Editorial projections, not confirmed plans — where each provider appears headed based on the public record, and the events most likely to move their grades. We re-grade against the today scorecard as these land.
Trajectories
Converting the most complete set of assets — models, silicon, OS, data — into a coherent agent OS, if product execution catches up to the hand it holds.
NowBest raw hand, coherence the question
Frontier Gemini, TPUs, Android, Workspace, and Maps under one roof.
Model StrategyPersonal ContextAgent Platform Potential
H2 2026Agents wired into Android & Workspace
Mariner/Astra-class agency moving from preview into the OS and productivity surfaces would lift cross-app-actions and agent capability.
Agent CapabilityCross-App Actions
Defending and deepening the enterprise control plane, while reducing OpenAI dependence with in-house models.
NowEnterprise leader, hedged models
Copilot everywhere and Azure as the default AI cloud; model strategy still split.
Enterprise ReadinessCross-App ActionsModel Strategy
H2 2026In-house models step up
Maturing MAI/Phi models reduce the OpenAI dependency and lift the model-strategy score.
Model StrategyAgent Platform Potential
From the default consumer assistant toward a full agent platform — the open flank is the device edge and a hardware surface of its own.
NowCapability + distribution leader, no edge
Frontier models, the best-known API, and a maturing agent mode — but no device story.
Assistant IntelligenceDeveloper APIsAgent Capability
H2 2026Agent platform + a hardware bet
Deeper agent infrastructure and a rumored dedicated device would attack the on-device gap that caps its scorecard.
Agent Platform PotentialOn-Device AI
Becoming the connective substrate of the agent era through MCP and best-in-class agentic models, while consumer surface and personal context remain deliberate gaps.
NowThe agent-builder's platform
Claude Code, computer use, and MCP set the standard for serious agentic work.
Agent CapabilityDeveloper APIsThird-Party Integrations
H2 2026MCP as default plumbing
As MCP adoption compounds across vendors, Anthropic's agent-platform and integration scores rise without it owning a consumer surface.
Agent Platform PotentialMyAGI Alignment
Pushing ambient, on-device AI through glasses and open weights — reach without (yet) a platform.
NowReach + open weights
Meta AI across billions of users and the open-weights ecosystem the field builds on.
Third-Party IntegrationsOn-Device AIModel Strategy
Connect 2026Glasses as the edge agent surface
Next Llama plus AR glasses could make Meta the ambient on-device leader.
On-Device AIConversational UX
From the most personal context with the least capable assistant toward an App-Intents-powered agent layer — if it can source frontier reasoning and ship a rebuilt Siri.
NowWidest asset-to-capability gap on the board
On-device leadership and unmatched personal context, paired with a bottom-tier assistant and almost no agency.
On-Device AIPersonal ContextAssistant Intelligence
WWDC 2026The catch-up keynote
A rebuilt Siri, an expanded App Intents action layer, and a clarified model strategy are the swing factors for half of Apple's dimensions.
Assistant IntelligenceCross-App ActionsModel StrategyDeveloper APIs
H2 2026App Intents as the cross-app substrate
If developers adopt an upgraded App Intents, Apple's cross-app-actions and agent-platform scores could move faster than its raw model intelligence.
Cross-App ActionsAgent Platform Potential
Racing up the capability curve with real-time data leverage; platform, enterprise, and edge remain unbuilt.
NowFast model, thin platform
Grok closed the capability gap; integrations, enterprise, and edge are minimal.
Assistant IntelligenceModel Strategy
H2 2026X as a distribution + data moat
Tighter X integration could turn real-time data into a distinctive agent surface — if a developer platform follows.
Agent Platform PotentialDeveloper APIs
