WED, 03 JUN 2026 · 18:35:45 UTC

Anysphere (Cursor)

FlagshipProduct

USA·HQ San Francisco·Est. 2022

The AI-first code editor — Cursor.

8.0

our score

Our take

Cursor is the AI-first IDE and autonomous coding agent platform that has become the default editor for thousands of engineering teams.

At a glance

Best known for
Cursor, the AI-first VS Code fork and autonomous coding agent
Biggest strength
Seamless autonomy slider from Tab completion to fully autonomous cloud agents
Biggest risk
Microsoft could subsume Cursor's innovations into GitHub Copilot/VS Code
Stage
Series B ($2.5B valuation)
Primary revenue
Subscriptions to Pro/Enterprise plans for AI coding assistance and cloud agents

What they do

Anysphere builds Cursor, an AI-first integrated development environment forked from Microsoft’s VS Code. Unlike traditional editors that rely on bolted-on AI extensions, Cursor is architected around a model-native user experience centered on an "autonomy slider"—ranging from lightning-fast Tab completion powered by a specialized model, to targeted Cmd+K edits, to fully autonomous agents capable of planning and executing complex, multi-file changes across a codebase. The platform integrates state-of-the-art large language models including Claude, GPT-5.5, and Opus 4.7, while offering teams the flexibility to bring their own model API keys. It serves a broad spectrum of users, from individual developers to massive enterprise engineering organizations that treat AI assistance as a core production utility rather than a peripheral add-on.

The product ecosystem has expanded well beyond the desktop IDE. Cursor now ships a CLI for terminal-based agentic workflows, a Slack integration for collaborative engineering task triage, and GitHub-integrated agents that autonomously review pull requests. The flagship agentic interface, Composer 2.5, enables users to delegate end-to-end feature development to cloud agents that build, test, and demo software on isolated computers for human review. These capabilities are backed by a rapid research-to-production pipeline that has delivered secure codebase indexing, semantic search, shadow workspaces, and multi-agent collaboration.

Cursor generates revenue through tiered subscriptions for individual professionals and enterprise teams, with costs scaling according to AI feature depth and cloud agent compute consumption. The company targets software builders at high-growth startups and established technology giants alike, positioning itself as the default editor for teams that prioritize velocity. It competes in the rapidly evolving AI-native developer tooling market, where differentiation increasingly hinges on agentic reliability, context depth, and seamless workflow integration.

Origin story

Founded in 2022 in San Francisco by a group of MIT researchers, Anysphere began life as an applied research lab focused on language model fine-tuning and fast inference optimization for code. The founders concluded that mainstream editors were architecturally incapable of delivering truly native AI collaboration, so they forked Microsoft’s VS Code and rebuilt the user experience from the ground up to center the model rather than treat it as a plugin. Early technical milestones in 2023—context-aware completions and intelligent code navigation—earned Cursor a devoted following among top-tier engineers and fast-moving startups.

The company’s evolution tracked the industry’s shift from autocomplete to agents. In 2024, Anysphere shipped shadow workspaces and multi-agent collaboration, letting AI agents operate in isolated sandboxes without destabilizing production code. A year later, reinforcement learning improved the precision of autonomous edits. By May 2026, Cursor had grown from a slick editor into a multi-surface software creation platform encompassing a desktop IDE, terminal CLI, Slack integration, and mobile agents. The $105 million Series B at a $2.5 billion valuation cemented its status as a major independent player in developer tools, even as the team remains lean at roughly 20–50 employees. Public details on earlier funding rounds are limited, but the company’s growth has been fueled by organic developer adoption and high-profile endorsements from industry leaders at NVIDIA, Stripe, and Y Combinator.

Key products

Cursor Desktop

The AI-first IDE forked from VS Code, offering model-native autocomplete, targeted edits, and deep context awareness for production engineering.

Composer 2.5

An agentic interface that plans and executes multi-file changes; users can delegate end-to-end feature builds to AI for review.

Cursor Agent / Cloud Agents

Autonomous agents that build, test, and demo features end-to-end on cloud computers, operating in parallel with human developers.

Cursor CLI

A terminal-based coding agent that brings Cursor’s autocomplete and agentic capabilities into shell and CI/CD workflows.

Cursor for Slack

A team collaboration integration that triages engineering tasks, answers code questions, and triggers agents from chat.

Leadership

  • MT

    Michael Truell

    Co-founder & CEO

    MIT researcher; previously worked on language model systems and led Cursor's model-native UX vision.

  • SA

    Sualeh Asif

    Co-founder

    MIT mathematics and computer science background; helped architect Cursor's early inference stack.

  • AL

    Arvid Lunnemark

    Co-founder

    MIT researcher focused on optimization and systems engineering.

  • AS

    Aman Sanger

    Co-founder

    MIT background; contributed to Cursor's editor core and agent infrastructure.

Strengths & risks

Strengths

  • +Model-native UX with an autonomy slider from Tab completion to fully autonomous agents
  • +Rapid research-to-product pipeline shipping semantic search and secure codebase indexing
  • +Multi-surface distribution: Desktop IDE, CLI, Slack, GitHub, and mobile agents
  • +Strong social proof from NVIDIA, Stripe, YC, and top open-source creators
  • +Bring-your-own-model flexibility reducing lock-in to a single LLM provider
  • +Forked from VS Code, minimizing developer onboarding friction

Risks

  • Microsoft can replicate agentic features directly into GitHub Copilot and VS Code
  • Dependence on third-party LLM providers like Anthropic and OpenAI for core intelligence
  • Small headcount (20-50) may struggle to support enterprise security and compliance demands
  • Open-source alternatives and well-funded competitors could commoditize AI agents
  • Viral adoption must convert to durable enterprise revenue to justify $2.5B valuation

Recent moves

  1. Shipped secure codebase indexing for enterprise customers

    2026

    Published research and shipped features enabling secure indexing of large private codebases for AI context retrieval.

  2. Launched semantic search across codebases

    2026

    Released semantic search capabilities to improve context-aware code navigation and agent comprehension.

  3. Released Composer 2.5 with autonomous cloud agents

    2026

    Introduced Composer 2.5 and cloud agents that build, test, and demo features end-to-end on isolated computers.

  4. Expanded to CLI and Slack integrations

    2026

    Brought Cursor agents into the terminal and team chat, enabling PR reviews and task triage outside the IDE.

  5. Introduced mobile agent capabilities

    2026

    Launched a mobile agent experience allowing users to delegate coding tasks from mobile devices.

Competitive position

Cursor occupies a premium position in the AI-native developer tools market as the insurgent alternative to Microsoft’s GitHub Copilot. Where Copilot is an extension layered atop VS Code and GitHub, Cursor is a fork rebuilt around an "autonomy slider" that lets developers move seamlessly from Tab completion to targeted edits to fully autonomous agents. This model-native architecture allows Cursor to iterate faster on UX paradigms—such as Composer 2.5’s cloud agents and shadow workspaces—than incumbents beholden to legacy editor constraints. The product has become the default choice for many top engineering teams, startups, and AI researchers who prioritize speed and depth of context.

However, Microsoft retains an overwhelming distribution advantage. Copilot is bundled with GitHub and Microsoft’s enterprise suite, reaching millions of developers through existing procurement channels. Cursor must continuously out-innovate to justify its standalone existence and $2.5 billion valuation. It wins on user experience and agentic depth today, but could lose ground if Microsoft integrates similar multi-agent, CLI, and Slack capabilities deeply into its own stack. The next 12–24 months will determine whether Cursor can convert its viral developer love into a durable enterprise moat before incumbents close the feature gap.

What to watch

  • 01Enterprise revenue growth and conversion from viral individual adoption to paid teams
  • 02Microsoft's Copilot roadmap for agentic features and native IDE integration
  • 03Gross margins as cloud agent compute costs scale with autonomous usage
  • 04Security certifications and compliance wins needed for large enterprise deals
  • 05Developer retention if open-source or bundled alternatives match agentic capabilities

Frequently asked questions

How is Cursor different from GitHub Copilot?

Cursor is a VS Code fork built natively around AI with an autonomy slider from Tab to full agents, whereas Copilot is an extension. It also offers cloud agents, CLI, Slack, and GitHub integrations.

Can I use my own API keys or models?

Yes. Cursor supports bring-your-own-model, letting teams use their own API keys for Claude, GPT-5.5, and other models alongside Cursor’s hosted options.

What is Cursor Composer?

Composer is Cursor’s agentic interface for planning and executing multi-file changes. Composer 2.5 can autonomously build, test, and demo features using cloud computers.

Is Cursor secure for enterprise codebases?

Cursor offers secure codebase indexing and semantic search. Enterprises should review Cursor’s security documentation and compliance posture for their requirements.

Does Cursor work outside the desktop IDE?

Yes. Cursor offers a CLI for terminal workflows, a Slack integration for team collaboration, and agents that review PRs in GitHub.

What models power Cursor?

Cursor integrates Claude, GPT-5.5, Opus 4.7, and a specialized Tab model, with support for bringing your own model.

How much does Cursor cost?

Cursor offers individual Pro plans and Business/Enterprise tiers. Pricing scales with AI features and cloud agent usage; see cursor.com for current rates.

The bottom line

Anysphere has successfully evolved Cursor from a smart VS Code fork into a multi-surface software creation platform spanning desktop IDE, CLI, Slack, GitHub agents, and mobile agents. Its model-native "autonomy slider"—from Tab completion to fully autonomous cloud agents—gives it a genuine UX moat that top engineering teams at NVIDIA, Stripe, and YC have embraced. At a $2.5 billion valuation and only 20–50 employees, the company is operating at extreme leverage, but its future depends on converting viral developer adoption into durable enterprise revenue before Microsoft closes the agentic feature gap in GitHub Copilot.

The next 12–24 months will be critical. Cursor must scale its security and compliance posture to land larger enterprise deals, prove that cloud agent compute costs allow healthy gross margins, and defend its independence as an open-ecosystem alternative to Microsoft’s bundled stack. If it succeeds, it could become the definitive software creation layer for the agentic era; if incumbents replicate its autonomy slider and multi-agent workflows natively, Cursor’s distribution disadvantage could become existential despite its superior UX.

Visit Anysphere (Cursor)

Key products

  • Cursor
  • Cursor Composer

Founders & leadership

All founders →

Latest announcements

20 entries
  1. Customer story detailing how Faire doubled pull request throughput using Cursor Cloud Agents.

  2. Cursor is recognized as a Leader in the 2026 Gartner Magic Quadrant for Enterprise AI Coding Agents.

  3. Josh Ma shares insights and lessons learned from building cloud agents at Cursor.

  4. Cursor 3.5 introduces Shared Canvases and the /loop skill for collaborative agent workflows.

  5. Cursor 3.5 brings improvements to Cursor Automations.

  6. Cursor in Jira

    productMay 19, 2026

    Cursor launches a new integration with Jira.

  7. Introducing Composer 2.5

    researchMay 18, 2026

    Cursor releases Composer 2.5 with substantial improvements in intelligence and behavior over Composer 2, particularly on long-horizon agentic tasks.

  8. New development environments for cloud agents are launched to improve agentic workflows.

  9. Cursor rolls out updates to Bugbot for both teams and individual users.

  10. PayPal shares how Cursor's AI tools expand what is possible to build beyond efficiency gains.

  11. The team details how they bootstrap Composer using autoinstall.

  12. Stefan and Jediah discuss ongoing improvements to Cursor's agent evaluation harness.

  13. Roshan Sadanani introduces the Cursor SDK for building programmatic agents.

  14. NAB shares how Cursor accelerated legacy system migrations.

  15. Cursor announces a partnership with SpaceX focused on model training.

  16. Keeping the Cursor app stable

    announcementApr 21, 2026

    Andrew and Kevin discuss strategies for keeping the Cursor application stable.

  17. Luke Melas-Kyriazi explores how better AI models allow developers to pursue more ambitious work.

  18. Users can now interact with agent-created visualizations directly in Cursor canvases.

  19. Amplitude shares how they increased production code shipping by 3x using Cursor.

  20. Meet the new Cursor

    productApr 2, 2026

    Cursor 3 is introduced as a unified workspace for building software with agents.

Related companies

All companies →