Microsoft has offered an early look at Project Solara, a new chip-to-cloud platform it says is designed to support a class of agent-first devices. The company describes the effort as a foundation for hardware and software built around AI agents rather than traditional apps and screens.
The preview comes from Microsoft’s Applied Sciences Group and centers on the idea that agents are becoming both a programming model and a primary way people interact with computers. In Microsoft’s framing, that shift could make it easier to build highly specialized devices for specific tasks, roles and environments without having to recreate the full technology stack each time.
Microsoft argues that computing has steadily moved closer to users and to the work they need to do, from mainframes to PCs, phones and wearables. Project Solara is meant to extend that evolution by creating devices where users invoke intelligence directly, rather than opening a sequence of apps and menus.
The company says agents can operate at different layers. In some cases they sit alongside existing apps. In others, they become part of the application itself. A third model places them outside individual apps entirely, allowing them to coordinate across services, devices and workflows while maintaining context over time. Microsoft says this broader model creates an opening for new device types.
Project Solara is intended to support that approach with a lightweight experience at the edge and cloud-backed state through Azure. Microsoft says the system is meant to function as a window into longer-running intelligent workflows, with the device acting as an interface between a person and a more distributed AI environment.
Microsoft is positioning the platform for enterprise use, emphasizing security, privacy, manageability and user control. The company says those protections are foundational rather than optional, given that agentic devices may handle work context and connect across data, identities and organizational boundaries.
The platform is also designed for an open ecosystem, according to Microsoft. Organizations would be able to use Microsoft agents where they are useful, while also bringing in agents they build themselves or source elsewhere for specialized needs. Microsoft says Project Solara must be able to bring those agents together in a coherent way.
The company says it is also working on just-in-time user interfaces, meaning the experience can adapt across devices and input modes without requiring developers to rebuild for each new form factor. Early versions rely on more structured UI elements, with more dynamic and generative interfaces expected over time.
Microsoft says it is previewing concepts in two broad categories, stationary and portable. Both are described as multimodal, with support for glanceable information, voice and vision, and the ability to reach the right agent at the right moment.
The company says it is exploring use cases in industries including healthcare, retail and financial services. It also says Project Solara is meant to support reference designs that device makers can adapt and customize.
Microsoft cautioned that the effort is still early. Even so, it said lower costs for building specialized devices could speed innovation and expand the range of computers designed for specific contexts.
The preview signals how Microsoft is thinking about the next platform shift. In its view, the move is from apps to agents, and from generic computers to devices built around intelligence delivered where users need it most.