Simular, a startup building AI agents for Mac OS and Windows, has raised a $21.5 million Series A funding round. The investment was led by Felicis, with participation from existing seed investors including NVentures, NVIDIA’s venture arm, and South Park Commons.
This startup stands out in the agentic AI space because its systems are designed to control the personal computer itself, rather than just a web browser. Agentic AI refers to systems that can autonomously complete complex tasks with minimal human intervention. The company’s technology can directly interact with the operating system, moving the cursor and clicking. This allows it to replicate a wide range of human digital activities, such as copying and pasting data into spreadsheets.
Simular recently announced the release of its 1.0 version for Mac OS. It is also collaborating with Microsoft to develop an agent for Windows. The startup is one of only five agentic companies selected for the Windows 365 for Agents program that Microsoft announced in mid-November. The other companies in the program are Manus AI, Fellou, Genspark, and TinyFish. While the timeline for the Windows version remains unclear, the company expects it to be as popular, if not more popular, than the Mac version.
The founders bring significant expertise to the venture. CEO and co-founder Ang Li is a continuous learning scientist who previously worked at Google’s DeepMind, where he met his co-founder, reinforcement learning specialist Jiachen Yang. Their work at DeepMind was applied, focused on improving products like Waymo, rather than being purely academic.
This product background is crucial because the path to functional AI agents involves solving major technical challenges. A primary issue is that large language models hallucinate some percentage of the time. Since agentic tasks can involve thousands or millions of discrete steps, a single hallucination can ruin the entire process, and the risk increases with the number of steps required.
A common proposed solution is to make the “non-deterministic” LLM “deterministic” by scripting its actions to be the same every time. However, this approach limits the agent’s creative problem-solving ability. Simular aims to marry both concepts. Its agent explores a task freely, with a human user providing course correction, until it achieves success. The user can then lock in that successful workflow, converting it into deterministic, repeatable code.
The company’s approach is enabled by what it calls “neuro symbolic computer use agents,” a technology it claims is not used by other agent companies. This system is not solely based on a large language model. Instead, it allows the LLM to write code that becomes deterministic. Once a successful workflow is found, it can be reliably repeated. An additional benefit is that this deterministic code resides with the end user, who can inspect, audit, and trust it.
Early beta customers include a car dealership automating VIN number searches and homeowners associations extracting data from PDF contracts. The company’s open source project, currently available only for Mac OS, has been used to create automations for content creation and sales and marketing.
Simular previously 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.