Problem

Cyberdesk addresses the last-mile automation problem created by legacy Windows software that lacks modern APIs, especially EHRs, ERPs, accounting systems, and logistics portals. Developers and operations teams otherwise have to perform repetitive data entry, extraction, reconciliation, and form-filing manually or maintain brittle integrations and screen scrapers. The economic pain is a combination of labor overhead and the speed, reliability, and cost limitations of conventional computer-use models and robotic process automation; YC describes traditional RPA as a $6 billion industry that still serves only a small fraction of automatable desktop work, within a much larger enterprise knowledge-work opportunity.

The clearest use case is healthcare: entering or extracting patient information across systems such as EHRs and OpenDental, eliminating repetitive chart work so clinical staff can focus on patients. Finance and logistics offer parallel use cases, including reconciling accounts, filing forms, and reading or writing data in legacy ERP and logistics applications.

Product / Service

Cyberdesk is a self-learning computer-use agent and developer toolkit for Windows. A user installs its driver on a physical or virtual Windows machine, describes a workflow in plain English, and triggers it through an API; TypeScript and Python SDKs, scheduling, retries, error handling, workflow logs, and machine management are built in. It can run on cloud or on-premises Windows environments without requiring firewall rules or a VPN.

The agent memorizes completed desktop tasks so repeated runs become cheaper, faster, and deterministic, while falling back to computer-use models when unexpected pop-ups or edge cases occur. Cyberdesk says this approach can repeat tasks three to five times faster and more cheaply than an initial agent execution, while minimizing exposure to the Windows environment and avoiding the fragility of conventional scripts.

Market

Cyberdesk competes in the emerging computer-use-agent and AI desktop-automation market, with YC categorizing it across artificial intelligence, developer tools, robotic process automation, workflow automation, and enterprise software. Its practical alternatives include traditional RPA suites, programmatic desktop-automation tools such as PyAutoGUI, and custom integrations or screen scrapers. Cyberdesk differentiates around legacy Windows coverage, self-learning repetition, developer APIs, and deployment on existing physical or virtual machines rather than requiring every target application to expose an API.

The company is an active Y Combinator Summer 2025 startup founded in 2025 by Mahmoud Al-Madi and Alan Duong, with a two-person team listed by YC. It is not purely pre-revenue: YC reports that the founders obtained a paid customer immediately after shipping the MVP and later closed additional deals, while Cyberdesk and LinkedIn identify customers or users in healthcare, finance, accounting, construction, and logistics. The evidence reviewed does not disclose a reliable revenue total, customer count, or pricing, so the best characterization is early commercial traction rather than established scale.