Workflow for AI-Assisted Model Generation

QuickField uses AI to help prepare simulation models, while model creation, verification, field calculation, and further editing remain transparent and under human control.

Step 1. Prepare the AI Assistant

Upload the QuickField Skills Archive to your AI assistant and instruct it to use the archive when generating QuickField Builder metafiles.

The Skills Archive contains documentation, examples, tests, and implementation details describing the .qfp, .qfs, and .qfd JSON metafiles used by QuickField Builders.

Step 2. Describe the Problem

Describe the required simulation model in natural language.

Example:

Prepare a DC magnetic analysis model of an air-core solenoid with external diameter 20 mm, internal diameter 10 mm, length 100 mm, and current 100 A.

The AI assistant generates the corresponding JSON metafiles:

.qfp — problem definition;
.qfs — geometry model;
.qfd — materials, sources, and boundary conditions.

Step 3. Optionally Review the Generated Metafiles

Before calculation, you may review the generated JSON files and check whether the geometry, materials, sources, boundary conditions, and analysis settings match the intended problem formulation.

Step 4. Create the QuickField Problem

Place all related JSON metafiles (.qfp, .qfs, and .qfd) in the same directory. Then run QuickField Problem Builder and open the .qfp file using the standard Windows file dialog, or specify the .qfp file as a command-line parameter.

problem_builder.exe solenoid.qfp

Problem Builder converts the JSON metafiles into native QuickField files using the QuickField API .

Step 5. Open and Continue Working

Depending on the .qfp settings, Problem Builder may also start the solution automatically.

You can then open and analyze the generated QuickField problem, continue working with it, or adjust it using the regular QuickField user interface or any automation methods, depending on user needs and preferences.

Human Supervision and Intervention

Human supervision and intervention are possible at every stage:

defining the task in natural language;
reviewing or editing the generated JSON metafiles;
inspecting or adjusting the generated QuickField model;

This allows AI to assist model preparation while keeping the model generation workflow transparent and controllable.

After the model has been generated, it becomes a regular QuickField problem and may be modified, solved, postprocessed, or automated using standard QuickField workflows.

For practical examples of this workflow, see AI Use Cases.