News from the AI & ML world
DeeperML - #microsoftbuild
|
@github.blog
//
Microsoft is aggressively integrating artificial intelligence into its developer tools and workflows, demonstrated by the announcements made at Microsoft Build 2025. A key focus is streamlining the development process by leveraging AI to accelerate the creation of issues and pull requests within GitHub, a fundamental aspect of software development. This involves utilizing GitHub Copilot to draft issues and assigning them to a coding agent for asynchronous execution, resulting in quicker generation of pull requests. This approach aims to maintain familiarity for developers while significantly improving efficiency and consistency. The importance of well-structured issues and pull requests remains paramount, even in AI-accelerated workflows, as they provide shared context, facilitate asynchronous coordination, support audit and analytics, and enable automation hooks.
BenchmarkQED, an open-source toolkit, has been introduced to automate the benchmarking of Retrieval-Augmented Generation (RAG) systems. This toolkit includes components for query generation, evaluation, and dataset preparation, all designed to support rigorous and reproducible testing. BenchmarkQED complements Microsoft's open-source GraphRAG library, enabling users to conduct GraphRAG-style evaluations across various models, metrics, and datasets. The toolkit addresses the growing need to benchmark RAG performance as new techniques emerge, particularly in answering questions over private datasets. BenchmarkQED facilitates the comparison of RAG methods, including LazyGraphRAG, against competing approaches like vector-based RAG with a 1M-token context window. Tests show that LazyGraphRAG demonstrates significant win rates, especially on complex, global queries that require reasoning over large portions of the dataset. This advancement aims to enhance the performance of RAG systems, particularly in scenarios where conventional vector-based RAG struggles with questions requiring an understanding of dataset qualities not explicitly stated in the text. The toolkit represents a major step forward in automating and scaling RAG benchmarking. Recommended read:
References :
@blogs.microsoft.com
//
Microsoft Build 2025 showcased the company's vision for the future of AI with a focus on AI agents and the agentic web. The event highlighted new advancements and tools aimed at empowering developers to build the next generation of AI-driven applications. Microsoft introduced Microsoft Entra Agent ID, designed to extend industry-leading identity management and access capabilities to AI agents, providing a secure foundation for AI agents in enterprise environments using zero-trust principles.
The announcements at Microsoft Build 2025 demonstrate Microsoft's commitment to making AI agents more practical and secure for enterprise use. A key advancement is the introduction of multi-agent systems within Copilot Studio, enabling AI agents to collaborate on complex business tasks. This system allows agents to delegate tasks to each other, streamlining processes such as sales data retrieval, proposal drafting, and follow-up scheduling. The integration of Microsoft 365, Azure AI Agents Service, and Azure Fabric further enhances these capabilities, addressing limitations that have previously hindered the broader adoption of agent technology in business settings. Furthermore, Microsoft is emphasizing interoperability and user-friendly AI interaction. Support for the agent-to-agent protocol announced by Google could enable cross-platform agent communication. The "computer use" feature for Copilot Studio agents allows them to interact with desktop applications and websites by directly controlling user interfaces, even without API dependencies. This feature enhances the functionality of AI agents by enabling them to perform tasks that require interaction with existing software and systems, regardless of API availability. Recommended read:
References :
@blogs.microsoft.com
//
Microsoft is unveiling new AI agent customization and oversight features at Build 2025, marking a significant step towards making AI agents more trustworthy and secure. This move aligns with the company's broader strategy of empowering businesses and individuals with custom-made AI systems. A core component of this initiative is extending Zero Trust principles to the agentic workforce, ensuring that AI agents operate within a secure framework. The announcements point to the tech giant's broader strategy of equipping businesses and individuals with their own, custom-made AI systems.
Microsoft is introducing Microsoft Entra Agent ID, a feature that extends identity management and access capabilities to AI agents. This aims to enhance trust when AI agents handle user data, addressing a critical concern as AI becomes more integrated into daily operations. The era of AI agents is upon us, driven by advancements in reasoning and memory, making AI models more capable and efficient. We’re seeing how AI systems can help us all solve problems in new ways. Microsoft is also launching tools to streamline AI agent development. One notable announcement is the general availability of the Azure AI Foundry Agent Service, which was announced last fall. The platform allows developers to build, manage and scale up AI agents that automate business processes. To that end, Microsoft is previewing several new developer tools for building agentic applications in Microsoft Teams. They support secure, peer-to-peer communication via the A2A protocol, agent memory for contextual user experiences and improved developme Recommended read:
References :
Michael Nuñez@AI News | VentureBeat
//
References:
techstrong.ai
, AI News | VentureBeat
,
Microsoft has significantly expanded its Copilot Studio platform at Build 2025, introducing multi-agent systems designed to revolutionize enterprise workflows. This expansion allows AI agents built with Copilot Studio, Microsoft 365, Azure AI Agents Service, and Azure Fabric to collaborate on complex business tasks, delegating tasks to each other to complete processes. According to Ray Smith, VP of AI Agents at Microsoft, this approach addresses the challenges of creating reliable processes within a single agent, improving maintainability, simplifying solution building, and enhancing overall reliability. An example scenario involves a Copilot Studio agent pulling sales data from a CRM, handing it to a Microsoft 365 agent to draft a proposal in Word, and then triggering another agent to schedule follow-ups in Outlook.
Microsoft is also committed to building an open agentic AI ecosystem, demonstrated by its support for Anthropic’s Model Context Protocol (MCP). GitHub and Microsoft have joined the steering committee for MCP, an open-source standard designed to enable AI tools to directly access data systems. This initiative aims to simplify data connections for AI models, allowing them to fetch information from apps and data stores more easily and safely. The MCP allows any AI client to communicate with any MCP server through a standard interface, eliminating the need for custom connectors for each data source. Microsoft plans to add first-party support across Azure and Windows to assist developers in exposing app features as MCP servers. Furthering its commitment to open AI, Microsoft is making available a Model Router and a Model Leaderboard to optimize model selection for specific tasks. The Model Router enables real-time selection of the best model for a given query, while the Model Leaderboard ranks AI models based on their performance across various categories. Additionally, Microsoft is offering pre-built agents, custom agent building blocks, and a software development kit (SDK) based on MCP. The company is also extending the observability capabilities of its Azure AI Foundry to agents, offering comprehensive tools and platforms to foster an open and collaborative AI ecosystem, referred to as the Agentic Web. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |