Cursor, a platform that was recently one of the largest API clients for OpenAI, Anthropic, and Google, has announced the release of Cursor 3. This is not a routine update to its development environment. The product, codenamed Glass, represents an aggressive bid for leadership in the niche of AI agents for programmers, a space where Cursor previously consumed third-party services. The company is betting that specialized coding solutions, such as Claude Code or Codex, have already proven their viability. Their popularity among millions of developers clearly signals a real demand for delegating complex tasks to AI. This, in turn, directly threatens the dominance of universal LLMs.
"Our profession has fundamentally changed in recent months," stated Jonas Nele, one of Cursor's engineering leads. "Many products that were the foundation of our success are losing their relevance." This is difficult to dispute. Over the past eighteen months, OpenAI and Anthropic have been heavily investing in their own, often subsidized, AI coding services. Their strategy is straightforward: bet on delegating entire blocks of work to AI agents, bypassing the universal assistant stage. Cursor 3 incorporates a new chat system that allows developers to formulate tasks in natural language. An AI agent takes these tasks and executes them, with the process visibly displayed in the IDE's side panel.
The core of Cursor 3, according to company representatives, is an "agent-first" approach integrated into a familiar development environment. Local preview of generated code after task execution in the cloud is, of course, a technical detail. However, it conceals a strategic maneuver: the Cursor team understands that developers choose tools that save time and increase productivity. Direct dependence on third-party LLMs, despite the purchase volumes, now appeared to be a vulnerability. By attempting to create its own specialized AI coding pipeline, Cursor aims to directly compete with OpenAI and Anthropic's solutions, but within its own "ecosystem."
Why this matters for business: Cursor, which grew by utilizing third-party LLMs, is now actively trying to embed itself into a new reality where specialized AI agents for coding are capturing market share from universal models. This is a direct challenge to OpenAI and Anthropic, whose own solutions have already earned developer trust. For CEOs, this signals that the competitive race for programmers is shifting from universal LLM assistants to specialized AI agents. This creates new niches for players like Cursor, but simultaneously intensifies pressure from tech giants ready to subsidize their offerings. Development teams now face a choice: use universal but less specialized LLMs, or take a risk by betting on agents that could become platform-dependent.
The market for AI-powered developer tools is fragmenting. Cursor's move suggests that the future lies in specialized agents tailored for specific workflows, rather than general-purpose LLMs, potentially forcing a strategic reevaluation from major AI providers.