The Reality Behind Cursor's AI-Generated Web Browser Claims

Cursor, an AI-integrated development environment (IDE) company, recently announced the development of a web browser purportedly built with its AI agents, specifically using GPT-5.2. CEO Michael Truell claimed that the project involved over three million lines of code and a custom rendering engine developed from scratch in Rust. However, the actual functionality of the browser has come under scrutiny, revealing significant discrepancies between the claims and the reality.

Cursor is better at marketing than coding

Despite the ambitious claims, reports indicate that the browser is not fully operational. Cursor engineer Wilson Lin's blog post about the project described the endeavor as "extremely difficult," and independent developers who accessed the code found that it often fails to compile and runs poorly. The project reportedly consumed between 10 to 20 trillion tokens and incurred substantial costs, yet the resulting codebase was far from functional. Many developers noted that the browser's performance was abysmal, with pages taking approximately a minute to load.

Cursor is better at marketing than coding

Lin defended the project on Y Combinator, asserting that the JavaScript engine used was a custom VM developed as part of the browser. However, critics argue that this does not align with the claim of building from scratch, as it appears to be derived from Lin's personal JavaScript parser project. The criticism extends to the overall design, which Lin described as a "uniquely bad design that could never support anything resembling a real-world web engine." This perspective highlights a broader concern regarding the marketing of AI capabilities, suggesting that the project was presented as a significant milestone rather than an internal learning exercise.

Cursor is better at marketing than coding

The situation with Cursor raises important questions about the current state of AI in software development. While tools like Cursor can assist with tasks such as autocomplete and refactoring, the marketing narrative often suggests that AI can handle entire projects autonomously. This disconnect between expectation and reality may lead to a decline in trust among developers, particularly as many enterprise AI initiatives fail to deliver meaningful results. Sarah Friar, CFO of OpenAI, emphasized the need for "practical adoption" of AI technologies, suggesting that the focus should shift towards achieving tangible outcomes before making grand claims.

Cursor is better at marketing than coding

There is a notable discrepancy between the optimistic portrayal of Cursor's browser and the critical feedback from developers. While Cursor's marketing suggests a groundbreaking achievement, the technical community's response indicates significant shortcomings in functionality and performance. The lack of passing Continuous Integration (CI) tests and reproducible builds further complicates the narrative.

Cursor is better at marketing than coding

• “We built a browser with GPT-5.2 in Cursor.” — Michael Truell, CEO of Cursor

Cursor is better at marketing than coding

• “building a browser from scratch is extremely difficult.” — Wilson Lin, Engineer at Cursor

Cursor is better at marketing than coding

• “So I agree this isn't just wiring up of dependencies, and neither is it copied from existing implementations: it's a uniquely bad design that could never support anything resembling a real-world web engine.” — Wilson Lin, Engineer at Cursor

Cursor is better at marketing than coding

The case of Cursor serves as a cautionary tale about the potential pitfalls of overhyping AI capabilities in software development, emphasizing the need for realistic assessments and validation of technological advancements.

Cursor is better at marketing than coding