Microsoft + CData Workshop: Build a Connected, Data-Driven Agent in Copilot Studio

Free hands-on session with Microsoft's Sabin Nair on January 28th

Your users want Copilot to do more than summarize and write emails. They want to answer real business questions: What's our pipeline by region? Which accounts have open support escalations and renewals this quarter? How does campaign spend correlate with closed-won revenue?

Those questions require live data from Salesforce, NetSuite, SAP, Snowflake — and the ability to correlate across systems in a single response. While you're figuring out the right architecture to deliver these capabilities, your users are already finding workarounds: exporting to CSV, uploading spreadsheets to agents, copying sensitive data into Copilot chat windows.

Join us for a free hands-on workshop on January 28th where you'll build a Copilot Studio agent capable of thinking across enterprise systems with Microsoft's recommended architecture pattern and guidance directly from their product team.

Best practices for agent building from Microsoft

Sabin Nair, Principal Group Product Manager at Microsoft, is co-hosting this workshop alongside CData's Jerod Johnson. Sabin leads product development for Copilot Studio's enterprise integration capabilities and will guide workshop attendees through:

• How to build a working AI agent that uses live data

• Best practices for agent design and deployment at enterprise scale

• The architecture patterns Microsoft recommends for connecting agents to business data securely

Sabin will share why CData Connect AI is the managed MCP platform featured by Microsoft for AI data integration across the Copilot ecosystem — including Copilot Studio, Azure AI Foundry, and the new Agents for Microsoft 365 platform.

Microsoft selected CData because Connect AI delivers what enterprises need to scale AI data access safely:

• Multi-source queries without ETL. Your agent can join data across Salesforce, NetSuite, Snowflake, Jira, and 350+ other sources in a single response — no data movement, no stale snapshots, no custom integration code.

• Passthrough security. Every AI query automatically enforces the user's existing source system RBAC at runtime. If a sales rep can only see their territory in Salesforce, that's exactly what Copilot can access. No parallel permission mapping required.

• Complete audit trails. Every AI-to-data interaction is logged with full user attribution — who asked what, which systems were queried, what data was returned. Exportable logs integrate with your SIEM for compliance reporting.

• No credential sprawl. OAuth/SAML passthrough means users authenticate with their existing enterprise identity. No shared service accounts or API keys floating around in automation tools.

Build agents in Copilot Studio without shadow IT risk

In 45 minutes, you'll build and deploy a working Copilot Studio agent that queries live enterprise data with your existing security controls intact.

Connected agent in Copilot Studio

• Connect multiple enterprise sources (Salesforce, Snowflake, NetSuite, or your own systems)

• Configure cross-system queries so Copilot correlates data across sources in a single response

• Test the agent with real queries against production data

Architecture and security foundation

• Passthrough security model with inherited RBAC

• Audit configuration for complete visibility into AI-data interactions

• Workspace isolation and least-privilege controls

Best practices from Microsoft

• Agent design patterns for enterprise deployments

• Production considerations for scaling across your organization

The entire session uses Copilot Studio's low-code interface. If you can write a prompt, you can build this agent.

Who should attend

This workshop is intended for technical leaders who are fielding requests to "connect Copilot to everything" and need strategies to deliver without compromising security and governance:

• Enterprise Architects designing secure AI data access patterns

• IT Directors evaluating how to scale Copilot beyond basic productivity use cases

• Copilot Studio Developers building agents that need multi-source enterprise data

Join to learn how to give users the AI-powered insights across enterprise systems they want without creating security gaps or encouraging shadow IT workarounds.

Workshop details

📅 Date: Wednesday, January 28th, 2025
⏰ Time: 9:30am PT / 12:30pm ET
⏳ Duration: 45 minutes
💻 Location: Virtual (link provided upon registration)

Prerequisites:

• Microsoft Copilot Studio access (trial accounts work fine)

• Admin or developer access to at least one enterprise system

• Basic familiarity with Copilot Studio interface

Register Now

Can't make it live? Register anyway to receive the recording — but join live for real-time Q&A with Sabin and Jerod.