Last week, I attended Microsoft Ignite in San Francisco. For four days I immersed myself in keynotes, demos, tech deep dives and lots of AI innovations. The scale is enormous, the pace even faster and the amount of new developments almost impossible to keep up with.
In this blog, I’ll take you through the key insights and themes that were central for me during Ignite.
The keynote at the Chase Center immediately set the tone. Microsoft no longer positions AI as a layer on top of existing products, but as the foundation on which to build in the coming years. This was substantiated with solid announcements, new tools and a clear vision of the future in which agents, models and context are central.
One of the biggest changes is that Copilot will become available by default within Word, Excel, PowerPoint, Outlook and Calendar for all Microsoft 365 licenses. This will make AI support a regular part of the daily work of millions of users.
In addition, Microsoft presented new AI models, multi-agent workflows, autonomous agents that perform entire task sets, and tools that allow developers to build applications based on prompts, but there are many other things that came up that I’m happy to share with you below.
Agents as digital colleagues
Agents were a major theme during Ignite. Microsoft sees agents as autonomous digital colleagues who perform tasks, make decisions and collaborate on workflows shape. They become part of the structure in which organizations will work and develop.
During several sessions, Microsoft showed how agents are used internally. Examples included ES chat, an internal agent that supports developers by gathering technical context from projects, and the Azure SRE agent that detects incidents, analyzes them, opens a GitHub issue and automatically prepares a pull request to resolve the issue. Microsoft emphasizes that the role of agents goes beyond tools. They see agents as partners to be involved in creation processes. In doing so, Foundry works as a central platform in which agents are built, combined and managed. Foundry also provides support for model routing, monitoring, feedback loops and multi agent workflows.
Agents in Entra ID
One notable development was that agents created their own identity within Entra ID. As a result, they no longer become just technical objects, but real entities within the organization.
Agents are given their own e mail address, position in the organizational structure and an owner or manager. This allows agents receive assignments via e mail and perform tasks linked to the organizational structure. An interesting case came from Commerzbank, which deployed a digital assistant with a realistic avatar. That avatar proved to be better accepted, especially by older customers, than a drawing style. As a result, adoption is higher and users perceive the assistant as more reliable.
Azure innovations
Azure received during Ignite much attention, especially in the areas of scalability and support for AI workloads. Microsoft showed how the infrastructure has evolved to support future AI applications to carry.
A session with Mark Russinovich included a new tour of the Atlanta data center, which runs almost entirely on renewable energy and uses a closed cooling system that only needs to be replenished after six years. Microsoft showed Azure Boost, which offloads workloads for higher efficiency. Also passing by was an introduction to optical computing, where computing is done with light instead of electricity. This is more energy efficient, faster and promising, but still premature. Finally, the Azure migration agent was demonstrated that automatically converts legacy environments to modern Azure architectures such as Kubernetes clusters.
Azure DocumentDB
Microsoft introduced Azure DocumentDB as a new NoSQL option. It is fully compatible with MongoDB and easier to manage than CosmosDB.
Azure DocumentDB offers less advanced and less scalable capabilities than CosmosDB, but is also cheaper as a result. It supports MongoDB syntax and drivers and offers a free variant with 32 gigabytes of storage. Visual Studio Code has a native interface that allows easy management of databases. Microsoft also showed a large language model that automatically advises which fields should be indexed.
AKS Automatic
Azure Kubernetes Services can be quite complex. Microsoft introduced AKS Automatic to Kubernetes more accessible and easier to manage.
AKS Automatic automatically configures default settings so users can deploy containers without in-depth knowledge deploy. A new feature is the Automatic Managed System Node Pool, in which node pools are preconfigured and cannot be manually modified. This prevents configuration errors and makes Kubernetes more user-friendly for organizations that do not have extensive in-house expertise.
SharePoint Embedded
SharePoint Embedded allows files within a proprietary application to be stored in SharePoint while only that application can access the files.
Files are stored within the Microsoft 365 tenant and benefit from all compliance, governance and security capabilities of SharePoint. During Ignite it became clear that features will soon be added for archiving and back up, allowing applications to have a full lifecycle for documents can support. One example mentioned was M Files, which already uses this integration.
SharePoint Advanced Management
SharePoint Advanced Management received several enhancements relevant to administrators and organizations with complex structures.
New features include reports for oversharing, lifecycle management, permissions reports per user and Microsoft 365 Storage Pay as you go. In addition, file-level back up will be available. Microsoft emphasized that archiving and backups are intended not only for recovery and storage management, but more importantly to prevent Copilot using old or deleted documents as source material for replies.
Transparency about Azure incidents
Microsoft showed at Ignite that transparency is central to dealing with incidents. They actively share what happens and what lessons they learn.
In every major incident, an Azure Incident Report published in which those responsible explain in a video what happened. In addition, Post Incident Reports available describing causes, consequences and improvement actions. With this, Microsoft provides insight into their processes and shows how incidents are analyzed and resolved.
New Copilot features
GitHub Copilot received several new features during Ignite that speed up and improve the development process. New keyboard shortcuts and workflows came along that provide more control and accuracy.
New features include Ctrl plus right arrow to accept suggestions word by word and Ctrl plus I for quick actions. The Copilot Instructions file allows for coding establish standards. There is a clear distinction between Agent Mode, which works autonomously and performs longer tasks, and Edit Mode, which focuses on controlled modifications. The recommended practice is to first generate a plan in Agent Mode and execute it incrementally in Edit Fashion. Microsoft also shared repositories with examples and standard prompts, including Awesome Copilot.
Custom agents and MCP
There was a lab in which participants built their own agents that linked to external tools via the Model Context Protocol.
The lab showed how to build a custom MCP service and link it to an agent created in the studio. Also on the program was an integration with Azure AI Search. Although the lab was not completely flawless, it gave a good idea of how custom agents can be linked to applications, databases and external services.
AI functionality without an internet connection
One of the labs showed how to implement powerful AI functionality run locally on Windows without cloud or Internet connection. This has benefits for privacy, security and cost.
In the lab, an application was built in which the DOS manual was indexed, converted to text and made searchable via semantic search. This demonstrated how search results are linked to a language model that neatly formulates the answer. It also showed how text and images can be indexed locally and how Windows AI interfaces can be used without connecting to the cloud.
Ignite brought together many new developments in AI, agents, Azure, development tools and Microsoft 365. The sessions showed what capabilities Microsoft has added or will soon make available and how these components are reflected in different products.
For organizations, this mainly means that new features and services are becoming available that they can use when relevant. It is valuable to keep track of which capabilities fit the way teams work and which features can play a role within their own environment.
Want to learn more about any of these topics or spar about what these developments can practically mean within your Microsoft 365 environment? Feel free to contact us. We’d love to think along with you.
