AI
AI Assistant
Use natural language to modify models, run simulations, and optimize in the web app
This guide introduces the AI assistant built into the web application. Use natural language to modify structures and optical parameters, run simulations and optimizations, and read results.
Open the AI Assistant
Click the floating button at the bottom-right of any page to open the chat panel on the right; click it again, or the close button, to hide it. Then type what you want, for example:
Add a 100 nm SiO2 layer at the endSweep wavelengths 400-800 nm with R / T / A enabledRun an optimization
What it can do
You can ask the AI Assistant to:
- Add, edit, and remove layers, layer groups, and surrounding media;
- Set wavelength, angle, polarization, detectors, incident spectrum, and color;
- Choose material indices (database / constant / custom);
- Configure and run parameter sweeps and optimizations;
- Run calculations and read results.
For irreversible actions such as removing a layer or clearing the structure, the assistant asks for confirmation before applying the change.
Limits
| Limit | Detail |
|---|---|
| Single-thread chat | One linear conversation; no branching, regenerate, or undo |
| No cross-session memory | Refreshing the page starts a new session; past conversations are not read |
| Text only | Accepts text instructions; no file or image upload |
| Built for interactive, step-by-step work | Designed for adjusting a model while you watch it in the browser — not for running large batches of simulations unattended |
AI Assistant or Simulation MCP?
| Your situation | Use |
|---|---|
| Working in the web app and want the AI to help edit the model and run one-off simulations | AI Assistant |
| Driving large, automated batches of modeling / sweeps / optimization from your own AI agent (e.g. Claude Code) | Simulation MCP |
Next
- Connect an external AI tool to build and simulate automatically: Simulation MCP
- Externalize state:
.tmm.jsonfile / URL sharing