or operate autonomous/agent-driven experiences that can respond to real-time signals.
AI integration: anomaly detection and Operations Agent GA
Tessa covers AI integration points, including:
- Anomaly Detector
- Operations Agent (GA)
Wrap-up and next steps
The session closes with guidance on where to learn more and how to get started with Microsoft Fabric Real-Time Intelligence.