Microsoft opens its AI infrastructure to external models and expands features for developers

During the Microsoft Build 2025 event, the company founded by Bill Gates reaffirmed its intention to make artificial intelligence an essential technology in everyday life.

The strategy goes beyond simple automated assistants and aims for an ecosystem where advanced models, customized capabilities, and collaborative tools converge in the cloud to facilitate the development and deployment of AI solutions in multiple environments.

One of the most notable announcements was the integration of Grok 3 and Grok 3 mini, models developed by xAI, Elon Musk's artificial intelligence company. Both will be available on the Azure AI Foundry platform, not only as hosted models, but also billed directly by Microsoft.

This marks an important step toward further opening up the company's AI infrastructure, which now allows developers to choose from more than 1,900 models offered by partners and Microsoft itself.

Grok Arrives on Azure: AI with Reasoning from First Principles

Elon Musk, present at the event, explained that Grok seeks to provide reasoned answers based on what he called "first principles." According to Musk, the approach of his model is to apply "the tools of physics to thinking," with the goal of reaching truths with a minimal margin of error.

This philosophy distinguishes it from other generative models, and its incorporation into Azure allows developers to customize, integrate, and manage AI agents with deeper reasoning capabilities according to their own needs.

Furthermore, this integration implies a direct collaboration between Microsoft and xAI, a striking situation considering that the previous year Musk had accused Microsoft and OpenAI of forming a de facto monopoly in the field of artificial intelligence.

The alliance suggests a change of tone and could mean greater diversity of options for developers using cloud infrastructure.

GitHub Copilot evolves with asynchronous agents and new features

Another innovation presented at Build 2025 was the introduction of the first asynchronous programming agent within GitHub Copilot. This agent not only provides real-time assistance but can also perform tasks such as fixing bugs, generating new features, and improving documentation without direct intervention, representing a breakthrough in software development automation.

Microsoft has also expanded the capabilities of GitHub Models, including tools for managing prompts, performing basic evaluations, and applying controls designed for business environments.

This ecosystem expansion seeks to offer greater control and adaptability to teams that need to integrate AI into demanding development cycles.

Microsoft also expanded the functionality of Copilot Studio, its environment for creating custom assistants using a low-code approach.

One of the main new features is the addition of the multi-agent orchestration feature, which allows developers to coordinate multiple agents to work together to solve complex tasks.

This feature allows combining different skills of specialized agents, achieving greater efficiency and adaptability in business environments.

Along these lines, Microsoft introduced the Microsoft 365 Copilot Tuning service, which allows users to use their own data, processes, and workflows to train custom models. This capability reinforces the trend of bringing AI closer to the specific needs of each organization, without the need for advanced programming knowledge.

With these announcements, Microsoft consolidates its vision of turning artificial intelligence into a cross-functional, adaptable, and increasingly integrated tool for work and personal dynamics.

The opening to third-party models like Grok, along with automation and collaborative development tools, outlines a strategy where the cloud, customization, and interoperability are central pillars of the future of AI.