Timo Weber
ServiceNow Employee

 

AI Agent Masterclass • Session 5

Converting Data into Intelligence

Workflow Data Fabric, Zero Copy & Interoperability

← Back to Masterclass Overview

Date
27 January 2026
Duration
~80 minutes
Speakers
Timo Weber, Chris Thorne, Julia Mulcahy, Javier Lombarda, Thomas Geering, Joost van Poppel
🎬 Coming soon
🎯 Key Takeaways
  • Workflow Data Fabric (WDF) is ServiceNow's purpose-built data fabric with three pillars: Connect, Contextualize, and Control — designed for action, not just data storage
  • Zero Copy Architecture enables AI Agents to query external data (Snowflake, Databricks, BigQuery) in real-time without duplicating it into ServiceNow
  • MCP (Model Context Protocol) and A2A (Agent-to-Agent) protocols enable interoperability — ServiceNow supports both as client and server
  • Process Mining can monitor agentic workflows to continuously improve AI Agent performance at scale

The Missing Piece: Data for AI Agents

Throughout Sessions 1-4 of the AI Agent Masterclass, we covered the "why," "what," and "how" of AI Agents. We discussed the AI Agency Gap, use case prioritization, building agents, and evaluation frameworks. But we missed a tiny, tiny bit — we didn't talk about data with intention.

Session 5 addresses this gap head-on. Because AI Agents are only as good as the data they can access. If your data is fragmented across systems, if it's stale, if it lacks context — your agents will make wrong decisions. This session shows how Workflow Data Fabric transforms data chaos into actionable intelligence.

Workflow Data Fabric: A Data Fabric for the Agentic Era

Unlike traditional data fabrics that simply move data from A to B, ServiceNow's Workflow Data Fabric is purpose-built for action. It doesn't just connect data — it contextualizes it and governs it directly in the flow of work.

"This is what transforms AI Agents from just being a chatbot into a context-aware co-worker — one that truly understands the business and delivers enterprise-wide outcomes in real-time."

— Chris Thorne, ServiceNow

📸 Image: Workflow Data Fabric Overview — A data fabric for the agentic era with Knowledge Graph at the center, connecting workflows, AI Agents, and AI Experiences

The 3 Pillars of Workflow Data Fabric

Pillar 1
Connect Your Data
  • Integration Hub: 200+ pre-built integrations, AI-powered spoke generator
  • External Content Connectors: SharePoint, Confluence, PDFs — indexed without copying
  • Stream Connect: High-volume Kafka streaming (20M+ events/day)
  • Zero Copy Connectors: Query Snowflake, Databricks, BigQuery in real-time
  • RPA Hub: UI automation for legacy systems without APIs
  • MCP Client: Connect to external MCP servers for extended AI capabilities
Pillar 2
Contextualize Your Data
  • Data Catalog: Single source of truth with metadata, lineage, and context (Data.World acquisition)
  • Knowledge Graph: Connect data sets for end-to-end ecosystem views
  • AI Data Explorer: Natural language questions → deep contextual insights (NEW)
  • Data Marketplace: Self-service data discovery for non-technical users
  • Coming Soon: AI Data Explorer will recommend and trigger agentic workflows based on insights
Pillar 3
Control Your Data
  • Governance: Automated lineage, policy-based access, metadata enrichment
  • Observability: Monitor data quality and usage patterns
  • Security: Semantic layer ensures compliance across AI strategies
  • API Management: Design, publish, and monitor APIs with visibility
  • Automation Center: Single pane of glass for all automations and ROI tracking

📸 Image: The 3 Pillars of Workflow Data Fabric — Connect, Contextualize, and Control with detailed capability breakdown

Zero Copy Architecture: Query Without Duplicating

One of the most powerful capabilities introduced is Zero Copy Connectors. Instead of replicating massive data sets into ServiceNow, AI Agents can query external data lakes (Snowflake, Databricks, Google BigQuery) in real-time using federated SQL queries.

0
Data Duplication
Live
Real-Time Data Access
In-Situ
Data Stays at Source

Zero Copy Use Case Patterns

Pattern Description Example
Insight-Driven Action Insights generated in data cloud trigger actions in ServiceNow Predictive maintenance alerts → automatic work orders
Real-Time Enrichment Pull contextual data when decisions are needed Customer 360° view during support interactions
Data Augmentation Enrich ServiceNow data with external master data Customer data, product catalogs, pricing without replication

📸 Image: Zero Copy Architecture — CSM Agent querying customer purchase history via Federated Query Engine and Zero-Copy Connectors

Integration Patterns for Optimal Performance

Not every integration pattern works for every scenario. The key is matching your requirements to the right approach:

Pattern Latency Best For Use Case
Synchronous API <500ms User-facing validations, instant lookups Form validation
Event Streaming Milliseconds Real-time alerts, operational intelligence IoT monitoring
Zero Data Copy ✓ Query-time Data Lakes, large-scale analytics AI enrichment, insights to action
Federated Query Variable Multi-source insights, cross-system joins 360° customer view
Asynchronous Sec to Min Bulk operations, background processing Data migration

📸 Image: Integration Patterns for Optimal Performance — Recommended WDF + ZDC use cases highlighted in green

WDF+ZDC Decision Framework: Qualification Factors

Use this framework to determine when Zero Copy makes sense for your scenario:

1. DATA VOLUME
Standard: <10K records
WDF+ZDC: >100K records ✓
2. DATA FRESHNESS
Standard: Historical
WDF+ZDC: Real-time ✓
3. BUSINESS RULES
Standard: Simple
WDF+ZDC: Complex ✓
4. LATENCY REQ.
Standard: Batch
WDF+ZDC: Sub-second ✓
5. DATA SENSITIVITY
Standard: Copyable
WDF+ZDC: Must stay in source ✓
6. CROSS-SYSTEM
Standard: Single source
WDF+ZDC: Multi-source ✓

💡 Key Takeaway: WDF is essential when data freshness and sensitivity are critical constraints. Score 4+ factors for WDF to maximize ROI.

📸 Image: WDF+ZDC Decision Framework — 6 qualification factors for choosing the right integration approach

Agent Interoperability: MCP and A2A Protocols

With the rise of AI Agents across platforms, interoperability becomes crucial. ServiceNow supports two key protocols for connecting with external AI ecosystems:

🔧 MCP (Model Context Protocol)

Purpose: Standardized way for an LLM to access external systems and their functions

Use When: You need context or data from external tools, but the logic stays in your ServiceNow agent

ServiceNow Support:

  • MCP Client (use external MCP servers)
  • MCP Server (expose ServiceNow tools to external clients)
  • Currently supports: Tools (Resources & Prompts coming)
🤖 A2A (Agent-to-Agent)

Purpose: Communication between AI Agents to collaborate on tasks toward a common goal

Use When: You want to leverage a fully-fledged external agent that can reason and act independently

ServiceNow Support:

  • Call external A2A agents (Workday, Salesforce, Google)
  • Agent discovery via agent cards
  • Orchestrator automatically picks the right agent

"In my head, how I distinguish these: If I need context or data — use MCP. If I want to leverage an external agent that can reason and do something more — use A2A."

— Thomas Geering, ServiceNow

📸 Image: What is MCP and A2A — MCP with ServiceNow as client vs. A2A with Google Vertex AI Agent

A2A in Action: Employee Transfer Example

Here's a practical example of how A2A enables cross-platform automation:

  1. ServiceNow AI Agent handles an employee transfer request ("John Smith is transferring from finance to logistics")
  2. Agent needs to update job history in Workday — searches for the right external agent via A2A protocol
  3. Workday Talent Mobility AI Agent is discovered and receives the handoff
  4. Workday agent performs the update and communicates back: "Job history updated. Please proceed."
  5. ServiceNow agent resumes workflow, updates software license ownership, completes the transfer

📸 Image: ServiceNow AI Agent with A2A Example — Employee transfer workflow showing handoff between ServiceNow and Workday agents

How to Set Up MCP in ServiceNow

Thomas demonstrated the MCP setup live in AI Agent Studio:

  1. Go to AI Agent Studio → Settings → Manage MCP Servers
  2. Add your MCP server with authentication (OAuth, API Key, or none)
  3. In your AI Agent, go to Tools → Add Tool → MCP Server Tool
  4. Select your MCP server — it fetches available tools automatically
  5. Important: Define which specific tools to use (don't preload all to save tokens)

Continuous Improvement with Process Mining

As organizations scale AI Agents, the complexity of managing, monitoring, and optimizing these systems increases significantly. Process Mining provides visibility into how AI Agents are actually interacting to solve use cases.

"There will be a shift from more deterministic workflows where an AI agent assists a human at a certain step, to more agentic systems where the entire use case could be accomplished by a team of AI agents. Process Mining becomes a powerful solution to help us get the visibility we need to continuously improve these agentic systems."

📸 Image: Process Mining for AI Agents — Visualizing how AI agents interact to solve use cases with process flow diagram

Live Demos

Session 5 featured two comprehensive demos showing WDF and Zero Copy in action:

Demo 1 — Julia Purvis
Electri: Cosmetics Supply Chain Optimization

Scenario: When music artists come to town, demand for fragrances skyrockets — but the company can't keep up (33% back-order rate).

Solution: AI Agent + Zero Copy + Knowledge Graph — automatically adjusts inventory orders and staffing based on concert data. Result: Back-order rate dropped from 33% to 2%.

Demo 2 — Javier Lombarda
Terraforce: Predictive Maintenance with Databricks

Scenario: Manufacturing company with massive sensor data across multiple systems — data is fragmented and hard to act on in real-time.

Solution: Zero Copy to Databricks feeds agentic workflow that analyzes sensor anomalies, interprets alerts, assesses risk, and automatically creates cases for critical warnings.

Ready to Evaluate? Start a POV

If you want to evaluate ServiceNow's agentic framework for your organization, ServiceNow offers Proof of Value (POV) engagements:

🎯 What You Get in a POV
  • Dedicated evaluation instance under your control
  • Load your own data to test with real scenarios
  • POV Hub with Application Tracker for plugin management
  • Use case discovery workshop (as covered in Session 2)
  • Build & evaluate agentic use cases running on your data

📸 Image: POV Application Tracker — Track installed applications, versions, and POV configuration

Interested? Contact your ServiceNow account team to express interest in a POV engagement.

What's Next? You Decide!

Good news — Session 5 is not the last session! Based on audience feedback, the Masterclass will continue. The top-voted topics for upcoming sessions are:

🥇 AI Control Tower
🥈 A2A Deep Dive
🥉 Voice Agents
MCP Deep Dive
Knowledge Graph
Agentic Desktop
"The rate of change has never been faster."
And it will never be this slow again!
Start NOW — See you soon!

📸 Image: Closing Slide — "The rate of change has never been faster. And it will never be this slow again! Start NOW."

Your Next Steps
  • Watch the Recording: See all demos in action and catch details you may have missed
  • Assess Your Data Landscape: Use the WDF+ZDC Decision Framework to identify integration opportunities
  • Explore MCP/A2A: If you have external agents or tools, start planning interoperability
  • Request a POV: Contact your account team to evaluate the agentic framework with your data
  • Stay Tuned: Session 6+ are coming — bookmark the Masterclass Overview Article

Last updated: January 2026 | By EMEA AI Solution Architect Team

Version history
Last update:
2 hours ago
Updated by:
Contributors