Building Agents for Microsoft 365 Copilot: From No-Code to Pro-Code

Session Date: May 19, 2025
Duration: 1 hour
Venue: Build 2025 Conference - BRK165
Speakers: Aaron Bjork (Product Management Director, Microsoft Copilot Studio Team), Abram Jackson (PM for M365 Copilot Extensibility, Microsoft)
Link: Microsoft Build 2025 Session BRK165

Introduction and Session Overview
1. Welcome and Speaker Introductions
2. Session Objectives and Agenda
Fundamental Concepts: Copilot and Agents
The Agent-Centric World: Paradigm Shift
Agent Architecture: Technical Deep Dive
No-Code Agent Builder: Democratizing AI Development
Copilot Studio: Professional Agent Platform
Industry Implementation: Nintex Partnership Demo
Professional Development: M365 Agents Toolkit and SDK
Session Wrap-Up and Resources
1. Key Takeaways
2. Additional Sessions and Learning

Introduction and Session Overview

Welcome and Speaker Introductions

Time: 00:00:00 - 00:01:30
Duration: 1m 30s
Speakers: Aaron Bjork, Abram Jackson

Aaron Bjork opened the session by welcoming attendees to the third floor venue on the first day of Microsoft Build 2025 at 6:00 PM. He introduced himself as a Product Manager on the Copilot Studio Team and welcomed Abram Jackson to the stage. Abram Jackson introduced himself as working on Microsoft 365 Copilot extensibility, expressing enthusiasm for the topic.

The speakers established the session’s focus on building agents for Microsoft 365 Copilot and acknowledged the comprehensive agenda ahead, including multiple demonstrations and technical deep dives.

Session Objectives and Agenda

Time: 00:01:30 - 00:02:00
Duration: 30s
Speakers: Aaron Bjork

Aaron outlined the session’s ambitious scope, mentioning “a lot of ground to cover” with “a bunch of demos” planned. The session was structured to move quickly through different development approaches, from no-code solutions to professional development tools.

Fundamental Concepts: Copilot and Agents

Understanding Copilot

Time: 00:02:00 - 00:04:30
Duration: 2m 30s
Speakers: Abram Jackson

Abram provided a foundational definition of Copilot, emphasizing Microsoft’s intentional naming strategy:

“We want Copilot to be your personal assistant, that you use it and it is helping you with your work. We’re pretty intentional about matching the products to that name. It isn’t ‘the pilot,’ it isn’t running off without you, it is your copilot and it is assisting you.”

Key Copilot Characteristics:

Personal Context Awareness: Knows about the user, their work, projects, and colleagues
Assistive Role: Designed to help users achieve more and “look great” using AI assistance
Work Integration: Connected to Microsoft 365 data and enterprise systems
User-Controlled: Always working with the user, not autonomously replacing them

Defining Agents

Time: 00:04:30 - 00:06:00
Duration: 1m 30s
Speakers: Abram Jackson

Abram defined agents as specialized entities with specific business focus:

“Agents are specialists to represent business processes or to understand a space. This is an excellent use for agents of the grounding information, the knowledge, and the APIs, and the data, to process and understand and work through some business process.”

Agent Characteristics:

Business Process Specialists: Understanding specific domains and workflows
Knowledge Integrators: Processing grounded information and real-time data
Implementation Engines: Actually executing business processes
Copilot Connectors: Enabling Copilot interaction with specialized systems

The Symbiotic Relationship

Time: 00:06:00 - 00:07:00
Duration: 1m
Speakers: Abram Jackson

The relationship between Copilot and agents was established as symbiotic, where Copilot provides the personal assistant interface while agents provide specialized domain expertise and system integration capabilities.

The Agent-Centric World: Paradigm Shift

Market Transformation Scale

Time: 00:07:00 - 00:09:30
Duration: 2m 30s
Speakers: Aaron Bjork

Aaron presented the transformational scale of agent adoption, citing IDC research:

“IDC, recently, if you are familiar with IDC, they’re an analyst, they projected that there’s over a billion new business process agents that are going to be created over the next four years.”

Transformation Metrics:

1 billion new business process agents over 4 years
1 agent per 7.5 people globally (including children)
Every function and process in organizations will be affected
Universal business impact across all industries and domains

Orchestra vs Jazz Quartet Analogy

Time: 00:09:30 - 00:12:00
Duration: 2m 30s
Speakers: Aaron Bjork

Aaron introduced a compelling metaphor to illustrate the paradigm shift:

App-Centric World (Orchestra):

Structured coordination: Musicians following sheet music
Specialized roles: Each instrument with specific purpose
Central conductor: Orchestrated control and coordination
Beautiful but rigid: Precise execution within defined parameters

Agent-Centric World (Jazz Quartet):

Improvised collaboration: Musicians playing off each other dynamically
Adaptive interaction: Listening and responding to other participants
No sheet music: Flexible, context-driven performance
Creative emergence: New possibilities through spontaneous collaboration

Copilot as the UI for AI

Time: 00:12:00 - 00:15:00
Duration: 3m
Speakers: Abram Jackson

Abram expanded on Satya Nadella’s concept of “Copilot is the UI for AI”:

“You’ve heard Satya say now lots of times, if you’ve been watching his keynotes, that Copilot is the UI for AI… But without agents, it’s not connected to your work systems.”

Key Points:

Bridge Function: Agents connect Copilot to work systems where business gets done
System Integration: Connection to systems of record, ticket tracking, customer databases
Microsoft 365 Context: Hundreds of millions of workers using Teams, Outlook, Word, Excel, PowerPoint
Security and Governance: Safest, most governed environment for AI deployment

Agent Architecture: Technical Deep Dive

The Five Core Ingredients

Time: 00:15:00 - 00:18:00
Duration: 3m
Speakers: Aaron Bjork

Aaron presented the fundamental architecture components that define any agent:

Agent Architecture Components:
├── Orchestrator: Control layer managing component interactions
├── Model: The "brain" providing reasoning and decision-making  
├── Knowledge: Grounding in verifiable, contextual information
├── Tools: Action-taking capabilities for real-world interactions
└── Triggers: Autonomous invocation and workflow initiation

Orchestrator and Model Synergy

Time: 00:18:00 - 00:19:30
Duration: 1m 30s
Speakers: Aaron Bjork

Aaron explained the relationship between orchestrator and model using an executive function metaphor:

“One of the ways to think about this is think of the model being the brain of the agent, but the orchestrator really being the executive function. They sort of work together hand-in-hand.”

Functional Relationship:

Model as Brain: Core reasoning and intelligence capabilities
Orchestrator as Executive Function: Decision-making about when and how to act
Coordinated Operation: Determining when to invoke models, APIs, or external systems
Strategic Control: Managing the flow of information and actions

Knowledge Grounding

Time: 00:19:30 - 00:20:30
Duration: 1m
Speakers: Aaron Bjork

Knowledge grounding was defined as the foundation of reliable agent operation:

“We call this grounding. And that’s the idea that when you bring knowledge into an agent, you’re really connecting that agent to real, verifiable and contextually relevant information.”

Grounding Benefits:

Prevents Hallucination: Ensures accuracy through verifiable information
Contextual Relevance: Information specific to business domain and use case
Real-time Data: Connection to live systems and updated information
Multi-source Integration: SharePoint, OneDrive, email, Teams, external databases

Tools and Actions

Time: 00:20:30 - 00:21:30
Duration: 1m
Speakers: Aaron Bjork

Tools were positioned as the action-taking components of agents:

“Tools are what we use to do things. So these are the things that actually take action.”

Tool Capabilities:

Real-time Data Access: Live information retrieval and processing
First and Third-party Systems: Comprehensive integration capabilities
Autonomous Action Execution: Tasks performed on behalf of users
API Orchestration: Complex workflows across multiple systems

Multi-Agent Collaboration

Time: 00:21:30 - 00:22:00
Duration: 30s
Speakers: Aaron Bjork

Aaron introduced the concept of agent teams, referencing Satya’s keynote announcement:

“One of the real goals is to build teams of agents where agents can work together, talk together.”

No-Code Agent Builder: Democratizing AI Development

Agent Builder Philosophy

Time: 00:22:00 - 00:25:30
Duration: 3m 30s
Speakers: Abram Jackson

Abram articulated the inclusive philosophy behind Agent Builder:

“The idea and the reason why we wanted to talk through those ingredients of an agent… is that anyone can create an agent with these ingredients. It doesn’t matter who you are and your persona and the way that you operate.”

Universal Accessibility:

All Skill Levels: From end users who “have never opened Visual Studio” to professional developers
Universal Deployment: Web, Windows, Office applications, mobile M365 Copilot app
License Flexibility: Works for both licensed and unlicensed M365 users
Team-Scale Solutions: Perfect for 3-6 person team processes
IT Independence: End users don’t need to request IT projects for automation

IDC Context:

Billion Agent Need: Referencing IDC’s projection of 1 billion agents needed
Personal Usage: Abram uses Agent Builder for his own agents despite having access to pro-code tools

New Features and Capabilities

Time: 00:25:30 - 00:28:00
Duration: 2m 30s
Speakers: Abram Jackson

Abram announced several brand-new features being discussed for the first time at Build:

Enhanced Knowledge Integration:

Team Chat History: Access to organizational conversation context
Email Integration: Grounding in communication patterns and content
Office 365 Entities: SharePoint, OneDrive for Business integration
Organizational Connectors: Whatever connectors the organization has configured
Website Grounding: Specific websites and web domains
Embedded Files: Direct file sharing with agent distribution

Agent Store Redesign:

Ground-up Rebuild: Completely new user experience designed for agent discovery
Unified Navigation: Agents, conversations, pages in integrated interface
Search and Filter: Easy discovery of relevant agents for specific processes
One-click Sharing: Instant distribution to teams and organizations

Live Demo: Build 2025 Concierge Agent

Time: 00:28:00 - 00:33:00
Duration: 5m
Speakers: Abram Jackson

Abram demonstrated the new Agent Store interface and Agent Builder capabilities through a practical example:

Demo Components:

Knowledge Sources: Build 2025 “Book of News” and build.microsoft.com website
Agent Creation Time: “Just a couple of minutes” to complete
Pre-built Capabilities: Code interpreter and conversation starters
Agent Store Interface: New navigation showing agents, conversations, pages
Knowledge Picker: Search functionality for information sources

Live Query Example:

User Query: "Tell me about the new features of Copilot Studio?"
Agent Response: Multi-agent orchestration, model fine-tuning, and many others
Grounding: Website data and official documentation
Result: Accurate, contextual information from verified sources

Multi-Agent Integration Preview

Time: 00:33:00 - 00:36:00
Duration: 3m
Speakers: Abram Jackson

Abram showcased a preview of multi-agent capabilities through a Lumen partnership demonstration:

Researcher Agent Integration:

Task: Create executive financial report on Lumen including financial performance and press releases
Multi-agent Coordination: Researcher agent coordinating with Lumen Quarterly Report Agent
Organizational Customization: “How do you want to structure this report?”
Seamless Integration: “It works the first try every single time”
Enterprise Implementation: Agents built by Lumen for their own organizational use

Future Availability:

Not yet available to all users but coming soon
Preview capability being developed for broader multi-agent support

Copilot Studio: Professional Agent Platform

Platform Positioning and Capabilities

Time: 00:36:00 - 00:38:30
Duration: 2m 30s
Speakers: Aaron Bjork

Aaron positioned Copilot Studio as an enterprise-ready agent development platform:

“Copilot Studio is a SaaS agent platform and it’s designed to help you really quickly and efficiently build agents that are ready to deploy in your organization. You can go from starting point to deployed with metrics, with analytics, with safety rails, with responsible AI literally in hours if not a number of days.”

Platform Capabilities:

Rapid Development: Hours to days vs. weeks for traditional development
Enterprise-Ready: Built-in metrics, analytics, safety rails, responsible AI
Multi-Channel Publishing: M365 Copilot and other destinations
Foundation Model Choice: Managed models plus Azure AI Foundry integration
Knowledge Customization: Multiple source integration and RAG configuration
Tool Integration: First and third-party system connections
Autonomous Workflows: Self-directed task execution

Architecture Control:

Full Access: All shaded areas in the architecture diagram
Model Selection: Managed models and Azure AI Foundry connections
Publishing Channels: Multiple deployment destinations
Knowledge and Tools: Complete customization capabilities
Orchestrator Limitation: Currently managed by Microsoft (future opening expected)

Advanced Model Integration

Time: 00:38:30 - 00:40:00
Duration: 1m 30s
Speakers: Aaron Bjork

Aaron highlighted four key features before the demonstration:

Key Features:

Azure AI Foundry Integration: Access to 1,900+ models through API key configuration
Computer Use Agents (CUA): Preview feature allowing human-like computer interaction
Agent-to-Agent Interactions: Multi-agent coordination and communication
Model Context Protocol (MCP): Connection to external servers and systems

Live Demo: Contoso Employee Resources Agent

Time: 00:40:00 - 00:50:00
Duration: 10m
Speakers: Aaron Bjork

Aaron provided a comprehensive demonstration of Copilot Studio’s enterprise capabilities:

Agent Configuration:

Contoso Employee Resources Agent:
├── Purpose: Employee onboarding assistance for fictitious company
├── Knowledge Source: Company SharePoint site
├── Model: GPT-4o with custom RAG configuration
├── Orchestration: Generative Orchestration enabled
└── Analytics: Complete usage tracking and business outcome measurement

Business Intelligence Features:

Session Analytics: User interaction patterns and usage metrics
Knowledge Source Tracking: Most accessed information and documents
Tool Utilization: Feature usage and workflow optimization data
Test Framework: Automated evaluation and outcome verification

Model Customization:

RAG Configuration: Full control over retrieval and generation parameters
Moderation Levels: Adjustable content filtering and safety controls
Custom Instructions: Specific guidance on model response behavior
Model Selection: Choice between managed models or Azure AI Foundry deployments

Agent-to-Agent Communication

Time: 00:50:00 - 00:55:00
Duration: 5m
Speakers: Aaron Bjork

Aaron demonstrated sophisticated multi-agent orchestration:

Connected Agents:

Contoso Tax Advisor: Specialized tax and financial guidance
Contoso Vacation Advisor: HR policy and time-off management
Dynamic Selection: Host agent choosing appropriate specialist

Live Interaction Example:

User Query: "How much vacation do I get as a new employee and how do I accrue more?"

Process Flow:
1. Employee Resources Agent receives query
2. Identifies vacation-related intent  
3. Invokes Contoso Vacation Advisor
4. Receives structured response with documentation
5. Presents integrated answer with source verification

Result: "New employees receive 2.5 weeks vacation upon hire. 
After one year of service, vacation increases to 3.5 weeks."

Activity Map Visualization:

Chain of Thought Tracking: Complete reasoning process display
Agent Invocation Visualization: Multi-agent workflow mapping
Source Document Linking: Direct validation and drill-through capability
Debugging Tools: Developer visibility into agent decision-making

Visual Studio Code Integration

Time: 00:55:00 - 00:58:00
Duration: 3m
Speakers: Aaron Bjork

Aaron announced and demonstrated a revolutionary development experience:

Major Announcement: New VS Code extension for direct Copilot Studio editing

Extension Capabilities:

VS Code Copilot Studio Extension:
├── Clone Agent: Download agent definitions to local development
├── YAML Editing: Direct manipulation of agent configuration
├── Language Server: IntelliSense and syntax support for agent definitions
├── GitHub Copilot Integration: AI-assisted agent development
└── Push to Server: Seamless deployment back to Copilot Studio

Developer Experience:

Local Development: Full editing capabilities offline
Version Control: Git integration for agent configuration management
IntelliSense Support: Type-ahead and error detection for YAML structures
AI-Assisted Development: GitHub Copilot suggestions for agent configuration
Bi-directional Sync: Seamless movement between VS Code and Copilot Studio

Industry Implementation: Nintex Partnership Demo

Employee Onboarding Use Case

Time: 00:58:00 - 01:00:00
Duration: 2m
Speakers: Video presentation (Nintex partner)

The Nintex demonstration showcased real-world enterprise agent implementation at Safalo Finance:

Business Challenge:

30-minute Manual Workflow: Document download, update, signature coordination
Legacy System Integration: On-premise financial system certification requirements
Multi-step Coordination: HR team managing complex approval workflows
Error-prone Processes: Inconsistent document handling and delays

Automated Solution Architecture

Time: 01:00:00 - 01:03:00
Duration: 3m
Speakers: Video presentation (Nintex partner)

Solution Components:

Employee Onboarding Agent Architecture:
├── Nintex Workflow: Document generation and template population
├── Nintex K2: On-premise system provisioning and certification
├── OneDrive Integration: File storage and organization
├── E-signature Workflow: Automated document signing process
└── Email Coordination: Notification and process management

Workflow Process:

Information Collection: Agent gathers new employee details through conversational interface
Document Generation: Nintex Workflow populates templates with employee data
File Organization: Automated folder creation and document storage in OneDrive
On-premise Provisioning: Legacy system access and certification setup through Nintex K2
E-signature Initiation: Automated delivery of documents for signature

Performance Results

Time: 01:03:00 - 01:03:30
Duration: 30s
Speakers: Video presentation (Nintex partner)

Transformation Metrics:

30 minutes → 2 minutes: 93% reduction in processing time
Zero Manual Errors: Automated accuracy and consistency
Integrated Experience: Single Teams interface for complete workflow
Scalable Process: Consistent experience across all new hires

Professional Development: M365 Agents Toolkit and SDK

Microsoft 365 Agents Toolkit

Time: 01:03:30 - 01:07:00
Duration: 3m 30s
Speakers: Abram Jackson

Abram introduced the professional development platform for complete agent control:

“The Microsoft 365 Agents Toolkit… gives you full control, full power over these agent ingredients… you can change out any part of this architecture.”

Complete Control Architecture:

Professional Developer Control:
├── Orchestrator: Custom logic and decision-making frameworks
├── Models: Any model from any provider, hosted anywhere
├── Knowledge: Custom knowledge sources and processing
├── Tools: Unlimited integration possibilities  
└── Deployment: Full control over hosting and distribution

Platform Features:

Teams Meetings Integration: Agents running directly in meeting contexts
Universal M365 Compatibility: Licensed and unlicensed M365 users
Visual Studio Code Integration: Professional development environment
TypeSpec Support: Simplified API specification management
Office Add-in Integration: Direct manipulation of Word, Excel, PowerPoint

Evolution from Teams Toolkit:

Enhanced Power: Significant additional capabilities beyond previous toolkit
Flexible Architecture: Choice of Microsoft or custom orchestrators, models, tools
Multi-platform Publishing: Teams meetings, M365 Copilot chat, external platforms

Office Add-in Integration

Time: 01:07:00 - 01:10:00
Duration: 3m
Speakers: Abram Jackson

Abram demonstrated breakthrough Office Add-in integration through LexisNexis partnership:

LexisNexis Legal Professional Integration:

Clause Rewriting Capability:

Context Awareness: Agent understands selected text in Word document
Domain Expertise: Legal language and formatting optimization
In-place Editing: Direct document modification with user confirmation
Professional Accuracy: Legal-grade precision and compliance

Shepardize® Feature:

Citation Validation: Automated legal citation checking and verification
Case Law Analysis: Determining if precedents are still valid or overturned
Document Enhancement: Adding visual indicators for citation status
Professional Workflow: Integration with existing legal research processes

Technical Implementation:

Context Transfer: Selected Word content automatically available to agent
Language Model Integration: Custom legal domain understanding
Document Modification: Direct text replacement in Word documents
Visual Enhancement: Automated icon and formatting additions

Microsoft 365 Agents SDK

Time: 01:10:00 - 01:12:00
Duration: 2m
Speakers: Aaron Bjork

Aaron introduced the comprehensive development framework:

“The SDK is a full developer framework and it’s designed to simplify the creation of full stack, multi-channel, enterprise-grade AI agents that can operate across M365, Teams, Copilot Studio and external platforms like Slack, Twilio and others.”

SDK Capabilities:

Full-Stack Development: Complete agent creation framework
Multi-Channel Publishing: M365, Teams, Copilot Studio, external platforms
Enterprise-Grade: Production-ready with enterprise requirements built-in
Multi-Language Support: C#, Python, JavaScript
Complete Control: Choice of orchestrator, models, knowledge sources, deployment

Integration Features:

Azure AI Services: Complete Microsoft AI stack integration
Semantic Kernel: Advanced AI orchestration framework
Third-party AI Services: Provider-agnostic model integration
External Platforms: Slack, Twilio, custom systems

Live Demo: Weather Agent Development

Time: 01:12:00 - 01:16:00
Duration: 4m
Speakers: Aaron Bjork

Aaron demonstrated the Visual Studio template system and rapid agent development:

Visual Studio Template System:

Visual Studio Agent Development Process:
├── Template Selection: Multiple agent patterns available
├── Service Configuration: OpenAI/Azure OpenAI integration
├── Scaffolding Generation: Complete project structure
├── Emulator Testing: Built-in testing and debugging
└── Deployment Options: Multiple distribution channels

Available Templates:

API Integration Agent: Working with external APIs
Basic Teams Agent: Simple Teams-focused functionality
Declarative Agent: Using Microsoft’s orchestration
Empty Agent: Starting from scratch
Weather Agent: Demonstration template with emulator

Development Experience:

Service Selection: Choice between OpenAI and Azure OpenAI
Configuration Input: API keys, endpoints, deployment names
Instant Scaffolding: Complete agent project in minutes
Built-in Emulator: Local testing and debugging environment
Adaptive Card Responses: Rich formatting and interactive elements

Demo Results:

User Query: "Compare the average rainfall of Seattle, Washington to Boston, Massachusetts"
Agent Response: Structured comparison with weather data, formatted as adaptive card
Development Time: Minutes from template to working agent
Integration: Direct Azure OpenAI model connection

Session Wrap-Up and Resources

Key Takeaways

Time: 01:16:00 - 01:17:00
Duration: 1m
Speakers: Abram Jackson, Aaron Bjork

The speakers acknowledged the comprehensive nature of the presentation:

“So I think that might have been a lot for everybody across Agent Builder, Copilot Studio, the Agent Toolkit and the Agents SDK. And we did not talk about all of the new features in all of these things.”

Coverage Summary:

Agent Builder: No-code democratization of agent development
Copilot Studio: Professional platform for enterprise agent deployment
Agents Toolkit: Full-control development for professional developers
Agents SDK: Complete framework for multi-platform enterprise solutions

Additional Sessions and Learning

Time: 01:17:00 - 01:18:00
Duration: 1m
Speakers: Aaron Bjork, Abram Jackson

The speakers directed attendees to additional learning opportunities:

Related Sessions at Build:

Building Agents: More detailed agent construction techniques
Architecting Agents: Advanced architectural patterns
Multi-agent Systems with Copilot Studio: Deep dive into agent orchestration
Agents SDK: Comprehensive SDK coverage
MCP and Copilot Studio: Model Context Protocol implementation
Labs and Hands-on: Practical implementation exercises

Interactive Opportunities:

Agent Instructions Contest: Community engagement activity
OpenHack: Hands-on development experience
Product Team Access: Direct interaction with Microsoft experts
Documentation and Examples: Comprehensive learning resources
Q&A Session: Continued discussion in the hallway after formal session

Availability: Both speakers committed to being available throughout the remainder of Build 2025 for questions and deeper discussions.

References

Microsoft Official Documentation

Microsoft Build 2025 Official Site - Complete conference content, session recordings, and announcements. Essential resource for accessing all Build 2025 materials and staying current with Microsoft’s latest AI and development announcements.
Microsoft 365 Copilot Documentation - Comprehensive guide to Microsoft 365 Copilot features, deployment, and best practices. Critical for understanding the platform foundation that agents extend and integrate with.
Copilot Studio Documentation - Official documentation for Copilot Studio platform, including tutorials, API references, and enterprise deployment guidance. Essential for implementing the professional agent development approach demonstrated in the session.

Development Tools and SDKs

Microsoft 365 Agents SDK - GitHub repository for the Microsoft 365 Agents SDK with code samples, templates, and community contributions. Provides hands-on resources for professional developers building enterprise-grade agents.
Visual Studio Code Extensions Marketplace - Location to find and install the new Copilot Studio extension for VS Code mentioned in the session. Enables professional development workflows with local agent editing capabilities.
Azure AI Foundry - Microsoft’s comprehensive AI platform providing access to 1,900+ models mentioned in the session. Critical for understanding model selection and custom AI service integration options.

Industry Analysis and Research

IDC AI and Automation Research - Source for the billion-agent projection cited in the session. Provides market context and validates the scale of transformation predicted for business process automation.
Nintex Process Automation Platform - Partner platform demonstrated for employee onboarding automation. Shows real-world enterprise implementation of agent-driven process automation and integration patterns.

Technical Standards and Protocols

Model Context Protocol (MCP) Specification - Emerging standard for agent-to-system communication mentioned in the session. Important for understanding how agents integrate with external systems and services.
OpenAPI Specification - Standard referenced for API integration and the TypeSpec alternative mentioned for simplifying agent tool development. Essential for understanding service integration patterns.

Appendix

Demonstration Details and Technical Specifications

Build 2025 Concierge Agent Configuration:

Agent: Build 2025 Concierge
Knowledge Sources:
  - Book of News (Microsoft Build 2025)
  - build.microsoft.com website content
Features:
  - Code interpreter capability
  - Conversation starters
  - Website grounding
  - File embedding
Development Time: "Just a couple of minutes"

Contoso Employee Resources Agent Architecture:

Agent: Contoso Employee Resources
Model: GPT-4o with custom RAG
Knowledge: SharePoint site integration
Connected Agents:
  - Contoso Tax Advisor
  - Contoso Vacation Advisor
Analytics:
  - Session tracking
  - Knowledge source utilization
  - Tool usage metrics
  - Business outcome measurement

Weather Agent SDK Template Configuration:

Template: Weather Agent
Language: C# (with Python/JavaScript options)
Service: Azure OpenAI
Components:
  - Agent instructions
  - Model connection configuration  
  - Emulator for testing
  - Adaptive card response formatting
Development Time: Minutes from template to working agent

Partnership Implementation Details

Nintex-Safalo Finance Integration:

Business Process: Employee Onboarding
Original Duration: 30 minutes manual process
Automated Duration: 2 minutes
Reduction: 93% time savings

Technology Stack:
  - Nintex Workflow (document generation)
  - Nintex K2 (on-premise system integration)
  - OneDrive (file management)
  - E-signature workflow
  - Email coordination
  - Teams interface