Simular’s AI Agent Revolutionizes Mac and Windows Automation with $21.5M Funding
In a significant advancement for personal computing, Simular, an innovative startup specializing in AI agents for Mac OS and Windows, has successfully secured $21.5 million in Series A funding. This round was led by Felicis, with participation from NVentures—Nvidia’s venture arm—and existing seed investor South Park Commons, among others.
Unlike many AI initiatives that focus on browser-based automation, Simular’s approach is to control the entire PC environment. Their AI agent is designed to autonomously perform tasks traditionally executed by users, such as navigating interfaces and managing applications. Co-founder and CEO Ang Li elaborated on this capability, stating, We can literally move the mouse on the screen and do the click. So it’s more capable of doing, repeating whatever human activities in the digital world, citing examples like copying and pasting data into spreadsheets.
On December 1, 2025, Simular announced the release of its 1.0 version for Mac OS. Simultaneously, the company is collaborating with Microsoft to develop a Windows-compatible agent. Simular is among five companies selected for Microsoft’s Windows 365 for Agents program, announced in mid-November. The other participants include Manus AI, Fellou, Genspark, and TinyFish. While a specific release date for the Windows version remains undisclosed, Li expressed confidence that it will be as, if not more, popular than the Mac version.
The expertise of Simular’s founders adds credibility to their ambitious goals. Ang Li, a continuous learning scientist, and Jiachen Yang, a reinforcement learning specialist, both have backgrounds at Google’s DeepMind. Their work there was not purely academic but aimed at enhancing Google products, including Waymo.
Developing AI agents capable of autonomously completing complex tasks presents significant technical challenges. One major issue is the tendency of large language models (LLMs) to produce inaccuracies or hallucinations. Given that agentic tasks may involve thousands to millions of discrete steps, an error at any point can invalidate the entire process. Moreover, the likelihood of such errors increases with the number of steps involved.
To address this, one approach is to make the inherently non-deterministic LLMs deterministic, ensuring consistent responses or actions. However, this method risks stifling the creative problem-solving abilities of the agent.
Simular proposes a hybrid solution. Their agent iterates freely on tasks, with human users providing guidance as needed. Once a successful workflow is identified, it is locked in as deterministic code, making it repeatable. Li explains, Our solution is, let agents keep exploring the successful trajectory. Once you found a successful trajectory, that becomes deterministic code.
This approach allows the agent to maintain creativity during the exploration phase while ensuring reliability in execution. Additionally, the deterministic code is accessible to end users, enabling them to inspect, audit, and trust the processes.
Early beta customers have demonstrated the practical applications of Simular’s technology. For instance, a car dealership has automated Vehicle Identification Number (VIN) searches, and Homeowners Associations (HOAs) are extracting contract information from PDFs. The company’s open-source project, currently available for Mac OS, has facilitated automations in content creation, sales, and marketing.
Prior to this funding round, Simular raised a $5 million seed round, bringing its total funding to approximately $27 million. Other investors include Basis Set Ventures, Flying Fish Partners, Samsung NEXT, Xoogler Ventures, and angel investor Lenny Rachitsky.
As Simular continues to develop and refine its AI agents, the potential for transforming personal computing through automation becomes increasingly tangible. By addressing key technical challenges and leveraging the expertise of its founders, Simular is poised to make significant strides in the realm of AI-driven automation.
