Microsoft’s Ignite 2025 conference is underway in San Francisco. The announcements coming out of Ignite support our customers’ core goal: boosting productivity through enterprise intelligence. In this post, we’ve summarized the most relevant updates from a strategic and architectural perspective.
The theme is clear: the era of AI experimentation is over. Now it's time to build solutions that can scale into production. As we already have more than enough AI models and development tools, the real focus now shifts to lifecycle and quality management of AI solutions. Not every release is available immediately, but these announcements offer strong guidance for long-term enterprise architecture planning. Also, many tools are ready to be tested today.
There’s clear demand for this next step. Microsoft’s own research highlights the same obstacles we've seen in real-world projects: poor data quality, overly IT-focused initiatives, difficulty managing AI tools, and a tendency to get stuck in pilots without the boldness to push into production and innovate in day-to-day work.
IQ: An intelligence layer across the entire organization
Microsoft unveiled three new IQ products that together form what could be described as an intelligence layer spanning the entire organization. This layer sits on top of foundational data stores and core business systems, connecting structured and unstructured knowledge across workflows.
- Work IQ captures knowledge embedded in everyday productivity tools, such as documents, emails, Teams chats. It also learns from users’ working habits.
- Fabric IQ lights up the semantic models built into Fabric or Power BI, enabling AI agents to tap into operational data with less friction and more impact.
- Foundry IQ centralizes data storage and retrieval, enabling AI agents to access shared, governed sources without duplicating effort. Instead of each agent independently performing retrieval-augmented generation (RAG), multiple agents can build on the same reliable data foundation.
This kind of intelligence layer is something many of our clients have asked for. Traditional data platforms excel at answering “what” questions with raw business data. But the “why” of business often lives in documents, emails, and conversations. Without access to that context, AI agents struggle to grasp the real processes that drive business.
A major advantage of Microsoft’s IQ suite is its built-in security and access control. With Microsoft Purview, permissions defined at the data source level are respected across all layers. This removes one of the biggest challenges in AI projects - ensuring protected content isn’t accidentally exposed and reducing the burden on developers managing data privacy.
Even for organizations not yet deploying agents, the path to AI readiness is now more practical. Companies can begin publishing data via Fabric IQ and Foundry IQ, gradually making knowledge accessible. Work IQ’s API support also allows insights stored in environments like SharePoint to be used across different apps and platforms. Together, these tools offer flexibility and a secure foundation for building enterprise-grade AI solutions.
The full spectrum of AI comes together
Until now, enterprise AI has followed two separate tracks. On one hand, personal productivity has been enhanced through Copilot tools. On the other, deeper process-driven AI has required more traditional, customized development.
The latest Ignite announcements show a clear effort to bring these together. From both the agent management and IQ platform perspectives, Copilot extensions, low-code solutions built in Copilot Studio, and fully custom AI applications now operate on equal footing. They can be managed centrally and all have access to the same governed data environment.
This is a meaningful shift. Copilot Studio’s evolution empowers innovators to automate more of their own work. At the same time, the most successful personal agents can now transition more easily into centrally managed IT solutions, because both the management layer and data stack treat all agents equally, regardless of how they were built.
New tools for managing AI agents
As AI agents start appearing everywhere—from SaaS tools to custom development and even those built directly by knowledge workers—organizations face a growing management challenge. Microsoft’s response is a new centralized control platform called Agent 365, designed to bring these diverse agents under one roof. Its mission is simple: prevent shadow AI from becoming a problem before it spirals out of control.
Agent 365 provides a unified service not only for maintaining an inventory of agents, but also for managing access rights, security, and organizational policies. For IT teams, the promise is clear: a single view of all agents running in the organization regardless of whether they were built using Microsoft technology or not. For developers, this experience is enhanced with Foundry Control Plane, a supporting service for content oversight, threat monitoring, and mapping how agents interact with each other.
The message behind these tools is unmistakable: the industry is well aware of how messy decentralized AI adoption can become, both through personal automation or agents bundled into every app. Robust management tools are essential. Not just for access control, but also for enterprise governance and troubleshooting. It’s encouraging to see Microsoft’s platform supporting third-party agents right from the first release.
Heading in the right direction, but not yet complete
Microsoft’s Ignite announcements show a clear intent: to remove barriers of AI adoption by making it easier to connect agents with broad enterprise data and reducing concerns around governance and control.
These are steps in the right direction. From an enterprise architecture perspective, this clarity is welcome. It’s now evident, for example, that consolidating business data in Fabric is a smart move because with the IQ layer, that data can power everything from Copilot-based tools to fully customized AI solutions.
However, technology alone doesn’t solve every challenge related to deploying AI agents. A core issue remains: too many AI projects begin without strong business involvement, resulting in thin, ineffective outcomes. Likewise, AI-related change management and meaningful impact tracking still fall squarely under classic leadership responsibilities—areas that Microsoft doesn’t address, but where Norrin actively supports our clients.
If you'd like to discuss how the Ignite releases could shape your organization or how to build practical enterprise intelligence, get in touch: jouni.heikniemi@norrin.com. We're here to help.
