DEV Community

Searchless
Searchless

Posted on • Originally published at searchless.ai

Microsoft Project Solara: The OS Where AI Agents Replace Apps — and What It Means for Brand Discovery

#ai #seo

Originally published on The Searchless Journal

Microsoft just announced something that makes the search-to-AI shift feel small by comparison. At Build 2026, the company unveiled Project Solara: a new operating system, built from the ground up, for devices that run AI agents instead of apps.

Not an app store. Not a browser. Not a search engine. An agent shell, sitting on hardware designed around a single premise: you speak, and agents act.

The platform is built on Android (technically, the Microsoft Device Ecosystem Platform, an enterprise-grade OS based on AOSP). It targets what Microsoft calls "agent-first devices" — a category that includes desk displays with facial recognition, wearable badges that record and transcribe conversations, and future form factors that haven't been invented yet. Target, Best Buy, CVS Healthcare, and AccuWeather are already planning pilots. Steven Bathiche, the Microsoft fellow leading the project, describes it as a "chip-to-cloud platform" where the operating system is "liminal, transcending the device and the cloud."

This is agentic commerce at the hardware layer. And for brands, the implication is stark: if AI agents are the new interface between consumers and products, then optimizing for search engines is necessary but no longer sufficient. You need to optimize for agents — and those agents are about to live on dedicated hardware that doesn't run Chrome, doesn't have an app store, and doesn't display blue links.

What Project Solara actually is

Project Solara is not a consumer product. It's a platform — a combination of hardware reference designs, an operating system, cloud infrastructure via Azure, and an agent framework that Microsoft wants hardware partners to adopt.

The core idea is simple and radical: the cost of creating specialized computing devices has historically been prohibitive because each new form factor required rebuilding the entire stack — hardware, OS, UI framework, app ecosystem, developer tools, security model. Agents collapse that cost. When an AI agent can generate its own interface on the fly, adapting to any screen size or input modality, you don't need a custom app for every device. Microsoft calls this "just-in-time UI."

The platform has three pillars, according to Microsoft's announcement:

  1. Enterprise readiness — privacy, security, and manageability through Microsoft Intune, Entra ID, and Hello for Business biometric authentication
  2. Agent-driven interaction with just-in-time UI — agents dynamically generate interfaces instead of loading static apps
  3. Extensibility — organizations can bring their own agents, not just use Microsoft's

The operating system layer is MDEP (Microsoft Device Ecosystem Platform), built on the Android Open Source Project. It's not Windows. It's not full Android. It's a stripped-down, enterprise-hardened version of Android designed for devices that don't need apps — they need agents.

Microsoft is previewing two reference design categories: stationary (think desk displays, Echo Show-like form factors with facial recognition) and portable (wearable badges with cameras and fingerprint scanners). These are concepts, not shipping products. Microsoft wants hardware partners — Samsung, Lenovo, HP, and others — to build the actual devices.

Why this matters more than another chatbot update

The search industry has spent the last two years processing a single question: what happens when AI replaces Google? That question matters, but it's too narrow.

Project Solara reframes the problem entirely. The question isn't "what happens when AI replaces search?" It's "what happens when AI agents become the primary interface between consumers and every service, product, and brand?"

Here's the difference. When a user searches Google for "best running shoes," brands can still compete. They optimize for keywords, buy ads, build content, earn backlinks. Even in AI Overviews and ChatGPT, the model synthesizes answers from the web. Brands have a shot at being cited.

But on an agent-first device, the interaction is different. A user doesn't search. They speak: "I need new running shoes for a marathon." The agent doesn't show a list of links or even a synthesized summary. It evaluates options, checks inventory, compares prices, reads reviews, and may complete the purchase — all in a single conversational flow. The user never sees a search result. They never visit a website. The agent is the entire interface.

This is what makes Project Solara different from the Copilot super app that Microsoft announced in late May. Copilot is a software consolidation story — unifying Microsoft's AI tools into one interface that lives inside existing platforms (Windows, mobile apps, browsers). Project Solara is a hardware story. It's about putting agents on dedicated devices that exist specifically to run agents. No legacy app frameworks. No browser. No search box. Just an agent shell that dynamically loads whatever AI the user needs.

The May 30 Copilot coverage focused on software consolidation and what it means for AI discovery inside existing surfaces. This is a different conversation: what happens when an entirely new category of hardware surfaces emerges, and agents are the only way to reach consumers on it?

The Target, Best Buy, CVS signal

The most important detail in Microsoft's announcement isn't the technology. It's the pilot partners.

Target, Best Buy, CVS Healthcare, and AccuWeather are planning to build agent-first experiences for Solara devices. These are not AI startups or enterprise SaaS companies. They're retailers and consumer service companies — the exact businesses that have spent decades optimizing for Google SERPs, Google Shopping, Amazon listings, and mobile apps.

Their involvement signals something specific: major retailers believe that agent-first hardware is a real enough surface to invest in now. Not in five years. Now.

When Target builds an agent experience for a Solara desk device, what does that look like? The user walks into their home office. The device recognizes their face via Hello for Business. They say, "I need to pick up a few things from Target." The Target agent — not the Target app, not the Target website — appears in the agent shell. It knows their past purchases. It suggests items. It checks inventory at the nearest store. It adds items to a cart. It confirms the order.

The user never opened an app. Never visited a website. Never saw a search result. The agent handled the entire discovery-to-transaction flow.

Now extrapolate. If Target has an agent on Solara, does Nike? Does a local coffee shop? Does a DTC brand that only sells through its website? The brands that show up first in an agent's recommendations aren't the ones with the best SEO. They're the ones with the best agent optimization — structured data, knowledge graph presence, entity authority, and direct agent API integrations.

The discovery gap: from SERPs to agent shells

There's a pattern in how discovery surfaces evolve, and Project Solara accelerates it.

In Web 1.0, discovery happened through portals — Yahoo, AOL, MSN. Brands bought banner ads and hoped for clicks. In Web 2.0, discovery moved to search engines. Brands optimized for Google's algorithm. In the app era, discovery moved to app stores. Brands built mobile apps and bought install ads. Now, in the agent era, discovery is moving to agent shells — AI interfaces that sit between consumers and every product, service, and piece of information.

Each transition made the previous surface less important but didn't kill it. Google didn't kill portals. Apps didn't kill Google. Agents won't kill apps. But each transition did shift where the growth was — where the new users, new revenue, and new attention went.

Project Solara represents the hardware instantiation of that shift. The agent shell is the new portal, the new SERP, the new app store — all at once. And brands that only optimize for the previous surface are invisible on the new one.

The technical implications are concrete. On a Solara device:

  • There is no browser. Traditional web SEO — title tags, meta descriptions, backlinks — doesn't apply because there's no page to render.
  • There is no app store. App store optimization is irrelevant because agents are loaded dynamically from the cloud.
  • There is no search box. Keyword targeting doesn't work because users speak in natural language, not query fragments.
  • The agent decides what to show. Discovery is mediated entirely by the agent's knowledge base, its training data, its real-time API access, and its reasoning about what's relevant to the user.

This is the most extreme version of zero-click discovery yet. Not just "the answer is in the SERP" but "the answer is on a device that doesn't even have a SERP."

What agent optimization looks like

If search engine optimization is about making your content rank, and app store optimization is about making your app install-worthy, then agent optimization is about making your brand agent-readable and agent-recommendable.

Based on what Microsoft has described about Solara's architecture, here's what agent optimization entails:

Structured entity data. Agents don't crawl web pages the way Googlebot does. They query knowledge graphs and structured data APIs. If your brand's information — products, services, locations, pricing, availability — isn't structured in machine-readable formats (schema.org, JSON-LD, knowledge graph entries), agents can't discover it. This is the foundation.

Knowledge graph presence. Microsoft's platform uses Azure for state management and agent orchestration. Agents will likely pull from Wikidata, Wikipedia, Google's Knowledge Graph, and proprietary Microsoft data sources. If your brand isn't a recognized entity in these graphs, it doesn't exist in the agent's world.

Agent API endpoints. The most visible brands on Solara won't be the ones with the best content. They'll be the ones with the best APIs. When Target builds a Solara agent experience, it's likely exposing product data, inventory, and transaction capabilities through an API that Microsoft's agent shell can call. Brands that expose clean, well-documented APIs for their products and services will be first-class citizens in agent ecosystems.

Conversational content architecture. Agents answer questions. Brands need content structured around questions, not keywords. FAQ sections, how-to guides, product comparison data, and pricing information — all structured in a way that an agent can extract, synthesize, and present to a user in a conversation.

Multi-agent compatibility. Microsoft explicitly designed Solara for a "multiple-agent world." Users won't only interact with Microsoft agents. They'll use agents from different providers, each with its own training data, citation patterns, and recommendation logic. Brands need to optimize for multiple agent ecosystems simultaneously — just like they learned to optimize for both Google and Bing, except the gap between agents is much wider than the gap between search engines.

The competitive landscape: not just Microsoft

Project Solara doesn't exist in a vacuum. The agent-first hardware category is emerging simultaneously from multiple directions.

Google is building AI features into every Android device — the June 2026 Android drop included digital wardrobes, AI try-on, and Circle to Search for shopping. These aren't Solara-scale (they're features on existing phones, not a new OS), but they point in the same direction: AI-mediated discovery replacing traditional search on consumer devices.

Meta is investing in AI-powered wearables — smart glasses with AI assistants that can see what you see, answer questions, and make recommendations. The Ray-Ban Meta glasses already have AI features; the next generation will have more.

OpenAI and Jony Ive are reportedly building a hardware device together. Details are scarce, but the direction is clear: a dedicated AI device, designed from scratch, with a conversational interface.

Apple's WWDC starts June 8, and the Siri redesign is expected to transform it from a voice assistant into an agent platform. If Apple puts a capable agent on every iPhone, that's 1.5 billion devices where agent-first discovery becomes the default.

The pattern is unmistakable. Every major tech company is converging on the same thesis: agents are the next interface, and they need hardware surfaces optimized for agent interaction. Microsoft is just the first to articulate it as a platform play with a specific OS, reference designs, and enterprise tooling.

Why brands should pay attention now

Project Solara is a preview, not a product. The reference designs are concepts. The pilot partners are early. The hardware partners haven't been named. Microsoft itself says "we are still early" and "I don't want to over-promise."

But the trajectory is clear enough to act on. Here's why brands should start thinking about agent optimization today:

First, the infrastructure decisions being made now will determine which brands are visible on agent-first devices for years. When Target builds its Solara agent, it's establishing the data architecture, API endpoints, and knowledge graph connections that will determine how the Target agent surfaces products and completes transactions. Early movers set the template.

Second, agent optimization is cumulative. Structured data, knowledge graph entries, and API endpoints aren't quick wins — they're infrastructure investments that take time to build correctly. Brands that start now will have a significant head start when agent-first devices reach consumer scale.

Third, the skills overlap with existing GEO work. The same structured data that helps ChatGPT and Perplexity cite your content also helps agent shells discover your products. The same knowledge graph presence that improves AI Overviews visibility also makes your brand a recognizable entity in Microsoft's Azure-backed agent ecosystem. Agent optimization isn't a separate discipline from GEO — it's GEO extended to hardware surfaces.

The brands that treat Project Solara as a curiosity — "Microsoft always has experimental hardware projects that go nowhere" — are making the same mistake that brands made when they dismissed mobile apps in 2008. The platform might not win. But the category will.

The new discovery stack

If you map the full discovery landscape in mid-2026, it looks like this:

Traditional search (Google, Bing) still drives the majority of intent-based discovery. SEO still matters. But growth is flat or declining across most verticals as AI Overviews and zero-click answers erode click-through rates.

AI search (ChatGPT, Perplexity, Gemini, Claude) is the fastest-growing discovery surface. Brands that optimize for AI citations — structured data, answer-first content, entity authority — are gaining visibility on surfaces that didn't exist two years ago.

Agentic software (Copilot, Siri, Google Assistant, Amazon Alexa) is becoming a transaction surface. Users don't just ask questions — they issue instructions. "Order more paper towels." "Book a table for two tonight." "Find me a gift for my mom." The agent handles discovery, evaluation, and transaction in one flow.

And now, agentic hardware (Project Solara, Meta glasses, the rumored OpenAI/Jony Ive device) is adding a dedicated physical layer. These devices don't have browsers or app stores. They only have agents. The discovery surface is entirely mediated by AI.

Each layer requires a different optimization approach, but they share a common foundation: structured data, knowledge graph presence, and machine-readable content. Brands that invest in that foundation can extend to every new surface as it emerges. Brands that don't will keep optimizing for a shrinking surface.

The shift from apps to agents is not a metaphor. Microsoft is building the operating system for it. Google is embedding it into Android. Apple is about to transform Siri into an agent platform. Meta is putting agents on your face.

The brands that will thrive in this landscape are not the ones with the best search rankings. They're the ones that make themselves agent-readable, agent-recommendable, and agent-transactable. That means structured data, knowledge graph authority, clean APIs, and content architectures built for conversations, not keywords.

Project Solara is a preview. But the direction is unmistakable: agents are becoming the interface, and hardware is being built to serve them. The question for brands isn't whether to optimize for agents. It's whether to start before or after the first Solara devices ship.

Is your brand visible to AI agents — or only to search engines? Run a free AI visibility audit and find out where you stand across ChatGPT, Perplexity, Gemini, and the emerging agent surfaces.

Sources

Find out what GEO services from Searchless can do for your brand's visibility across AI engines, agent platforms, and the new discovery surfaces.

Top comments (0)