MotorAI
JPEN
Menu

Release Notes

v1.0.1

Bug fixes

  • JMAG-Designer detectionFixed an issue where JMAG-Designer installed on a drive other than C: or in a non-standard folder was not found, making JMAG unavailable. It is now detected automatically without any path configuration.

  • License switching when sendingFixed an issue where the only option was to quit the app when sending a message while the license was in use on another device. You can now choose "Use on this device" and continue sending.

  • Local server left running after exitFixed an issue where the background local server kept running after the app was closed due to a license check.

v1.0.0

New features

  • Multiple-choice question cardsWhen the AI needs a decision from you, it now asks through a question card with options. You can type your own answer under "Other" if none of the options fit.

  • Task plan (to-do list) displayThe AI's work plan is now shown as a to-do list, so you can follow the progress and elapsed time of each step.

  • Progress explanations in the conversationThe approach and findings the AI writes between steps now appear in the conversation as they are written. The summary right before a question card stays visible, so you can see why the AI is asking.

  • Plan / Manual / Auto run modesYou can now switch between the Plan, Manual and Auto run modes before sending or while a task is running. Plan only investigates and proposes a plan, Manual asks before every change, and Auto asks only before irreversible or long-running operations.

  • Clearer approval cardsApproval cards now state what will be done to what, such as "Delete the study 'Case3'". For irreversible or long-running operations, the card also shows the impact.

  • Choosing a skill with "/"Type "/" at the start of the input box to see the list of skills and pick the one to use for your request.

  • Creating your own skills through chatCreate your own skills just by asking in chat. They can be used from the next message, without restarting the app.

  • Additional instructions for the AISettings now include "Additional instructions for the AI", where you can register rules the AI should always follow, such as which units to use. A ready-to-use example is filled in by default.

  • English displayThe display language can now be set to Japanese or English. The AI also responds in that language, or in the language you write in if it differs.

  • Image pasting and zoomPaste images into the chat with Ctrl+V or drag and drop. Use Ctrl+"+"/"-" to zoom the whole window, and zoom the conversation history and previews separately.

  • Motor characteristics and efficiency mapsThe AI can now drive the analysis software on its own to measure dq parameters and derive motor parameters, speed-torque characteristics and efficiency maps.

  • Baseline model proposal from a specificationStarting from a specification (JSON, PDF or text pasted into the chat), the AI can now take you through the feasibility check and desk sizing to a baseline JMAG file proposal with a minimum of analyses.

  • Conversion and grafting for JMAG ExpressJMAG files built outside JMAG Express can now be converted to the standard JMAG Express structure while keeping their geometry. Template grafting supports PMSM, SynRM, squirrel-cage induction and wound-field synchronous machines with any pole and slot count, and broken geometry is repaired automatically where possible.

  • Batch runs for analysis group optimizationAnalysis groups created in the JMAG-Designer GUI can now be reconfigured in place and optimized as a batch run on the scheduler.

  • 3D analysis of axial flux motorsFor axial flux motors (one rotor, two stators), you can now go from model creation and material setup to a 3D analysis run and torque report through chat.

Improvements

  • Refreshed layout and designThe layout and icons have been redesigned, for example with category-based Settings and the file tree moved to the right edge. A disabled "New conversation" button now appears dimmed, and image previews can be zoomed and dragged freely.

  • Faster display after sendingYour message now appears as soon as you send it, and "Preparing" is shown until the AI starts responding. Your text is no longer lost if sending fails.

  • Faster JMAG startupJMAG-Designer now starts up significantly faster.

  • Lower token usageWe revised how the analysis AI loads its instructions, cutting token usage by about 20%.

  • Problem reports for analysis group optimizationAnalysis group optimization now reports when Range Finder collapses a design variable's range or stops with an error.

  • Expanded motor theory databaseThe motor theory knowledge the AI refers to has been expanded.

  • Expanded motor patent databaseThe motor patent knowledge the AI refers to has been expanded.

  • LLM connection settingsUnder Connection in Settings, choose between the MotorAI default endpoint and your own endpoint, such as an in-house LLM. Pick a model from the suggestions or type one in, and the connection is checked on save when you change the endpoint.

Bug fixes

  • Garbled text in CSV previewFixed an issue where UTF-8 and Shift-JIS files were not displayed correctly in the CSV preview. The character encoding is now detected automatically.

  • Cluttered answersFixed an issue where one request produced a stack of answer bubbles, and an issue where in-progress notes were mixed into the final answer.

  • Error shown after a successful responseFixed an issue where an error message appeared in the conversation even though the response had completed normally.

  • Sending stuck after an errorFixed an issue where, after the server returned an error, the next message got stuck.

  • Previous server left runningFixed an issue where, after using JMAG and restarting the app, the previous local server and JMAG kept running in the background.

  • Analysis group optimization not progressingFixed an issue where analysis group optimization was configured successfully but never ran, producing zero individuals, and an issue where a run was wrongly judged complete partway through.

  • Windows opening repeatedlyFixed an issue where many console windows opened during JMAG operations, and an issue where the JMAG-Designer selection dialog kept appearing.

  • Missing characters in reportsFixed an issue where Greek letters, symbols and Japanese text were dropped or replaced in PDF and Word reports, and an issue where Korean and Chinese characters were not displayed.

  • Effect of Python environment variablesFixed an issue where Python run by the AI did not work correctly when PYTHONHOME or PYTHONPATH was set on the PC.

  • Chat screen issuesFixed issues where chat text could not be selected or copied, tables and equations were not copied correctly, line breaks shifted the input position, and previewing a malformed CSV closed the app.

  • Equations in reportsFixed an issue where equations were not rendered correctly in PDF and Word reports.

v0.4.5

New features

  • Expanded JMAG Express integrationCreate a new motor JPROJ file from specifications using a template, or replace the rotor and stator through chat instructions. Analysis execution is supported for outer- and inner-rotor PMSMs, SynRMs, induction motors, and wound-field motors.

  • Zoom controls for conversation history and previewsConversation history and preview screens can now be enlarged or reduced. Image previews can be zoomed and dragged freely instead of simply following the screen size.

  • Review and edit all JMAG analysis model settingsMore settings can now be reviewed and changed, including parts, materials, windings, analysis conditions, circuits, meshes, and results. When necessary, the agent can consult JMAG Help to perform a task or prepare an answer.

  • Multi-study optimization with analysis groupsConfigure and run optimization across multiple analysis studies, then review results for each candidate, entirely through chat.

  • Japanese and English displayThe application interface can now be switched between Japanese and English.

Improvements

  • Expanded motor theory databaseThe motor-engineering knowledge base has been expanded, with particular emphasis on design theory.

  • Faster response startSent messages appear immediately, and the preparation time before a new turn begins has been significantly reduced.

  • Clearer progress messagesProcessing status is now shown with clear user-facing descriptions instead of internal terminology.

  • Improved settings screenSettings are now organized and displayed by category.

  • Much faster case configurationCSV loading for case settings and design tables, which previously could take tens of minutes, now completes within a few minutes.

  • Shorter waits after analysis failureIf a solver stops during analysis, the application now detects it promptly and terminates the process instead of continuing to wait.

Bug fixes

  • Project closing when generation was stoppedFixed an issue where stopping response generation could also close the open JMAG project and cause work to be lost.

  • Processes shown as complete while still runningFixed cases where long-running operations were shown as complete before finishing or the conversation stopped midway.

  • CSV encoding and crashes in Japanese environmentsFixed an issue where case-setting CSV files became garbled in Shift-JIS JMAG environments and could cause the application to close unexpectedly.

  • Selecting and copying chat textText in the chat can now be selected with the mouse and copied.

  • Crashes when previewing malformed CSV filesMalformed CSV files no longer cause the application to close unexpectedly. The preview is stopped and the affected row number is displayed.

  • Geometry edits that did not completeFixed operations that could fail to finish after changes such as modifying dimensions following a hole operation, as well as cases where holes disappeared during later edits.

  • Settings lost when applying templatesFixed cases where applying an analysis template could remove or reset materials, magnetization directions, current, rotational speed, and other analysis conditions.

  • Unneeded design variables remaining in optimizationFixed an issue where variables excluded from optimization could remain and cause an execution error.

  • Geometry restoration after pole or slot count changesFixed model corruption during geometry restoration after changing pole or slot counts. When geometry cannot be restored automatically, the application now asks the user how to proceed.

  • Agent execution errors in the production applicationFixed issues that prevented some functions from running or caused processing to stop responding.

Earlier release notes are currently available on the Japanese release notes page.