Cursor Builds 'Sand' Agent to Rival Claude, ChatGPT Work

The interesting thing about Cursor quietly building a general-purpose office assistant is less the product itself and more what it says about where the coding-tools winners think the next fight actually happens. According to The Information, Cursor has been developing a personal AI agent internally referred to as Sand, designed to reply to emails and texts and organize spreadsheets, and it is the company's first product aimed at general users rather than developers.

The reporting says work began in April as Cursor leased computing capacity from SpaceX's AI division, SpaceXAI, and an internal version rolled out to Cursor employees in late June 2026. If it ships, Sand would compete directly with Anthropic's Claude Cowork and OpenAI's ChatGPT Work, the two horizontal AI-for-office products already picking up early enterprise pilots. In other words, the company best known for AI pair programming is putting a foot in the door of the general knowledge-worker assistant category.

The strategic subtext is the pending SpaceX acquisition. As CNBC reported, SpaceX will spend $60 billion to acquire Cursor in an all-stock deal expected to close in the third quarter, and Sand could plausibly serve SpaceXAI's enterprise ambitions as much as Cursor's own. That reframes the question from "is Cursor pivoting away from developers" to "is SpaceXAI assembling a full productivity stack under the Cursor brand." Either reading is a real change for buyers who assumed Cursor would stay in coding lanes.

The honest caveat is that this is single-source reporting on an unreleased product. The article makes clear it is not certain that Sand will make it to market, and what the reporting does not give you is any detail on which models power it, how it differs functionally from Cowork or ChatGPT Work, or what pricing and general availability might look like. Internal dogfooding is not the same as a shipped product, and pending acquisitions have a way of quietly rerouting roadmaps.

Still, the direction is worth watching. If Sand ships in something like its reported form, enterprise buyers get a third serious general-purpose contender at exactly the moment Anthropic and OpenAI are trying to lock in seats. And if it does not ship as a Cursor-branded product, that itself will tell you something useful about how SpaceXAI plans to organize the assets it is acquiring. Take the specifics as reported, not settled.