-
Suggestion
-
Resolution: Unresolved
-
Not Evaluated
-
None
-
QDS 4.7.2
-
None
Current challenge — why coding becomes the blocker
- Drag‑and‑drop is easy, logic is hard
Engineers at Japanese automakers can already place images, buttons, and gauges in Qt Design Studio (QDS) by dragging them onto the canvas.
The hurdle appears the moment they need the UI to do something—for example:- Rotate a speedometer needle when the speed value changes
- Show or hide a list when the user taps it
- Keep two properties in sync with a binding expression
- QML concepts feel foreign to non‑programmers
These engineers are domain experts, not coders. property alias, signal handlers, JavaScript snippets, and state machines look like "real programming," so they slow down or abandon the prototype and ask the Tier‑1 supplier to fill the gap. - Result: "living specifications" are neither fast nor frictionless
Instead of handing a ready‑to‑run prototype to the supplier, the OEM must attach written instructions or sketches and hope the Tier‑1 interprets them correctly—adding cost and rework later.
Proposed solution — AI that fills the coding gap
| Step | What the engineer does | What the AI assistant in QDS does |
|---|---|---|
| 1 | Designs the static layout as usual with drag‑and‑drop. | - |
| 2 | Opens an "AI Prompt" panel and types a natural‑language request, e.g. "Bind the speedometer needle to the speed value." | Finds out what is "speedometer needle" and "speed value" in QML by scanning the entire project and binds them. |
| 3 | Types another request, e.g. "Make this list collapse by default and expand like a dropdown when clicked." | Adds the necessary states, transitions, and click handler, or replaces the ListView with a ready‑made dropdown component, if that makes sense. |
| 4 | Runs QDS Preview to confirm everything moves as expected—without touching any code. | Generates clean, commented QML/JS files and tags them with the original prompt for traceability. |
| 5 | Exports the project bundle and sends it to the Tier‑1 supplier. | - |
Key benefits
- No coding required: bindings, signals, and small scripts are produced automatically.
- Faster feedback loop: engineers can verify behavior immediately in the preview.
- Reduced miscommunication: suppliers receive an executable prototype, not an abstract spec.
- Traceable: each AI‑generated block is labeled, so suppliers can see exactly what was auto‑generated and why.
Practical considerations
- Language support: prompts in Japanese or English. Japanese support is Must have for the customer.
- Performance: aim for sub‑2‑second responses to keep the workflow fluid.
- Offline fallback: allow on‑prem models or cached templates if the workstation is isolated.
- Version control: generated files and codes live in the normal QDS project tree, so Git diff works as usual.
In short: the feature turns free‑form, human language into the bits of QML code that non‑programmers currently find impossible, letting them finish a truly "living" specification on their own before they pass it to the Tier‑1 team.