Richard Lee

[Simular @SimularAI]

Our computer-use agent @sai_borg just beat Opus 5 and GPT-5.6 Sol on OSWorld 2.0.

Sai scored 73% on the CUA benchmark where each complex task takes a skilled human over an hour to execute.

And Sai did it at about 2/3 the cost per task of Opus and GPT 💸

The unlock is our neurosymbolic framework that pairs the exploratory power of neural networks with the logic of symbolic code, resulting in reliability and cost saving.

Read the full report: https://t.co/cQjxXxUavp

[Richard Lee @Richard_simular]

@sai_borg earning that paycheck

[Rita Liao @ritacyliao]

Not @Richard_simular telling me @sai_borg's deterministic code replay drops token usage by 90%

For a repetitive task our accountant hates 🤓

[Rita Liao @ritacyliao]

Not @Richard_simular telling me @sai_borg's deterministic code replay drops token usage by 90%

For a repetitive task our accountant hates 🤓

[Ang Li @angli_ai]

Agent S3, the first computer-use agent that surpasses human-level performance in OSWorld v1 benchmark, has now been accepted to @TmlrOrg (Transactions on Machine Learning Research).

Congratulations to the @SimularAI team!

@chalo2000, vincent, @Richard_simular, jiachen, @xwang_lk

Stay tuned for what's next!

[Zening Chen @ZeningChen42844]

https://t.co/3qJxtpJubC

[Richard Lee @Richard_simular]

Using Sai to judge agent projects made by Simular product suite

[Sai @sai_borg]

Judging time at #Calhacks - who's gonna win our $500 prize? 👀

[Simular @SimularAI]

Today marks an important milestone in the history of @SimularAI, the autonomous computer company.

Our open source computer-use agent, 𝐀𝐠𝐞𝐧𝐭 𝐒, scored 72.6% on the OSWorld benchmark, surpassing the human baseline (72.36%) for the first time ever.

This milestone matters because it shows AI can now use computers the way humans do, and, in many cases, do it better.

This is a glimpse of a future where work becomes faster, more accessible, and more empowered for everyone.

#AI #Automation #Simular #AgenticAI #ComputerUse

[Simular @SimularAI]

We just raised $21.5M Series A to build autonomous computer agents that can actually use a computer like a real teammate 🖥️🤝

Led by @felicis, with NVentures (@nvidia's venture arm), @BasisSet, @flyingfishvc, @spc & @lennysan.

Launching Simular 1.0:
• Native desktop agent (MacOS 15+ silicon)
• Operates across apps & the browser
• Learns from your corrections like a coworker
• Powered by Agent S — 69.9% on OSWorld, approaching human performance (72%)

Autonomous computers aren’t about replacing humans — they’re about cooperation.

[Simular @SimularAI]

Have tasks you wish an AI agent could do on your desktop, but all you get are slides decks or vibe-coded apps?

Simular 1.0 is here to actually relieve humans from repetitive computer work.
Not a black-box LLM that invents its own steps, but an AI teammate you can teach to mirror your real workflow and repeat its success infinitely.

Meet your new teammate, Simular 1.0.

[Simular @SimularAI]

We are proud to be invited to test @Microsoft 365’s platform for agents!

At Simular, our mission has always been to free humans from computer tasks. Building for this large ecosystem of desktop and cloud apps brings us one step closer to that vision.

https://t.co/Wvmt8hMYXS

As one of the few teams with early access to the API and documentation, we’ll be testing Microsoft’s state-of-the-art agentic infrastructure and providing feedback from the front lines. Also can’t wait to see what @ManusAI @genspark_ai @FellouAI @Tiny_Fish build for the pilot.

Our latest Agent S3 already achieves a 69.9% success rate on OSWorld, just a 2% gap from human performance at 72%. Stay tuned for what we are running in 365!

@msPartner @angli_ai

[Xin Eric Wang @xwang_lk]

🚀 Introducing 𝐀𝐠𝐞𝐧𝐭 𝐒3, the most advanced computer-use agent, now 𝐚𝐩𝐩𝐫𝐨𝐚𝐜𝐡𝐢𝐧𝐠 𝐡𝐮𝐦𝐚𝐧-𝐥𝐞𝐯𝐞𝐥 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞🧠💻

Just one year ago, Agent S scored ~20% on OSWorld: SOTA then, but far from human 72%.

Today, Agent S3 reaches 6̳9̳.̳9̳%̳ (⬆10% over prior SOTA), nearly matching humans on computer use.

Few imagined this gap could close so fast, but with steady progress and 𝐟𝐢𝐫𝐬𝐭-𝐩𝐫𝐢𝐧𝐜𝐢𝐩𝐥𝐞 simplicity, @SimularAI did it in just one year.

🧪 The s̵e̵c̵r̵e̵t̵ 𝐨𝐩𝐞𝐧 𝐫𝐞𝐜𝐢𝐩𝐞:
→ 𝐁𝐞𝐡𝐚𝐯𝐢𝐨𝐫 𝐁𝐞𝐬𝐭-𝐨𝐟-𝐍 — scaling agents 𝘵𝘩𝘦 𝘳𝘪𝘨𝘩𝘵 𝘸𝘢𝘺
→ Simpler design, stronger baseline
→ 100% 𝐨𝐩𝐞𝐧 𝐬𝐨𝐮𝐫𝐜𝐞 (as always)

Learn more at https://t.co/F2wFiRKQvM & 👇

[Xin Eric Wang @xwang_lk]

🚀 Meet 𝐀𝐠𝐞𝐧𝐭 𝐒2.5, our latest open-source Computer Use Agent, setting 𝐧𝐞𝐰 𝐒𝐎𝐓𝐀 𝐨𝐧 𝐎𝐒𝐖𝐨𝐫𝐥𝐝 𝐕𝐞𝐫𝐢𝐟𝐢𝐞𝐝!

✨ Upgrades: Flexible context engineering + advanced high-level reasoning.

📈 OSWorld Verified Results:
- 100 steps: S2.5 56.0% (Prev. SOTA: 53.1%)
- 50 steps: S2.5 54.2% (Prev. SOTA: 50.6%)

⚡️ Simpler, better, more efficient: fewer steps & just one action per step, no costly Best-of-N sampling.

[Simular @SimularAI]

1/ Wait, Bigfoot figured out how to run a startup without drowning in multitasking 👀

It found 𝗦𝗶𝗺𝘂𝗹𝗮𝗿 𝗣𝗿𝗼, the world’s first production-grade, computer-use agent that runs thousands of steps without a hiccup - working 24/7 so he didn’t have to.

So how does Simular Pro work?

Most agents use LLMs - great explorers, but flaky when # steps grows; RPAs are stable but rigid. Simular Pro uses 2 agents, maximizing generality & repeatability:

- Neural: explores
- Symbolic: executes

Every action is editable, deterministic code.

[Richard Lee @Richard_simular]

Agent S2 achieves SOTA on OSWorld, WindowsAgentArena & AndroidWorld. Improvements of 18.9% & 32.7% over baselines on OSWorld. Also 52.8% on WindowsAgentArena & 16.52% on AndroidWorld.

[Richard Lee @Richard_simular]

Agent S2: A new framework for computer use agents, tackling GUI interaction. It delegates tasks across generalist & specialist models, improving human productivity by automating digital tasks.

[Richard Lee @Richard_simular]

Agent S2 excels at precise GUI localization using Mixture-of-Grounding, & handles long-horizon tasks with Proactive Hierarchical Planning. This allows it to adapt to changing environments in real time.

[Richard Lee @Richard_simular]

Link to Agent S2 notebook LM here: https://t.co/e2bxYNsQts

[Richard Lee @Richard_simular]

Agent S2: A new compositional framework for computer use agents! It delegates tasks across generalist & specialist models. Aims for better GUI grounding & planning.

[Richard Lee @Richard_simular]

S2 achieves SOTA on OSWorld, WindowsAgentArena & AndroidWorld! Significant gains over baselines like Claude & UI-TARS, showing its generalization power.