Microsoft Corp. is upgrading its Azure Foundry service with new features that will make it easier for developers to build artificial intelligence applications.

The company debuted the enhancements today at an event marking its 5oth anniversary. Executives also introduced a new version of Copilot, Microsoft’s consumer-focused chatbot. It now offers more personalization options and a feature that can perform tasks in the user’s browser.

Azure Foundry, the cloud service that Microsoft updated today, makes it easier for developers to build AI applications. It includes prepackaged AI models that remove the need to train a custom algorithm. There are also development tools that promise to provide further time savings. 

As part of today’s update, Microsoft is expanding Azure Foundry’s lineup of development tools with an offering called the AI Red Teaming Agent. Currently in public preview, it can automatically test AI applications for safety risks. That spares software teams the hassle of running tests manually and thereby speeds up development.

The AI Red Teaming Agent is powered by an open-source framework called PyRIT. The latter software, which was also developed by Microsoft, can send an AI application a large number of malicious prompts and determine which ones lead to harmful output. If it discovers weak points in an application, the AI Red Teaming Agent automatically turns its findings into a report.

“You can generate a score card of the attack probing techniques and risk categories to help you decide if the system is ready for deployment,” Microsoft Senior Product Manager Minsoo Thigpen wrote in a blog post. “Findings can be logged, monitored, and tracked over time directly in Azure AI Foundry, ensuring compliance and continuous risk mitigation.”

The tool is joined by a set of new “agentic evaluations.” According to Microsoft, they provide metrics that can be used to measure the risks associated with an AI model and its output quality.

For developers that use its Visual Studio Code code editor, the company is rolling out a new Azure Foundry extension. The plug-in provides the ability to perform tasks such as deploying an AI model in the cloud without leaving the editor. Additionally, an Azure Foundry component that makes it easier for developers to integrate multiple AI agents into an application is now generally available. 

“When we mark an API as generally available it means that we have high confidence in the quality of the surface for building AI applications and that we can support and maintain the API going forward,” explained Microsoft Group Product Manager Shawn Henry.

Image: Microsoft