streugut.ch
Archive

The direction

Where the AI race is heading

AI competition is moving from the frontier model alone to control of the systems that turn models into governed work. That shift is being carried by agent platforms tied to enterprise data, identity, and workflow, by partner-led rollout channels, and by infrastructure buildout spreading from chips into capacity, financing, and power. What is particular now is how quickly these layers are being packaged into managed stacks across rival clouds, with Microsoft pressing to bind more of that path together.

The move

What actors actually did

Microsoft launches new agentic AI tools while rivals expand coding models and plug-ins

This week, Microsoft rolled out several new AI tools at its Build conference. The company launched Microsoft Scout, an always-on personal agent for Microsoft 365 users, designed to handle tasks across email, calendar, and files with enterprise security controls. Alongside Scout, Microsoft introduced Rayfin, an open-source SDK and CLI to help developers generate agentic app backends, and MDASH, a scanning harness that integrates Defender and GitHub Code Security to orchestrate AI agents across the development lifecycle. Other launches from Microsoft include Work IQ APIs for integrating business context into agentic workflows, Discovery—a platform for research teams and individual use with GitHub Copilot—and Project Solara for embedding AI agents in wearable devices like enterprise badges.

Outside Microsoft, the competition was active as well. OpenAI expanded Codex with six new role-specific plugins, in-app annotations, and a preview of Sites, bringing Codex capabilities to a broader set of knowledge workers. Anthropic grew Project Glasswing, nearly tripling its reach to support over 150 organizations globally in AI-driven vulnerability scanning for critical infrastructure. xAI released Grok Build 0.1, a new coding model now available in public beta, signaling further innovation in agentic coding tasks.

Among hardware partners, NVIDIA announced an expanded collaboration with Microsoft to deliver a unified computing stack for AI on Windows, Azure, and local devices, including new hardware and runtimes. These moves highlight an accelerating pace of agentic AI development, both in software tools and supporting infrastructure.

Sources