Transforming software delivery process with intelligent agent orchestration
ReleaseManager_Intro.mp4
In modern software development, Release Managers play a pivotal role bridging the gap between development and operations. As the orchestrators of software deployment, they ensure that releases are timely, efficient, and risk-mitigated. However, increasing complexity and fragmentation across systems have made this role more challenging than ever.
The Release Manager Assistant (RMA) is a solution accelerator designed to augment release managers with AI-driven intelligence, multi-system integration, and real-time decision support. It simplifies the release lifecycle from planning to post-deployment analysis, all through a unified and contextual interface.
Watch a quick demo below!
Release.Manager.Demo.mp4
- Fragmented release data across platforms
- Lack of real-time release health visibility
- High manual effort in compiling readiness reports
- Difficulty managing cross-team/service dependencies
- Visualizing complex data across multiple systems
- Notifying stakeholders on change in the system
- Real-time dependency mapping and release health assessment
- Visualization of complex data
- Smart notifications
- Issue and stakeholder updates via integrated channels (emails)
- Seamless integration with tools like:
- JIRA
- DevOps (MySQL database and/or API integration)
- Microsoft Graph APIs
- Unified, contextual insights from disparate systems
The architecture is built on a modular and secure AI-native design leveraging:
- Azure AI Foundry service for agents in cloud and supporting tools (Code Interpreter).
- Semantic Kernel for agent orchestration, agent memory and tools integration.
- Microsoft Graph API for enterprise identity and email notification integration*.
- Azure Key Vault & Azure Storage for secure secret and data management.
For setting up email notifications, additional setup is required. For instructions, please refer to this guide: Microsoft Email Notifications Setup
Release.Manager.Architecture.Walkthrough.Low.Bitrate.mp4
This diagram provides a visual representation of how the JIRA Agent, DevOps Agent, Visualization Agent and Notification Agent collaborate to streamline release planning and execution for a release manager.
Diagram above highlights components in the end-to-end system run that run locally (highlighted in green) and components that run in the cloud (highlighted in blue).
Note: This solution accelerator is designed to be adaptable. You can customize integrations and workflows based on your internal tooling landscape.
For detailed setup instructions, please follow the guide here: SETUP INSTRUCTIONS
- Azure Subscription
- Access to Azure OpenAI and Azure AI Foundry
- Local/Cloud instance for JIRA, DevOps (MySQL connection), etc.
- python >= 3.12
- Docker runtime for running service containers locally (For more information on local execution, refer to this guide: LOCAL EXECUTION IN DOCKER)
Synthetic dataset is included in the solution under /data directory for reference purposes.
- ✅ Multi-system integration baseline
- ✅ Natural language release query support
- 🚧 Azure DevOps integration
- 🚧 Teams notifications support
The dataset in this project is released under the Community Data License Agreement – Permissive, Version 2.0 - CDLA, see the LICENSE-DATA file.
This project is licensed under the MIT License. See LICENSE for details.
- Azure AI Foundry Agent Service
- Semantic Kernel Agent Framework
- Semantic Kernel Memory
- Azure AI Foundry - Code Interpreter
Developed with ❤️ by Microsoft