Simular: Pioneering Agentic AI for Mac OS and Windows with $21.5 Million Series A

Simular, an innovative startup developing AI agents for Mac OS and Windows, has recently announced a significant achievement: a $21.5 million Series A funding round led by Felicis. Notable players like NVentures (Nvidia’s venture arm), existing seed investor South Park Commons, and others are also part of this funding milestone. This funding is poised to accelerate Simular’s vision of creating AI agents that operate at a higher capability and autonomy than typical solutions.

What Makes Simular Unique?

Unlike other AI solutions that primarily focus on browser interactions, Simular is ambitious in its goal to control the entire PC environment. This innovative approach taps into what is known as "agentic AI," which refers to AI systems that can autonomously complete complex tasks with minimal human intervention. Co-founder and CEO Ang Li elegantly explains, "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." For instance, the AI can perform tasks such as copying and pasting data into a spreadsheet seamlessly.

Release of Version 1.0 and Upcoming Developments

Just recently, Simular unveiled its 1.0 version for Mac OS, generating buzz among early adopters. The company is also in collaboration with Microsoft to develop an agent specifically for Windows, as part of the Windows 365 for Agents program. This partnership places Simular among an elite group of five agentic companies, including Manus AI, Fellou, Genspark, and TinyFish. Although details about the Windows version timeline remain vague, Li assures users that it will be as popular—if not more so—than the Mac OS version.

Founders with Proven Expertise

What sets Simular apart is also its founders’ exemplary backgrounds. Ang Li, a continuous learning scientist, formerly honed his skills at Google’s DeepMind. Partnering with him is Jiachen Yang, a reinforcement learning specialist. Their experience is crucial as they transition their extensive research capabilities into actionable products aimed at enhancing user experiences—not just theoretically but also practically.

Navigating Technical Challenges

As promising as the future of agentic AI may be, there are considerable technical hurdles to overcome. A primary issue exists with large language models (LLMs), which are known to "hallucinate" information, leading to inaccuracies. This becomes particularly concerning for agentic tasks that could involve thousands to millions of discrete steps. A single hallucination could undermine the entire task, raising concerns about reliability.

One potential solution is to make LLMs "deterministic," scripting their responses to ensure consistent outcomes. However, this could reduce the creativity that characterizes agentic problem-solving. To address this, Simular proposes a hybrid approach: while letting agents explore and iterate freely, users can step in to correct the course, locking in successful workflows that then become repeatable and deterministic.

Unveiling Neuro-Symbolic Computing

The revolutionary aspect of Simular’s technology lies in what they call “neuro-symbolic computer use agents.” Ang Li emphasizes that their approach goes beyond typical LLM-based models. Instead, the agents are designed to generate code that enables deterministic behavior while also preserving the flexibility needed for creative problem-solving. This capability allows for workflows that not only succeed on the first attempt but can also be replicated successfully in subsequent executions.

Empowering Users with Control

What truly differentiates Simular’s offering is that the deterministic code generated during task execution is accessible to the end-user. This model inspires trust, as users can inspect and audit the code, ensuring transparency. Li points out, "Once they have the code, they can trust it, because they can inspect it, they can audit it, they can see what’s going on."

Real-World Applications and Open Source Initiatives

Simular’s innovative methodology is already capturing attention from various sectors. Early beta customers include a car dealership automating VIN number searches and homeowners’ associations extracting contract details from PDFs. Their open-source project—currently only available for Mac OS—has resulted in useful automations spanning content creation to marketing and sales.

Funding and Future Prospects

Having previously secured a $5 million seed round, Simular now boasts a total funding of approximately $27 million, with investments from notable names such as Basis Set Ventures, Flying Fish Partners, and Samsung NEXT. With such robust backing, the startup is well-positioned to continue influencing the landscape of agentic AI.

Stay tuned for the latest developments in AI technology as Simular ventures into uncharted territories, promising to revolutionize how we interact with our computers and automate tasks with unprecedented ease.

Inspired by: Source