Rescale Brings Autonomous Agents to Simulation Engineering
Rescale has released three new capabilities designed to automate the manual work that slows product development teams: autonomous agents for simulation workflows, AI-ready surrogate models, and cost controls for cloud computing resources.
The updates address a specific productivity bottleneck. In simulation-driven engineering, experts spend hours on repetitive tasks-validating inputs, diagnosing failed runs, generating reports, configuring iterations. Autonomous agents now handle this work while keeping engineers in control through checkpoints and approval gates.
Agents for Simulation Work
Rescale's agentic digital engineering uses prebuilt agents for common simulation tasks: input validation, job failure diagnosis, hardware configuration, and report generation. Engineers can assemble these into multi-step workflows that run automatically and notify teams at defined decision points rather than requiring constant monitoring.
The platform enforces security and governance. Agents operate under explicit autonomy levels-from recommending actions to executing within guardrails-allowing teams to expand automation incrementally. Execution ties to existing access controls and organization policies.
McLaren Automotive, an early adopter, reports a 3x productivity boost for engineers evaluating design iterations. The company runs agents on a platform trained on its own engineering data, allowing it to evaluate thousands of design variations across multiple physics domains in hours.
AI Physics Operating System
Surrogate modeling has been on engineering roadmaps for years. The practical barrier isn't understanding the technology-it's converting raw simulation data into production-ready models that engineering teams actually trust and use.
Rescale's AI Physics operating system handles the full pipeline in one environment: data management, model training, versioning, and deployment. Engineers get access to model architectures including NVIDIA PhysicsNeMo without deep machine learning expertise.
The system embeds surrogate model inference directly into Autodesk Alias and Blender, allowing design teams to evaluate thousands of variations in their native tools. Iteration compresses from days or weeks to near real-time.
Customers report measurable gains: 4x more design candidates evaluated per cycle, 30x cost-efficiency improvement, and 60% faster product development.
Compute Cost Control
Cloud HPC gives engineering teams access to enormous computational scale. Without controls, that scale creates cost exposure and hardware selection friction that consumes engineering time.
Rescale's compute economics layer adds automated, rules-based controls and real-time visibility. Organization-wide cost controls let leaders make transparent tradeoffs between cost savings and resource availability without manual configuration. Curated hardware groupings deliver equal or better performance than individual core-type selections, removing benchmarking work from engineers.
Platform Extensions
The release also includes a Data Fabric that connects engineering knowledge across systems-SharePoint, AWS S3, Azure Blob, and other platforms-to power AI-assisted decisions. A new Workflow Builder gives simulation managers a drag-and-drop canvas to encode validated processes into repeatable templates with structured data handoffs between steps.
Daikin Industries, which manufactures HVAC and industrial equipment, has deployed Rescale across R&D sites and is advancing toward broader agentic capabilities across its global organization. "What excites us is how directly Rescale's capabilities align with our vision for the future of industrial manufacturing," said Satoru Takanezawa, Senior Engineer and Group Leader at Daikin's Technology and Innovation Center.
To see these capabilities in action, visit Rescale's Spring 2026 Showcase.
Your membership also unlocks:
