Rescale puts engineering agents to work on Hermes 2 reactor

By Amara Osei

Rescale disclosed on November 10, 2026, that Kairos Power and Oak Ridge National Laboratory’s Manufacturing Demonstration Facility use its platform for Hermes 2, a 50-megawatt fluoride-salt-cooled reactor. The company introduced a framework for customizing engineering AI agents as the teams test advanced manufacturing and construction methods with support from Rescale AI developer engineers.

For Hermes 2, Rescale said agents ran thermal, flow and shielding simulations and produced related reporting on reactor-cavity and bioshield-plug designs. An engineer approved every agent action, keeping the work under human sign-off.

Rescale said Oak Ridge National Laboratory is encoding its simulation and additive-manufacturing expertise in agents. The aim is to let engineers beyond the facility adopt those methods.

Kairos Power projects that agentic workflows built from past simulations could raise end-to-end simulation-analysis efficiency by up to 3x. The company said the workflows must retain the traceability needed for regulatory compliance and described the efficiency target as prospective.

What does the agent control layer do?

Rescale’s Fall 2026 Showcase account makes Agentic Projects the control layer for Edison, its AI assistant and agent orchestrator. Teams can set objectives, tools, skills and guardrails, then add agent nodes to validate inputs, analyze results or prepare evidence.

According to Rescale, projects can be saved and shared across programs. A suggested learning enters project memory only after an engineer accepts it.

Rescale’s Engineering Intelligence feature links requirements, design, simulation, test and process records in knowledge graphs. The graphs give agents the relationships and configurations needed for complex work, while a separate work graph maps people, data and tasks across a workflow.

That work graph lets engineers trace how work was performed and where agents participated, according to Rescale. The November 10 event did not introduce every component for the first time.

What had shipped before the event?

Rescale’s Agentic Projects release note, dated October 26, 2026, describes validated baseline jobs, linked standards and files, reusable skills, and citations when Edison answers from a linked document. Projects start private, can be shared with a workspace and require Agentic Engineering to be enabled.

A November 5, 2026 release note introduced Rescale Desktop for macOS on Apple silicon, Windows x64 and Linux x64. The app moves files between local and cloud environments, tracks jobs and provides Edison chat without a browser.

According to Rescale, Edison searches only user-approved folders. It can see file names, sizes and dates, but not file contents, and it asks before every transfer.

Desktop chat requires Advanced Modeling & Simulation or Agentic Engineering. Rescale says the feature is not yet available on Rescale ITAR.

How do external AI clients connect?

Rescale expanded its Model Context Protocol connectors so outside AI clients can issue natural-language requests to the platform. The company says the connectors expose specialized agents and more than 1,350 pre-installed engineering applications in Claude and other MCP-compatible tools. Rescale MCP is also listed in the Claude and Codex directories.

Rescale’s MCP documentation limits that route to interactive use by a person working through an agent. Each connection is bound to one workspace and the signed-in user’s existing permissions, with actions attributable to that user. Rescale directs production integrations to its REST API or command-line interface.

What changes in simulation work?

Rescale’s AI physics additions address surrogate models, which are AI approximations trained from simulation data. A dataset-creation agent turns existing simulation data into training-ready datasets.

Guided pre-processing identifies outliers and coverage gaps and recommends data splits, according to Rescale. Auto Tune adjusts model configurations during training.

Rescale says Design Explorer provides a guided local interface for evaluating thousands of design options against constraints set by simulation engineers. It can link multiple surrogate models into multiphysics optimizations.

Uncertainty quantification adds confidence and out-of-distribution signals to each prediction, Rescale said. The tool identifies results that need a full-fidelity simulation check and cases where the model needs expansion.

The releases described by Rescale keep three controls explicit: suggested learnings need engineer approval, Hermes 2 agents require human sign-off, and uncertain surrogate predictions are flagged for full-fidelity checks. The Hermes 2 collaboration will test whether reusable agent workflows can reduce analysis effort while preserving an auditable engineering record for regulatory review.