Global search has been expanded to include RFQs and purchase orders, making it easier to find sourcing and procurement records quickly.
Shops can now hide built-in shared material types and categories that don't apply to their operations, keeping material lists clean and relevant. Hidden materials are also excluded from AI suggestions.
The quotes list now loads 50 rows at a time from the server, keeping the page fast even with large volumes of quotes.
You can now toggle a dimensioned bounding box overlay directly in the STEP file viewer, making it easy to verify part dimensions at a glance without leaving the quoting workflow.
The number of purchase orders shown on the home dashboard now correctly matches the count shown on the full purchasing page.
The quote pipeline and PO pipeline values on the dashboard are now correctly converted to and displayed in your company's home currency.
Header fields that you haven't changed on a quote will no longer be overwritten when fresh quote data is loaded in the background.
The job detail page will no longer disappear or show an error if a background data refresh encounters a problem.
When adding a new material type while working on a quote, it is now correctly associated with your company only.
Suppliers visiting an expired or unavailable RFQ response link will now see a clear, translated message instead of a blank or broken page.
Leftover custom material rows that no longer belong to any company are now removed, keeping the database tidy.
3D part previews were not fitting properly in portrait layouts. Models now frame correctly regardless of viewer orientation, and thumbnails are generated more reliably in the background.
The company-wide summary, pending RFQ counts, and purchasing pipeline are now calculated directly in the database, reducing load times on the home dashboard.
Quote list and quote detail pages now load more quickly by doing more processing on the server, paginating large lists, and fetching data in parallel.
Job detail pages and the jobs list are now significantly faster, with data fetched in parallel and updates handled without full page reloads.
Global search now runs multiple search strategies in parallel with an automatic fallback, returning results more quickly.
Snoozing or editing an ETA now updates only the relevant job and board data in the background instead of reloading the entire page.
The overnight ETA sweep now processes multiple jobs simultaneously and skips inactive companies, completing much faster.
The production board and purchasing requirements page no longer reload data that was already loaded on the server, making initial display instant.
Team member and subscription data are now loaded in parallel during permission checks, reducing delays on protected actions.
Multiple file signing requests made at the same time are now grouped into a single server call, reducing wait time for file previews.
The Forge AI assistant now delivers its final answer immediately and runs information-gathering steps in parallel for quicker responses.
The Google Places library now loads on demand only when an address field is present, reducing unnecessary overhead on other pages.
Importing large CSV files is now faster as rows are processed in batches of 500.
Numerous under-the-hood optimizations reduce unnecessary data fetching, improve database query efficiency, and keep UI components stable across re-renders.
Several internal security hardening measures have been applied to purchasing, search, and database functions to better enforce data isolation between companies.