Microsoft used Build 2026 to introduce Project Solara, a chip-to-cloud platform for devices built around AI agents rather than traditional apps.
The announcement is still early and deliberately framed around concept hardware, not finished consumer products. But the direction is specific: Microsoft wants agents to move out of the laptop, phone, and browser, and into purpose-built devices for workplaces such as stores, clinics, offices, and field environments.
Project Solara starts with two reference concepts. One is a wearable badge designed for portable, on-the-go use. The other is a desk hub for stationary work. Microsoft says early pilots will involve companies including Best Buy, CVS Health, Levi’s, Target, and AccuWeather. Qualcomm and MediaTek are the first silicon partners.
What Microsoft Is Actually Building
Project Solara is best understood as a platform stack, not a single gadget. Microsoft describes it as “chip-to-cloud,” meaning the device, edge operating system, cloud state, identity, management, and agent interface are meant to work as one system.
The edge operating system is Microsoft Device Ecosystem Platform, or MDEP, an enterprise-grade OS built on Android Open Source Project. On top of that sits an Agent Shell that can load and tailor multiple cloud-based agents. The enterprise pieces are familiar Microsoft territory: Intune for device management, Entra ID for identity, Hello for Business for biometric authentication, and privacy controls such as physical mic mute buttons and recording indicators.
That matters because agent-first devices are not just small screens with chatbots attached. If they are worn by staff, placed on desks, or used around customers and patients, they need to answer basic questions before anyone gets excited about the interface: Who is signed in? What data can the agent reach? Is the device listening? Can IT disable it, patch it, or audit it?
Microsoft is clearly aiming Project Solara at those questions first, which is why the initial pitch is enterprise-heavy rather than consumer-friendly.
The Reference Devices Show The Use Case
The badge concept is a lightweight wearable with a touchscreen, fingerprint authentication, far-field microphone array, speaker, side-facing camera, WiFi, Bluetooth, GNSS, 5G connectivity, and Qualcomm wearable silicon. Microsoft describes it as a form factor for information workers, nurses, front-line workers, and others who already carry badges.
The desk concept has a touchscreen, face authentication, privacy lock buttons, microphone mute and volume controls, dual far-field microphones, a speaker, a UWB presence sensor, USB-C ports, WiFi, Bluetooth, and MediaTek IoT silicon. Microsoft says it can work on its own, act as a companion to a Windows PC, or become a Windows 365 client when connected to an external display.
Those details make the strategy easier to read. Microsoft is not trying to replace the PC with a badge. It is testing whether a smaller device can become the most convenient doorway into a set of agents at a particular moment.
A store manager, for example, might not want to open a laptop between customer conversations. A badge-style device could provide a quick view of urgent items from a priority agent, then let the manager capture an in-person discussion through a facilitator-style agent, with follow-up actions handled later in Microsoft 365. That scenario only works if authentication, recording permissions, audio quality, and enterprise data boundaries are handled cleanly. Otherwise it becomes another workplace device people avoid.
Why “Just-In-Time UI” Is The Real Bet
The most important technical idea in Project Solara is not the badge or the desk hub. It is Microsoft’s argument that agents can reduce the cost of building specialized computers.
Historically, new device categories have struggled because they require more than hardware. They need app frameworks, interface conventions, developer tools, management systems, security models, and a reason for software makers to support yet another screen size and input method.
Project Solara tries to loosen that dependency with what Microsoft calls just-in-time UI. The idea is that an agent experience can adapt its presentation to the device, task, and modality, whether the user is speaking, tapping, glancing, or interacting across multiple modes.
Microsoft is not claiming fully generative interfaces are ready to replace structured UI. Its current description sits in the middle: more flexible than standard responsive design, but not dependent on unconstrained AI-generated screens. That is a sober distinction. Enterprise users need predictable controls, not surprise layouts during a clinical handoff or store operation.
If the approach works, the same agent could appear differently on a desk hub, wearable badge, or future vertical device without every developer redesigning the experience from scratch. That would make niche hardware less risky for device makers and more plausible for businesses with specific workflows.
What Changes For Developers And Businesses
For developers, Project Solara extends the meaning of “agent distribution.” An agent built for Microsoft 365, Copilot Studio, the Microsoft 365 Agents SDK, or Microsoft Agent Framework may eventually need to show up outside a conventional app surface.
For businesses, the question is more practical: which workflows are valuable enough to deserve their own agent-first access point?
- Healthcare: Microsoft says Dragon Copilot is exploring agent-first experiences for physicians and nurses, including capturing interactions, surfacing relevant context, and following through on tasks.
- Retail: Pilot partners such as Best Buy, CVS Health, Levi’s, and Target suggest Microsoft sees front-line work as an early proving ground.
- Office work: The desk concept points to daily briefings, priority cards, voice access to Microsoft 365 Copilot, and task handoff with a Windows PC.
- Developer work: GitHub Copilot is exploring ways to keep developers closer to the progress of coding projects through voice and other modalities.
The common thread is not novelty hardware. It is reducing friction in workflows where opening a full app is too slow, too awkward, or too disconnected from the physical environment.
The Risks Are As Clear As The Opportunity
Project Solara also raises hard adoption questions. Wearable and desk-based agent devices will need strong social cues around recording, listening, and camera use. Microsoft mentions physical mute controls and clear indicators, which are essential. They may not be sufficient in every workplace.
There is also the problem of agent quality. A device built around agents is only as useful as the agents it can summon. If they miss context, surface noisy priorities, or require too much correction, the form factor will not save the experience.
Then there is ecosystem timing. Microsoft says the concept devices will inform reference designs and turnkey solutions, with pilots beginning in the coming months. That means the next stage is not a broad launch. It is learning whether real businesses want these devices after the demo glow fades.
What To Watch Next
The most important signal will be what pilot customers actually deploy. A badge used by a narrow group of workers for one high-value workflow would be more meaningful than a vague general assistant device. The same goes for the desk hub: its value depends on whether it becomes a faster working surface or simply another notification endpoint.
Watch also for OEM participation beyond Qualcomm and MediaTek reference work, because Project Solara needs device makers if it is to become a category rather than a Microsoft experiment. Developer support will matter too, especially whether third-party agents can reach these devices without heavy custom work.
Project Solara is early, but it is a useful marker for where Microsoft thinks enterprise AI is headed. The company is no longer treating agents only as features inside existing apps. It is preparing for a world where agents become the reason a device exists in the first place.