SimPilot uses a tool-driven execution loop behind a conversational interface.
Step 1: Define the requirement
Describe the scenario in plain language. Include whatever you know:
- Geometry and dimensions
- Physics and operating conditions
- Solver preferences (optional)
- Desired outputs or decisions
You can also attach files (geometry, PDFs, images, data tables, archives) to ground the request.
Step 2: Agent plans and executes
The agent turns your prompt into a runnable workflow, typically including:
Research and retrieval
Pull relevant references from internal knowledge and, when needed, external search tools.
Case and mesh setup
Create directory/file structure, generate config files, and prepare meshing inputs.
Execution
Submit simulation runs through the configured runtime path and stream progress.
Failure handling
Parse logs, diagnose problems, rewrite targeted files, and retry within configured limits.
Analysis preparation
Extract metrics, prepare artifacts, and expose outputs for follow-up questions.
Step 3: Analyze and iterate
When results are available, you can:
- Review structured summaries and artifacts
- Open interactive 3D visualization
- Compare runs and run parameter sweeps
- Generate exportable reports
- Branch the conversation and test alternatives
Continuous loop
SimPilot is iterative by design. The output of one run becomes context for the next question, so you can move from rough idea to refined decision in one conversation thread.
Be explicit about success criteria (for example target pressure drop, drag reduction, or thermal limit). Clear criteria help the agent choose better defaults and comparisons.