Skip to main content
Back to timeline
Navier AISource publication:

Navier AI's Prism puts an engineering agent on simulation setup, running OpenFOAM, SU2 and CalculiX locally

Synopsis

Navier AI has released Prism, a desktop app in which an engineering agent handles simulation setup and execution. Commercial desktop suites are built around a dedicated simulation specialist and usually license each analysis type separately, while open-source solvers require each case to be built by hand in text files and scripts, so most teams test only a handful of design variants. Prism runs OpenFOAM, SU2, CalculiX and Navier AI's own FEA solver Ferro on the user's machine, with the agent building geometry, setting up the case, running it, checking results and writing the report, while every input file stays in the project to open and rerun.

Source-provided article image: Simulation Software for Hardware Teams
Navier AI

Interpretation

The time-consuming part of simulation shifts from the solve to setup, meshing and reporting, and an engineering agent is placed on that chain to carry those steps. Commercial desktop suites are built around a dedicated simulation specialist and usually license each analysis type separately, while open-source solvers require users to build each case by hand in text files and scripts. Prism is described as running OpenFOAM, SU2, CalculiX and Navier AI's own FEA solver Ferro on the user's machine, with the agent covering the chain from geometry to report.

The user describes the problem, and the agent builds the geometry, sets up the case, runs the solve, checks the result and writes the report. The agent drives established physics solvers rather than replacing the physics, placing it in the class of AI simulation tools that use an agent to set up and run real solvers. Every input file stays in the user's project, where it can be opened and rerun to audit what the agent set up.

Perspective

Prism is a desktop app for hardware teams that runs OpenFOAM, SU2, CalculiX and Ferro on the user's own machine and covers common physics such as fluids, structures and thermal; the text points to validation studies against wind-tunnel and towing-tank measurements but gives no specific comparison results.

The text does not state how accurately the agent sets up cases, how much human intervention is needed, or the specific figures from the validation studies; whether the agent's result checks are reliable still requires the reader to inspect the retained input files.