Simular AI
Category: AI Agents
Builds autonomous computers / computer-use agents: flagship Sai is an always-on AI co-worker that operates a secure cloud (or BYOD) desktop like a human—GUI, browser, terminal, APIs—plus open-source Agent S frameworks for research and developers. Simular AI was founded in 2023. The company is led by Ang Li. Based in Palo Alto, California, United States. Team size: 10-50. Total funding raised: ~$27M total (all rounds). Latest round: Series A. Key investors include Basis Set Ventures, Flying Fish Partners, Samsung NEXT Ventures, South Park Commons, Felicis Ventures, NVentures, Xoogler Ventures, Lenny Rachitsky.
- Founded
- 2023
- Headquarters
- Palo Alto, California, United States
- Team size
- 10-50
- Total funding
- ~$27M total (all rounds)
Value proposition
Enables developers and users to deploy reliable, research-driven AI agents that can handle real-world digital complexity through improved reasoning, collaboration, and robust automation capabilities.
Products and solutions
Sai (always-on AI co-worker on cloud remote desktop / BYOD), Agent S3 (open-source computer-use agent framework), Simular for macOS / Windows desktop agents, Simular Cloud / Windows 365 for Agents enterprise deployment
Unique value
Neuro-symbolic computer-use agents: neural exploration plus conversion of successful runs into deterministic reusable code (Simulang), cutting cost/token use ~100x. Agent S3 / Sai first surpassed human baseline on OSWorld (72.6%) and leads OSWorld 2.0 at 73% vs GPT/Opus at lower cost.
Target customer
Enterprise developers and individual users building or utilizing applications that require advanced autonomous agents for task automation on computers.
Industries served
Enterprise Software, Insurance, Healthcare, Business Development / Sales, Venture Capital, Knowledge Work / Productivity
Technology advantage
Proprietary architecture inspired by reinforcement learning techniques from DeepMind. The system combines large language models (LLMs) for exploration with the ability to convert successful task completions into stable, symbolic code for reliable and repeatable execution.
How they differentiate
Focuses on multi-agent, collaborative, and robust architectures for complex environments, as opposed to single-task automation or code generation. The architecture converts successful agent actions into symbolic, reusable code for increased reliability.
Main competitors
Anthropic (Claude Computer Use / Cowork), OpenAI (Operator / computer-use), Manus AI, Cognition (Devin), Microsoft (Windows 365 Agents / Copilot agents)
Key partnerships
Microsoft (Windows 365 for Agents — early pilot partner; Sai demonstrated at Microsoft Build 2026; platform now GA), Samsung NEXT (seed investor; Korea ecosystem), Undisclosed enterprise pilots
Notable customers
Car dealership (automating VIN number searches), HOAs (extracting contract information from PDFs)
Major milestones
Achieved #1 ranking on the OSWorld Computer Use Benchmark., Received Best Paper Award at the ICLR 2025 Agentic AI for Science workshop., Featured in Wired, TechCrunch, The Register, MIT Tech Review, IBM Think., Launched Simular for macOS / Simular 1.0 desktop agent., Raised $21.5M Series A led by Felicis (Dec 2025); ~$27M total., Accepted into Microsoft Windows 365 for Agents; demoed Sai at Microsoft Build 2026 (GA)., Launched Sai (Mar 2026) — always-on cloud AI co-worker., Agent S3 first to surpass human OSWorld baseline at 72.6% (Dec 2025); Agent S3 paper accepted TMLR 2026., Sai tops OSWorld 2.0 at 73% beating GPT-5.6 Sol and Opus 5 at ~2/3 the cost (Aug 28 2026)., Team scaled to ~30 across Palo Alto and Singapore (Aug 2026); SIMULAR PTE. LTD. incorporated Jul 2025.
Growth metrics
Team of ~30 across Palo Alto and Singapore (Aug 2026); shipping weekly updates for Sai (always-on cloud AI co-worker launched Mar 2026); neurosymbolic stack cut token usage ~100x vs industry average; Sai leads OSWorld 2.0 at 73% (Aug 2026).
Market positioning
Positioned as a provider of advanced, research-driven infrastructure for desktop and browser-based AI agents, targeting both developers and enterprise users.
Geographic focus
Global, with HQ and primary presence in Palo Alto, United States, and a major engineering/product hub in Singapore (~half of team); Asia engagement via Samsung NEXT and Korea market interest.
Patents and IP
Information not publicly available.
About Ang Li
Ang Li is the Founder and CEO of Simular, a company focused on creating AI agents that can operate computers like humans. His work aims to build lifelong autonomous agents. Prior to Simular, he contributed at Google DeepMind, Baidu Apollo, Facebook AI Research, CMU Robotics, Apple, Google Streetview, and Comcast Labs DC. Ang holds a Ph.D. in computer science from the University of Maryland College Park, focusing on spatial perception from visual and linguistic information.
Latest news about Simular AI
More AI Agents companies
Official website: https://simular.ai