Microsoft’s Project Solara: How Agentic Devices Are Rewriting the Rules of Work

Project Solara is Microsoft’s boldest bet yet on agentic computing: a platform for low-power, purpose-built devices that run AI agents instead of traditional apps. Built on an enterprise Android layer (MDEP) with chips-to-cloud integration and Azure services, Solara aims to move AI from screens into everyday tools—and into the flow of work.

TL;DR

Project Solara replaces app-centric devices with agent-first hardware—think wearable badges and desk hubs—that invoke AI to handle tasks across identity, security, and enterprise data. Powered by the Microsoft Device Ecosystem Platform (MDEP) and Azure AI, Solara supports multi-agent workflows, over-the-air management, and strong privacy-by-design, promising week-long battery life for frontline form factors and streamlined productivity for knowledge work.

What is Microsoft’s Project Solara, in plain terms?

Project Solara is a platform for devices that use AI agents—not apps—as the primary interface. It’s built on an enterprise version of Android (MDEP), integrates with Azure, and supports devices from desk hubs to wearable badges. The aim is to embed agentic architecture into physical tools that work faster, cost less, and fit more contexts than PCs and phones.

Rather than tapping and swiping across apps, Solara devices accept voice, touch, and vision inputs, then trigger agents to read data, take actions, and coordinate with enterprise systems. The result is an intent-first experience: ask, confirm, and complete. Microsoft is piloting Solara with major retailers, healthcare providers, and knowledge-work teams to validate value in inventory, clinical documentation, customer service, and office coordination.

How does Solara work across chips, devices, and cloud?

Solara spans silicon to cloud: low-power hardware runs MDEP (an enterprise Android), while AI workloads are orchestrated via Azure. Identity, policy, and telemetry flow through Intune device management, Entra ID, and Defender for device security, enabling secure, over-the-air operations at enterprise scale.

This stack is designed for speed and resilience. A wearable badge can capture audio for clinical notes, scan a QR for inventory, or summarize a client visit; the desk hub can invoke agents by voice, sign you in with facial recognition, and even hand off a full Windows session via Windows 365 cloud streaming. OEMs can use off-the-shelf components, and Microsoft’s team has demonstrated rapid porting—moving a badge to the platform within days—showing the ecosystem’s agility.

Why MDEP (enterprise Android) is the OS layer Solara needs

The Microsoft Device Ecosystem Platform (MDEP) is a hardened, enterprise-optimized Android layer that favors low power, small footprints, and secure, policy-driven management. It bundles OTA updates, privacy controls, and identity, making it a pragmatic base for AI-first devices that must operate for a week on a single charge.

Because MDEP aligns with familiar enterprise controls, IT can provision devices with Entra ID, enforce policies in Intune, and monitor with Defender from day one. This dramatically reduces the time between pilot and production—and it standardizes a new class of endpoints, from badges and scanners to hubs and earbuds.

Prototype devices at a glance

Microsoft’s reference designs illustrate how agent-first devices can slot into daily work—without replacing PCs or phones where they still shine.

These are not consumer gadgets. They’re role-defined tools that compress multi-step app juggling into quick, reliable agent prompts—making real-world work faster and less error-prone.

What makes Solara different from past “single-assistant” plays?

Instead of one assistant for everything, Solara supports multiple, organization-approved agents that are permissioned and governed by IT. That means a clinical documentation agent, a knowledge-base agent, and a service-desk agent can coordinate—each with distinct scopes—under enterprise policy.

This multi-agent model builds on Copilot extensions and orchestration principles to minimize hallucinations, respect data boundaries, and route requests to the right skills. It’s intent-first and role-specific, with UIs generated as needed by an Agent Shell, not pieced together from siloed apps.

Privacy and security by design—especially for sensor-rich wearables

Solara’s hardware and OS design adopt privacy-by-default: physical shutters and indicators, consent-aware capture, and policy gates that govern what devices can see, hear, or store. Combined with Defender, Entra ID, and Intune, enterprises get unified auditing and response.

Organizations can enforce retention limits, disable sensors per role, and require explicit user acknowledgement for sensitive tasks. For regulated environments, a privacy-by-design playbook for AI wearables helps teams formalize consent and documentation flows before devices ever hit the floor.

How Solara changes work in healthcare, retail, and the office

Solara removes the friction of “find the right app, then the right field.” A clinician can capture a conversation with consent, have it summarized, coded, and filed; a retail associate can scan, locate, and restock without leaving the aisle; an executive can request a brief, join a meeting, and action follow-ups from a single voice prompt.

Early pilots span healthcare documentation, inventory and planogram checks, customer service triage, weather and field reporting, and knowledge-worker desk orchestration. Because agents are tuned to roles—not job titles—organizations can standardize repeatable flows while adapting to local context. For many teams, this is the first credible path to hands-free productivity at scale.

For developers and IT: from apps to intents

Developers don’t ship full-screen apps; they publish skills, tool use, and compact UI fragments that the Agent Shell assembles on demand. A lightweight, intent-driven UI cuts build time and keeps experiences consistent across form factors. IT inherits a familiar model: enroll devices, assign policies, verify compliance, and monitor posture.

Teams can integrate enterprise data via Azure AI at the edge and in the cloud, bring existing Copilot skills, and compose agents for frontline or knowledge roles. The bar for success is operational reliability: predictable invocation, quick disambiguation, and graceful failure when the agent can’t complete the task.

A simple way to pilot Project Solara

• Define two high-value workflows where “hands-busy, eyes-busy” is common (e.g., bedside note capture; aisle restock).
• Map permissions, data sources, and privacy constraints, then set policy baselines in Intune.
• Build or adapt agents as skills and Copilot extensions, targeting MDEP devices.
• Run a two-week sprint with success metrics (time saved, error rate, satisfaction), capturing failure cases for iteration.
• Harden for production: OTA updates, observability, Defender alerts, and Entra ID access reviews.