Simular

Market position

Simular positions itself against both LLM-only agents, which it characterizes as good explorers but unreliable executors, and traditional RPA, which it describes as deterministic but rigid. It claims to be the only agent topping browser, computer and smartphone benchmarks simultaneously, and cites a #1 ranking on the OSWorld computer-use benchmark.

Stated differentiators are the neuro-symbolic combination of LLM reasoning with symbolic code for repeatability, continual learning that adapts to changing interfaces, human-in-the-loop supervision that lets the agent learn to repeat successful workflows, transparent and inspectable execution, cross-platform operation without app-specific APIs, and guardrails requiring confirmation before critical actions.

Technology

Simular uses a neuro-symbolic, hybrid agent architecture that combines LLM-based neural exploration ("System 1") with symbolic, deterministic execution ("System 2") and continual/lifelong reinforcement learning, aiming to pair LLM flexibility with the repeatability of symbolic code. Agents interact directly with UI elements across macOS, Windows and Linux without requiring APIs, and support replayable, deterministic execution. Developer-facing automation uses Simulang, a JavaScript-based DSL, with webhook integration into production pipelines. Agent S is the company's open agentic framework; the team publishes papers including "On the Reliability of Computer Use Agents" and "The Unreasonable Effectiveness of Scaling Agents for Computer Use."

Go-to-market

Distribution is direct through the company website with downloads, an invite code flow for Sai and a Windows waitlist; the site asks new users how they heard about Simular, listing Product Hunt, TikTok, Instagram, X, Google search, LinkedIn, YouTube, Reddit and word of mouth. The company builds developer and research awareness through open-source releases on GitHub and Hugging Face, published papers, benchmark leaderboard results, a research blog, community channels, and press coverage (Wired, TechCrunch, MIT Technology Review, The Register, IBM Think, Microsoft Build). A business offering targets operations teams.

Knowledge workers who spend large amounts of time on repetitive on-screen work, developers seeking programmable desktop automation, and business/operations teams; the company cites user tasks in insurance, recruiting and travel, plus use cases in sales and marketing, administration, recruiting, healthcare, e-commerce, real estate, engineering and finance.

Geography

The Series A announcement is datelined San Francisco, California; a startup directory lists the HQ as San Carlos, United States. The company refers to a "global team" photo dated December 2025. Products run on macOS, Windows and Linux.

History

Simular's seed round of $5 million was led by Basis Set Ventures, with earlier backers including Samsung NEXT, Xoogler Ventures, Lilian Weng and Abhinav Gupta. In 2025 its Agent S framework reached 69.9% on the OSWorld benchmark and the team received a Best Paper Award at the ICLR 2025 Agentic AI for Science workshop. On December 2, 2025 the company announced a $21.5 million Series A led by Felicis Ventures and simultaneously debuted Simular 1.0, its native desktop agent. In December 2025 it reported computer-use results exceeding human performance on OSWorld, and it continued publishing product and research updates into 2026.

Risks & controversies

No controversies are reported in the sources. One directory listing (Dealroom-sourced) states "no known external funding" and describes the company differently, conflicting with the funding disclosed by the company and press.