From Apps to Agents: Microsoft Just Made Its Boldest Move Yet
Earlier this week, Microsoft reshaped its leadership and if you read between the lines, it wasn’t just about org charts.
It was a statement about the future of software.
Ryan Roslansky, CEO of LinkedIn, has now taken on a broader role overseeing all of Microsoft’s consumer AI efforts, including Copilot. Why does this matter?
Because LinkedIn has quietly become one of Microsoft’s most successful AI experiments, blending personalized recommendations, career agents, and content generation tools at scale.
This leadership move signals that Microsoft sees the next frontier not in standalone apps, but in AI agents that know you, grow with you, and work on your behalf.
In Fact, Patrick Moorhead (Founder, CEO, Chief Analyst at MoorInsightsStrategy) even wrote that “The Microsoft reorg today means more than the headlines are indicating. This is big and is about the ‘SaaS is dead’ narrative we’ve heard from Satya which has become and industry conversation.”
We already had hints about where Microsoft is going, Satya Nadella, Their CEO publicly spoke about it some time ago:
We’re entering the post application era.
Instead of opening five apps to get things done, you’ll speak to one intelligent agent.
Instead of interfaces, we’ll rely on intent and context.
Instead of learning software, software will learn you.
Microsoft’s restructuring reflects this shift:
From tools you use, to teammates you trust.
From GUIs to conversations.
From productivity apps to personal AI ecosystems.
This changes everything, from product design to business models to skill sets.
The age of apps brought us here. The age of agents will take us forward.
This transformation is already under way, with leading organizations enacting plans to augment their human workforce with thousands of agents per employee, redefining efficiency and productivity.
In the Agent Era, systems don’t just run code, they reason, plan, and act. Agents don’t follow scripts, they respond to context, use tools, and adapt in real time.
The problem? This shift will completely break the underlying infrastructure, built for the Application Era.
The transition from applications to agents brings entirely new levels of scale and performance organizations need to take on:
- Moving from 10s of applications to potentially trillions of agents.
- From monolithic application stacks to distributed data fabrics across cloud, data centers and edge.
- From hundreds of CPUs to millions of GPUs needed to power these agents.
- From batch processes to data being embedded in real time.
- From terabytes of structured data to exabytes of natural, unstructured data agents will use to perceive the world around them.
Here’s the thing: this future is not decades away, it’s already here.
Whether customers are exploring copilots, deploying ChatGPT style solutions, or building toward an agent workforce, they must prepare for the most demanding AI implementations.
The world needs a new OS to face these challenges, so VAST introduced the VAST AI Operating System, the first platform built from the ground up to power the Agent Era.
It’s a data centric, hyperscale operating system that gives agents all the resources they need to thrive.
The VAST AI OS turns your infrastructure into a thinking machine and makes it achievable for every organization to realize the benefits of AI.
The VAST AI Operating System
So is what VAST sells really an operating system? When you layer it against the services of a traditional OS built for applications, you can see how the VAST AI OS delivers services modernized for the new requirements.
Click on each icon on the graphic below to learn how to explain how the VAST Operating System works together to serve
VAST DataStore
The VAST Datastore replaces the file system with a an efficient hyperscale data environment, giving agents fast access to virtually any type of data, at any scale.
VAST DataBase
The Database allows them to index data, store context as vector embeddings and search across that data in real-time.
VAST DataEngine
The Data Engine provides the distributed run time for agents, with built in fault tolerance to make them resilient. It also provides the capability to execute functions and the messaging infrastructure for agents to communicate with each other and the outside world.
VAST InsightEngine
The Insight Engine provides a real-time embedding service.
VAST AgentEngine
The AgentEngine provides an intuitive interface to build and deploy agents at scale.
AI Agents
Software systems that leverage data to interact with users, other systems, or their surroundings.
Security & Observability
The security to safely deploy agents with the right controls and the observability needed to get visibility to what they are doing, is inherently built into the platform. With agents making decisions in complex, evolving environments, clear transparency into their reasoning paths is critical. The VAST AI OS includes a Explainability & Transparency layer that allows you to track tool usage, memory states, and trace execution plans.
VAST DataSpace
The DataSpace allows you to seamlessly extend agent deployments from the edge, to the data center to the cloud.
DisAggregated, Shared-Everthing (DASE)
This is all powered by the VAST’s DASE architecture.
Why is this so powerful?
Legacy architectures have fundamental limitations, preventing them from providing the transactional performance, at scale, agents need to act in real-time.
DASE was built to provide the parallelism needed to address these challenges, while also bringing new levels of resilience, efficiency and simplicity.
That transactional performance is key for scaling agent embeddings, where other vector databases struggle with real-time updates and limit deployments to billions of embeddings, VAST’s vector database supports real-time updates, real-time search and scales to trillions of embeddings, allowing you to unleash the full potential of AI across all your data.
The VAST AIOS give you everything you need to ingest structured, unstructured and event data, embed and add context to that data in real-time and connect agents with the tools they need to act on this data.
The ultimate result is :
One compute platform that supports all your AI workloads, enabling the continuous loop of inference and fine tuning
One system that enables any deployment option, from edge and embedded systems, to the data center to the cloud
One unified data environment for any type of data, structured or unstructured
One tier, giving real-time access to all data
Why This Matters Now
Microsoft’s elevation of AI and its unification of Copilot and LinkedIn leadership signals that every major software platform is becoming an AI platform. Agents will replace dashboards, copilots will replace clicks.
But agents are only as powerful as the context they can access.
That context lives in your data.
VAST’s AIOS is the foundation for context rich, high speed, secure data access at scale, powering the next generation of agentic computing across enterprises, scientific research, life sciences, and more.