Who is building the agent operating system?
Twelve dimensions across seven providers. Re-graded as announcements land, with the market's reaction shown alongside.
Anthropic
#4 · 7.0/10 overallThe agent-builder's platform, and the author of MCP.
The reference platform for serious agentic work: Claude Code, computer use, and the Model Context Protocol that's becoming the industry's cross-app standard. Thin on consumer surface, personal context, and the device edge.
- Assistant
- Claude
- Flagship model
- Claude Opus / Sonnet 4.x
- Public ticker
- Private (backers include Amazon: AMZN, Alphabet: GOOGL)
Claude is frontier-class, with a particular edge in coding and careful reasoning.
Claude Code and computer use are the most credible shipped agentic systems today.
Excellent text chat; no first-party voice or consumer-grade multimodal interaction layer.
MCP turns Claude into a cross-app actor and is becoming the industry's connection standard.
Projects add some memory, but there's no deep, persistent personal-context substrate.
MCP is the breakout: an open protocol the whole industry is adopting for tool/app connectivity.
The developer's favorite API: clean tool use, strong agents, available via AWS and GCP.
Owns its frontier models; relies on partner clouds for compute rather than its own silicon.
Authoring MCP positions Anthropic as the connective substrate of the agent ecosystem.
No device or local-inference presence at all.
Claude for Work plus Bedrock/Vertex availability make it a serious enterprise option.
Earned-capability ethos and MCP align tightly; missing personal context and a revocable runtime as product.
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.
- 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.
- NewsAnthropic says Claude can now use your computer to finish tasks for youCNBC · 2026-03-24
Anthropic expands Claude's computer-use as part of a push toward agents that operate screens and apps for you.
- AnalysisAgents Over BubblesStratechery · 2026-03-16
The AI value chain shifts to integrated model-plus-harness systems whose orchestration creates defensible moats.
- AnalysisA Guide to Which AI to Use in the Agentic EraOne Useful Thing (Ethan Mollick) · 2026-02-18
A models/apps/harnesses framework: the harness that lets a model take real actions now matters as much as IQ.
- MarketLeaders, gainers and unexpected winners in the Enterprise AI arms raceAndreessen Horowitz (a16z) · 2026-01-30
An enterprise survey: OpenAI still leads adoption but Anthropic and Google gain fast, a multi-vendor oligopoly.
- AnalysisAnthropic's Claude Code is having its 'ChatGPT' momentUncoverAlpha · 2026-01-26
Claude Code's terminal-native agent design drove rapid enterprise adoption and measurable productivity gains.
- NewsOpenAI, Anthropic, and Block join Linux Foundation effort to standardize the AI agent eraTechCrunch · 2025-12-09
Anthropic donates MCP to the new Agentic AI Foundation, so MCP shifts from vendor project to neutral infrastructure.
- PrimaryDonating the Model Context Protocol and establishing the Agentic AI FoundationAnthropic · 2025-12-09
Anthropic hands MCP to a Linux Foundation fund to keep it open amid 97M+ monthly SDK downloads.
- PrimaryOne Year of MCP: November 2025 Spec ReleaseModel Context Protocol · 2025-11-25
MCP's rise from open-source experiment to de facto standard adopted across OpenAI, Microsoft, GitHub, and Google.
- PrimaryCode execution with MCP: Building more efficient agentsAnthropic Engineering · 2025-11-04
Presenting MCP servers as code APIs cuts agent token usage by up to 98.7% by loading only needed tools.
- AnalysisAI Engineering Goes MainstreamLatent Space · 2025-06-13
AI engineering matures into a vertical; agents, reasoning, and orchestration tooling are the practical frontier.
- AnalysisClaude 4, Anthropic Agents, Human-AI AgentsStratechery · 2025-05-27
Reads Claude 4 as evidence Anthropic is betting on scaffolded agents while leaving the consumer chatbot opportunity open.
- AnalysisAgents are models using tools in a loopSimon Willison · 2025-05-22
A crisp working definition of an agent that clarifies the core mechanism behind the agentic-platform shift.
- AnalysisThe Agentic Era: Orchestration and Distribution, Not Model Power, Define AI MoatsDecoding Discontinuity · 2025-05-02
As frontier models converge, advantage shifts to orchestration quality and distribution reach over raw benchmarks.
- AnalysisModel Context Protocol has prompt injection security problemsSimon Willison · 2025-04-09
How MCP's open tool model enables prompt-injection and 'rug pull' mutations: the security tax of the standard.
- AnalysisWhy MCP WonLatent Space · 2025-03-10
MCP became the de facto agent-integration standard via AI-native design, Anthropic's developer brand, and first-party tooling.
- PrimaryIntroducing computer use, a new Claude 3.5 Sonnet, and Claude 3.5 HaikuAnthropic · 2024-10-22
The first frontier model to navigate a screen by moving a cursor, clicking, and typing.
- NewsAnthropic says 80% of its new production code is now authored by ClaudeVentureBeat
Anthropic claims Claude now authors most of its own production code, and what agentic coding means for enterprises.
- PrimaryAnthropic raises $30 billion in Series G at $380 billion post-money valuationAnthropic
The capital base underpinning Anthropic's enterprise and agent-platform ambitions.