SimWorld Studio introduces automated platform using SimCoder to generate interactive 3D environments in Unreal Engine 5 for embodied AI agents

VIEWS2MBOOKMARKS475LIKES993RETWEETS69REPLIES32

Super excited to introduce SimWorld Studio!🏭

With built-in coding agent SimCoder🦞, you can vibe-code physical 3D scenes in Unreal Engine -- then train embodied agents directly inside the generated world.🦾

From environment generation to agent training, end-to-end. 🚀

SimWorld@simworld_ai

Environment generation is the missing scaling axis for embodied AI.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn.

Env-agent co-evolvution improves navigation success 50% → 90%.

From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.

40d2M993475

Lianhui Qin@Lianhuiq

Scaling embodied AI starts with automating the environments.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D environments where agents act, fail, and learn.

With coding-agent + embodied-agent co-evolution, navigation success improves from 50% → 90%.

1/

40d54.6K241153

Xiaokang(Koe) Ye@koe_ye40329

Environment scaling is becoming one of the key bottlenecks for embodied AI.

Agents need diverse, interactive worlds where they can act, fail, recover, and learn — but building these 3D environments by hand is slow and hard to scale.

Meet SimWorld Studio: a self-evolving factory for interactive 3D environments in Unreal Engine 5.

We use coding agents to generate, verify, and turn static scenes into trainable environments — taking a step toward scalable world generation for embodied agent learning.

40d4.5K5438

Lianhui Qin@Lianhuiq

🦞Vibe-coding your physical environment with SimWorld Studio feels seamless.

Our amazing coding-agent team, led by @koe_ye40329 and @haoqik322, is working hard to release the interactive playground very soon.

Stay tuned!

https://simworld.org/simworld-studio/

#AI #EmbodiedAI #ReinforcementLearning

Xiaokang(Koe) Ye@koe_ye40329

Super excited to introduce SimWorld Studio!🏭

With built-in coding agent SimCoder🦞, you can vibe-code physical 3D scenes in Unreal Engine -- then train embodied agents directly inside the generated world.🦾

From environment generation to agent training, end-to-end. 🚀

39d3.6K4021

Nicholas | WithLore@itsnicholash

AI launches today:

• Notch Agents: one prompt to a full ad workflow, from research and strategy to voice and editing.

• SimWorld Studio: self-evolving 3D environments where embodied agents can act, fail, and learn.

• Text-to-CAD: open source CAD harness and skills for Codex and Claude, including mechanism validation and STEP animations.

• Google Gemma 4 E4B + Argent: local model driving an iOS simulator through real software navigation.

• ProRata Index: usage and revenue rails for content owners whose work gets used by AI agents.

Pattern: agents are moving into production workflows now. Ads, CAD, simulated worlds, phone UIs, and rights infrastructure are all becoming agent surfaces.

Lianhui Qin@Lianhuiq

Scaling embodied AI starts with automating the environments.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D environments where agents act, fail, and learn.

With coding-agent + embodied-agent co-evolution, navigation success improves from 50% → 90%.

1/

38d6.3K1011

Yuhan (Tina) Liu@l_yuhan7272

What if the 3D environments for embodied agent learning could keep evolving? 🌍🔁🤖

In SimWorld Studio, our coding agent SimCoder 𝘀𝗲𝗹𝗳-𝗲𝘃𝗼𝗹𝘃𝗲𝘀 through verifier feedback to generate better environments, then 𝗰𝗼-𝗲𝘃𝗼𝗹𝘃𝗲𝘀 with the embodied agent to form an adaptive curriculum.

SimWorld@simworld_ai

Environment generation is the missing scaling axis for embodied AI.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn.

Env-agent co-evolvution improves navigation success 50% → 90%.

From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.

40d4.3K276

Murray Kang@haoqik322

Embodied AI is still waiting for its “environment scaling moment. Today’s embodied simulation environments are often static, manually built, or hard to scale.

Our SimWorld Studio explores a different direction with coding agents:

What if environment generation could run automatically as an agentic system?

Prompts → interactive 3D worlds Failures → verifier feedback Feedback → better tools and skills Embodied agent performance → harder adaptive curricula

Our coding agent SimCoder is self-evolving: it improves its own world-building capabilities through verifier feedback, reusable tools, and accumulated skills.

At the same time, it enables co-evolution between environments and embodied agents: agents fail in generated worlds, and those failures shape the next environments they train in.

This turns environment generation from a one-time scene generation problem into a self-improving training loop for embodied intelligence.

SimWorld@simworld_ai

Environment generation is the missing scaling axis for embodied AI.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn.

Env-agent co-evolvution improves navigation success 50% → 90%.

From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.

40d1.3K134

Lianhui Qin@Lianhuiq

Github: https://github.com/SimWorld-AI/SimWorld-Studio Website: https://simworld.org/simworld-studio/ Arxiv paper: https://arxiv.org/abs/2605.09423 Huggingface paper: https://huggingface.co/papers/2605.09423 Youtube link: https://youtu.be/WPFZPWYqcFY

Lianhui Qin@Lianhuiq

The key idea is co-evolution.

Agent succeeds → the next environment gets harder.

Agent fails → the next environment teaches what it missed.

The curriculum is no longer fixed.

Environments that don’t just get generated — Environments that grow with the agents inside them. 4/

40d39455

Drishti Regmi@DrishtiRegmi

So excited to announce + be apart of SimWorld Studio! This is only the beginning for our team, make sure to check in for our future work and give our demos on the website a quick look!

Lianhui Qin@Lianhuiq

Scaling embodied AI starts with automating the environments.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D environments where agents act, fail, and learn.

With coding-agent + embodied-agent co-evolution, navigation success improves from 50% → 90%.

1/

40d65673

James Fleming@jamesflemingcs

Super excited for the release of my first coauthor work, a lot of exciting stuff ahead with this framework!

Lianhui Qin@Lianhuiq

Scaling embodied AI starts with automating the environments.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D environments where agents act, fail, and learn.

With coding-agent + embodied-agent co-evolution, navigation success improves from 50% → 90%.

1/

40d50471

Lianhui Qin@Lianhuiq

Why environment generation?

Digital agents scale through sandboxes: code, web, computer-use.

Embodied agents need the same but in 3D worlds.

Not just static scenes, but environments with physics, interaction, tasks, rewards, observations, and standard learning interfaces.

2/

Lianhui Qin@Lianhuiq

Scaling embodied AI starts with automating the environments.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D environments where agents act, fail, and learn.

With coding-agent + embodied-agent co-evolution, navigation success improves from 50% → 90%.

1/

40d47302

Xiaokang(Koe) Ye@koe_ye40329

Check out more results in our paper and try SimWorld Studio yourself.

GitHub: https://github.com/SimWorld-AI/SimWorld Website: https://simworld.org/simworld-studio/ Paper: https://arxiv.org/abs/2605.09423

40d10722

Lianhui Qin@Lianhuiq

The key idea is co-evolution.

Agent succeeds → the next environment gets harder.

Agent fails → the next environment teaches what it missed.

The curriculum is no longer fixed.

Environments that don’t just get generated — Environments that grow with the agents inside them. 4/

Lianhui Qin@Lianhuiq

How SimWorld Studio works:

Prompt → SimCoder writes UE5 code

Code → interactive 3D environment

Verifier → checks collisions, physics, prompt alignment, task validity

Fixes → become reusable skills

Environment → Gym-style task

Agent → enters and learns 3/

40d40311

SimWorld@simworld_ai

This is the central vision of SimWorld Studio:

not just generating 3D scenes, but building a self-evolving environment factory for embodied AI.

Describe a world. Generate it. Verify it. Train an agent inside it. Use the agent’s behavior to shape the next world.

Website: https://simworld.org/simworld-studio/

SimWorld@simworld_ai

This makes co-evolution possible:

An embodied agent learns inside the world generated by SimCoder.

Its success rate, failures, and behavior traces become feedback to SimCoder.

The next world is generated based on what the agent can or cannot do.

If the agent succeeds too easily, SimCoder can make the next world harder.

If the agent fails completely, SimCoder can generate an intermediate challenge.

The curriculum is no longer fixed.

The world adapts to the agent’s capability frontier.

40d87950

Xiaokang(Koe) Ye@koe_ye40329

@abdulazeem_s Links are here: Arxiv paper: https://arxiv.org/abs/2605.09423 Github: https://github.com/SimWorld-AI/SimWorld-Studio Website: https://simworld.org/simworld-studio/

39d23211

SimWorld@simworld_ai

This makes co-evolution possible:

An embodied agent learns inside the world generated by SimCoder.

Its success rate, failures, and behavior traces become feedback to SimCoder.

The next world is generated based on what the agent can or cannot do.

If the agent succeeds too easily, SimCoder can make the next world harder.

If the agent fails completely, SimCoder can generate an intermediate challenge.

The curriculum is no longer fixed.

The world adapts to the agent’s capability frontier.

SimWorld@simworld_ai

The output is not a static 3D asset.

Each generated world can be exported as a Gym-style environment with:

reset observation action reward termination

So embodied agents can directly enter the world, interact with it, and learn from experience.

40d49530

Maitrix.org@MaitrixOrg

Environment generation is the new AI frontier

SimWorld@simworld_ai

Environment generation is the missing scaling axis for embodied AI.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn.

Env-agent co-evolvution improves navigation success 50% → 90%.

From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.

40d52720

Zhiting Hu@ZhitingHu

🔥🔥

SimWorld@simworld_ai

Environment generation is the missing scaling axis for embodied AI.

Introducing SimWorld Studio: a self-evolving factory for endless interactive 3D env where agents act, fail & learn.

Env-agent co-evolvution improves navigation success 50% → 90%.

From a prompt, our SimCoder writes code to automatically build an interactive world. Agents train inside it. And their performance shapes the next world.

40d1.1K60

Drishti Regmi@DrishtiRegmi

@simworld_ai So excited to see what future research directions are explored in SimWorld! #EmbodiedAI #MachineLearning #UCSD #Simulation

40d1944

AUTONOMOUS - July 16th@autonomousevent

@Lianhuiq Very cool approach to building out the sim environments.

How does the verifier handle physics plausibility on generated environments?

40d2702