DataMesh Launches Executable Digital Twins to Train Industrial AI

SINGAPORE – January 15, 2026 – Digital twin technology provider DataMesh today announced the launch of DataMesh Robotics, a new platform designed to bridge the critical gap between virtual simulations and the chaotic reality of industrial environments. The solution introduces an "Executable Industrial Digital Twin," a dynamic virtual world that allows artificial intelligence systems to learn and adapt to complex, evolving factory and facility conditions before being deployed in the physical world.

As industries increasingly turn to robotics and embodied AI—intelligent systems that can perceive and act within their environment—they face a significant hurdle: training these systems effectively. Robots that perform flawlessly in clean, static simulations often fail when confronted with the unpredictable nature of a real factory floor. DataMesh Robotics aims to solve this by creating training environments that don't just look real but act real.

"At the heart of industrial embodied AI is the need for a training world that changes like the real world," said Jie Li, CEO of DataMesh, in the announcement. "We go beyond industrial-grade scenes and synthetic data by making the environment executable — so processes evolve, events are triggered, and task objectives become explicit and measurable. DataMesh Robotics aims to become the industrial training environment and data engine for robotics teams."

From Static Visualization to Executable Worlds

For years, the concept of a "digital twin" has primarily involved creating a static 3D replica of a physical asset, overlaid with real-time sensor data. While useful for monitoring and visualization, these models lack the dynamic behavior needed to train an AI for complex, multi-step tasks. An AI learning to navigate a warehouse needs to understand not just the layout, but how that layout changes with moving forklifts, shifting inventory, and human workers.

DataMesh's Executable Digital Twin, built on its FactVerse platform, represents a significant evolution. These virtual environments are not merely passive models; they are active simulations where industrial logic is enforced. In this framework:

• Processes Evolve: Manufacturing, inspection, and maintenance workflows unfold over time, forcing the AI to learn sequences and adapt to changing operational phases.
• Events Trigger Reactions: Alarms, equipment state changes, and task transitions can be triggered, training the AI to respond to unexpected but plausible events.
• Objects Interact: Virtual machinery and components can move and interact based on simulated physics and pre-defined business logic.

This dynamic capability allows the platform to generate training data that more accurately reflects the complexities of real-world industrial operations, including scenarios with safety constraints and partial observability, where the robot doesn't have a complete view of its surroundings. While competitors like Siemens and Dassault Systèmes offer powerful digital twin ecosystems, and game engines like Unity and Unreal Engine provide photorealistic simulation, DataMesh is carving out a specific niche by positioning its solution as a dedicated data product for embodied AI training.

Solving the Industrial Data Bottleneck

One of the most significant challenges in developing industrial AI is the scarcity of relevant training data. Real-world data from factories is often difficult and expensive to collect, and it may not cover the full range of potential failures or hazardous situations needed to create a robust AI. DataMesh Robotics addresses this by focusing on the generation of high-quality, industrial-grade synthetic data.

The platform provides an end-to-end stack for modeling industrial scenes, simulating physics, and producing scalable synthetic data. This includes not only visual data for perception tasks but also non-visual variables like process states, temperature fluctuations, and pressure readings, giving the AI a richer understanding of its operational context. All data is generated with automated ground-truth labeling, drastically reducing the time and cost associated with manual data annotation.

Furthermore, the solution tackles a notoriously difficult aspect of AI development: defining task objectives and reward signals. In an industrial setting, a task's success is not a simple binary outcome. It often involves adhering to strict tolerances, following precise sequential workflows, and prioritizing safety above all else. Designing a reward function that captures these nuances is complex and prone to error. DataMesh Robotics offers a configuration-driven approach, allowing developers to define goals, success conditions, and penalty structures through a more intuitive interface, leading to clearer training objectives and more stable learning for the AI.

Integrating into the Modern Robotics Ecosystem

Recognizing that robotics development relies on a complex toolchain, DataMesh has designed its new platform for seamless integration with mainstream ecosystems. The solution supports the export of its executable digital twin assets and synthetic data to leading robotics simulation environments, including NVIDIA Isaac Sim and Omniverse.

This integration is a strategic move that allows robotics teams to incorporate DataMesh's dynamic, data-rich environments into their existing workflows without starting from scratch. By connecting with NVIDIA's powerful platforms for physically-based rendering and physics-accurate simulation, users can enhance the fidelity of their training scenarios. This approach helps solve common interoperability challenges and allows developers to leverage their existing investments in the NVIDIA ecosystem, accelerating the entire research and deployment pipeline.

The solution is built for enterprise-level demands, supporting on-premises, private cloud, and hybrid deployments to meet varying security and infrastructure requirements. This flexibility, combined with enterprise-grade governance, makes it a viable option for large-scale industrial applications.

Paving the Way for the Autonomous Factory

With this launch, DataMesh is contributing to the broader vision of Industry 4.0 and the fully autonomous factory. The ability to train more intelligent, adaptable, and reliable robots has profound implications for manufacturing, logistics, and facilities management. Use cases range from complex workstation operations and navigation in busy warehouses to remote inspection and maintenance in hazardous or restricted environments where human safety is a primary concern.

The company's credibility is bolstered by its recognition in multiple Gartner research reports for its work in intelligent simulation and spatial digital twin technologies. The new Robotics solution has already moved beyond the conceptual stage, with prototype validation complete and pilot projects now running with enterprise partners, including telecom operators and data labeling providers.

As DataMesh continues to expand its library of industrial assets, task templates, and ecosystem integrations, the platform stands to become a critical enabling technology. By providing a safer, faster, and more effective way to train the next generation of industrial AI, it helps de-risk the transition to autonomous operations and brings the promise of smarter, more efficient, and safer factories one step closer to reality.