Project: [enterprise agents] - Multi Agent using Agents Toolkit to find and manage volunteering opportunities

Track

Creative Apps (GitHub Copilot)

Project Name

MultiAgent using Agents Toolkit to find and manage volunteering opportunities

GitHub Username

reshmee011

Repository URL

https://github.com/reshmee011/AgentsToolkit_IntelligentApp/

Project Description

This project is an intelligent Orchestrator Agent that interprets a user’s intent and dynamically coordinates actions across multiple Microsoft 365 and Azure services. It acts as a central decision-maker, selecting the appropriate connected agent or service to fulfil the user’s request automatically.

Problem It Solves
Modern organisations rely on multiple systems (SharePoint, email, workflows, monitoring tools, etc.), but users often need to manually switch between platforms to retrieve information, send communications, update records, and track activity.

This project solves that fragmentation by:

• Automating multi-step workflows from a single user request
• Reducing manual effort and context switching
• Ensuring consistent data handling and traceability
• Enabling seamless integration across Microsoft 365 and Azure services

Instead of users navigating different tools, the orchestrator understands the intent and executes the necessary actions behind the scenes.

How It Works
Based on the user’s intent, the Orchestrator Agent selects and triggers the appropriate connected capability:

• Information Retrieval
  Retrieves relevant content from Azure AI Search connected to a SharePoint document library.

• Email Automation
  Calls a Power Automate flow to generate and send emails automatically.

• Data Management
  Connects to Microsoft Graph to create and retrieve list items within Microsoft 365.

• Monitoring & Observability
  Connects to Azure Logic Apps to log prompts and responses into Application Insights for tracking, analytics, and auditing.

Key Features

• Intent-based orchestration
• Multi-agent action routing
• Secure OAuth authentication
• Seamless Microsoft 365 integration
• Automated workflow execution
• Activity logging and telemetry tracking
• Scalable and modular architecture

Demo Video or Screenshots

Demo Video:https://www.youtube.com/watch?v=C4yJfcK1kew&t=18s

Primary Programming Language

Other

Key Technologies Used

Technologies Used

• Microsoft 365 Agents Toolkit – Agent framework and orchestration
• TypeSpec
• OAuth Authentication – Secure access control to M365 Graph and Power Automate
• Key based authentication to Azure AI Search and Logic App
• Power Automate – Workflow automation
• Azure Logic Apps : Process integration and logging
• Microsoft Graph : Microsoft 365 data access
• Azure AI Search : Intelligent document retrieval
• Connects to other OOB declarative agents like SharePoint Page Agent

Submission Type

Individual

Team Members

No response

Submission Requirements

• My project meets the track-specific challenge requirements
• My repository includes a comprehensive README.md with setup instructions
• My code does not contain hardcoded API keys or secrets
• I have included demo materials (video or screenshots)
• My project is my own work with proper attribution for any third-party code
• I agree to the Code of Conduct
• I have read and agree to the Disclaimer
• My submission does NOT contain any confidential, proprietary, or sensitive information
• I confirm I have the rights to submit this content and grant the necessary licenses

Quick Setup Summary

1.Clone the repo
2. Configure each individual declarative agent using the respective read me file and deploy
3. Update the Ids of the connected agents within the orchestrator agent to refer to each standalone agent and deploy

Technical Highlights

Rather than creating predefined flows for every scenario, I used an orchestrator agent that determines the appropriate action based on intent classification in a multi agent scenario

This decision:

• Reduced workflow duplication
• Made the system extensible
• Allowed new actions to be added without redesigning the core architecture

Challenges & Learnings

One of the main challenges was that there were no existing samples demonstrating how to connect to a Power Automate flow from the orchestrator. I had to explore and experiment to determine the correct integration pattern and authentication flow.

Additionally, the original scaffolded solution for connected agents referenced an older version that did not support the worker_agents decorator. This decorator is not supported in TypeSpec, and my objective was to build the solution entirely using TypeSpec. As a result, I had to redesign parts of the architecture to ensure compatibility while maintaining the intended multi-agent behavior.

Adopting a multi-agent approach enabled a modular and extensible design. Each capability can function independently, making the system easier to maintain and allowing components to be reused across different solutions.

However, working with multiple agents introduced behavioral differences. For example:

• Citations from connected agents are not always returned consistently.

• Response times can sometimes be slower due to cross-agent coordination and orchestration overhead.

Despite these challenges, the modular multi-agent architecture provides greater flexibility and long-term scalability compared to a monolithic design.

Contact Information

No response

Country/Region

United Kingdom

This is a project submission rather than a scoped change in the agentsleague repository. Start with the linked project's README and quick setup summary, then review the declarative agents, orchestrator configuration, and connected service integrations; completion means the agents are deployed, connected IDs are configured, and the demonstrated volunteering workflow operates end to end.