Download the Onshape Mobile App

IOS Android
Illustration with an overlay showing a pumpkin as a 3D model and undergoing a simulation analysis.
READ TIME:
05:55

Summary

  • Onshape and SimScale work together with a tight integration built on cloud-native architectures.
  • With a cloud foundation, AI tools can be better leveraged for designing and for validation.
  • The FeatureScript MCP server from Onshape Labs and the SimScale Engineering AI Agent accelerate design-to-analysis loops.

It can take weeks for validated simulation results to return to the designer. AI is changing that.

With Onshape and SimScale, an Onshape partner, design and validation now happen in one loop. Product developers can create a parametric CAD model from a prompt, then use an AI agent to simulate and test it, all in the same cloud-native environment. The integration between cloud-native CAD and simulation tools creates a tighter loop between design and validation.

The Problem Between Design and Analysis

In a typical workflow, validating a design might take weeks. Results can be held up by a simulation expert’s availability, access to CAE licenses, and the work required to prepare a CAD model for analysis.

For hardware engineers, the design may continue moving forward while they wait, increasing the risk that problems are discovered later, when the cost of change is higher.

This problem area can be addressed by closing the gap between design and analysis. Cloud-native tools and AI are making it possible.

Onshape is a cloud-native CAD and PDM system. The FeatureScript MCP server, from Onshape Labs, lets a prompt generate editable, parametric CAD. Built-in simulation capabilities cover quick checks on that model.

But when the question needs deeper analysis, SimScale runs the full study. SimScale provides cloud-native simulation across CFD, FEA, thermal, electromagnetics, and other physics. Its Engineering AI Agent can set up simulations, run optimization studies, interpret results, and produce reports from a prompt.

Together, they create a connected design and simulation workflow, allowing engineers to evaluate design decisions earlier, while the cost of change is still low.

Between them, the pieces fit:

  • Associativity: When the design changes in Onshape, the simulation setup carries forward, and most parameters stay the same
  • One agent, two tools: The SimScale Engineering AI Agent is available in both Onshape and SimScale, with the same chat history and messages mirrored between them.
  • Versions and branches: Every study ties back to a saved Onshape version
  • Open to automation: Both tools expose APIs, so the workflow can also run headless

How would this look in practice?

From Prompt to Parametric CAD in Onshape

It all starts with the design. The FeatureScript MCP server from Onshape Labs, connects an LLM to FeatureScript, the programming language behind Onshape’s native features. A plain-language prompt generates FeatureScript code that can become a native, parametric feature.

The parameters that matter, such as dimensions, counts, and clearances, can be exposed as configurations. Specifying the parameters of interest in the prompt makes them easier to find and adjust afterward. Those configurations become the design space for everything that follows.

Why code instead of asking AI to do the job? Building robust parametric CAD requires reliable references to geometry such as faces, edges, and vertices. AI is already effective at generating code, and FeatureScript gives it a structured way to create those references while preserving a parametric feature tree.

Onshape holds text-code-to-CAD workflows to three rules:

  1. Humans stay in the loop to verify and edit what the AI creates.
  2. Results should be deterministic and preserve engineering intent.
  3. The result lives in the CAD system where engineers work.

Once created, the feature behaves like any other parametric feature in Onshape. It can be shared across a company, and reusing the feature does not require generating it again.

The next question is whether the model actually works.

SimScale’s Integrated AI Agent Runs the Simulation

That is where SimScale comes in. SimScale is a cloud-native simulation platform for CFD, FEA, thermal, and electromagnetic analysis, and it connects to Onshape through a separate integration in the Onshape App Store. The Engineering AI Agent comes with it. The same agent that runs inside SimScale is enabled in Onshape and opens in a panel next to the model.

A prompt describes the goal, such as which parameters to vary, what to optimize, and what constraints apply. The agent can use custom materials saved in the SimScale library, apply saved views, and produce a PDF report at the end.

From there, the agent handles much of the simulation setup traditionally performed by a specialist. It proposes the physics, prepares geometry that isn’t yet simulation-ready, meshes the model, and chooses boundary conditions. For details not specified in the prompt, the agent determines the setup based on the study. Before any compute is spent, it asks for approval of a simulation budget.

For optimization studies, the agent works in batches. It runs a first set of designs across the design space, evaluates the results, and determines the next set of simulations based on the results. Studies can vary CAD parameters, such as geometry, as well as simulation parameters, such as materials or operating conditions.

Engineers can step in at any point to adjust a setup, override a boundary condition, or review assumptions, then hand control back. They can also leave the agent to continue working and return later to review the completed report. The matching SimScale project keeps the full trail: parameter bounds, each batch of simulations, and how the agent moved from one round to the next.

The Results Land Back in Onshape

The final report summarizes the design space, simulation results, parameter influence, assumptions, and recommended configuration, giving the engineer a record of how the optimization reached its result.

Back in Onshape, the simulation can be documented alongside the design. The winning configuration is set on the model, the PDF report is imported into the same document, and a version is saved. That creates a traceable record connecting the design configuration, optimization study, and resulting engineering decision.

From there, a branch can be created for the next design question, and the agent picks up in the same panel to run the next study. Each study builds on the last, with the history intact.

Instead of waiting weeks for the design-to-analysis loop to close, engineers can move between design, simulation, and iteration within the same connected workflow.

The Loop in Action

The demonstration below walks through the workflow for optimizing a parameterized pumpkin lantern. The study varied candle height, flame height, and wax chemistry to evaluate their effect on the pumpkin lid temperature. After an initial optimization, the team ran another study to determine the tallest candle that could be used while keeping the lid below 100°C.

Put the Workflow into Practice

AI is changing how products are designed and validated. With Onshape and SimScale, AI can support a connected loop between design and simulation: a prompt to help create parametric CAD, an agent to explore the design space through simulation, and Onshape versions to preserve the resulting design decisions.

SimScale and the FeatureScript MCP server are both available through the Onshape App Store. New Onshape users get six months of SimScale’s professional platform through the Discovery program.

The Onshape Discovery Program

Learn how qualified CAD professionals can get Onshape Professional for up to 6 months – at no cost!

Latest Content