Microsoft Unveils IQ and Rayfin to Centralize Enterprise AI Agent Governance and Context

Microsoft used its Build conference to tackle a growing problem in enterprise AI: agents that work in isolation, create duplicate data stores, and operate without a shared understanding of company rules or systems. The company’s answer centers on two new pieces of technology, Microsoft IQ and Rayfin, which are designed to give AI agents a common layer of context and governance, according to VentureBeat’s reporting from the event.

The issue Microsoft is trying to solve is straightforward but significant. As companies deploy more AI agents for coding, workflows, and office tasks, each tool often starts from scratch, with no memory of where business data lives or what policies apply. That creates silos and makes it harder for firms to manage security, compliance, and consistency, VentureBeat reported.

According to the same report, Microsoft IQ is intended to provide agents with a persistent knowledge layer so they can draw on business context instead of relearning it every time. Rayfin, meanwhile, is being positioned as part of Microsoft’s broader effort to let developers connect and govern agent behavior more consistently across systems. The company is framing both tools as infrastructure for enterprises that want AI agents to be useful without becoming fragmented islands of data and logic.

The push comes as Microsoft and its rivals race to make agents more capable and more enterprise-ready. OpenAI, for example, announced new Codex tools aimed at white-collar work, including role-specific plug-ins and enterprise features, while Microsoft also introduced new tools for controlling agent behavior and testing it through portable policy files and evaluation frameworks, according to TechCrunch. Together, those releases show a broader industry shift from experimental assistants toward business software that can be managed like other corporate systems.

Microsoft’s Build announcements also extended beyond software. The company unveiled the Surface RTX Spark Dev Box, a local AI development machine built for running and fine-tuning large models on the desktop instead of relying entirely on cloud computing, according to Microsoft’s own blog and multiple reports. That device is meant to help developers work on agentic AI locally, while still connecting to the cloud when needed.

The hardware and software launches suggest Microsoft is trying to control both sides of the enterprise AI stack: where models run, and how agents access data and follow rules. That matters for companies that want faster deployment of AI tools without losing visibility into security, compliance, or data governance.

Other Build-related announcements pointed in a similar direction. Microsoft is also testing a wearable access badge and a desktop device for workers, according to the BBC, underscoring how the company is exploring AI and workplace hardware together as part of a wider effort to reshape office productivity.