๐ Version 1.11.0 is now available!
==================================
This release supercharges the Human in the Loop batch review experience โ faster triage, smarter filtering, and bulk actions at scale. Here's everything that's new!
๐งโโ Human in the Loop
๐ Review large batches in a fraction of the time with a redesigned batch review page
- Batch Preview field: add a templatable Batch Preview on the Human in the Loop node. Its resolved value shows up on every review card so reviewers can decide from a short summary without opening each task. Supports
{{variables}}, markdown, and multi-line content.
- Custom filters: declare one or more filters on the node (label + templatable value). At pause time, each task is tagged with its resolved values, and the review page renders a dropdown per filter with the distinct values from the batch. Perfect for slicing large batches by category, language, product line, etc.
- Bulk approve/reject scoped to filters: the existing ยซ Approve / Reject filtered ยป actions now honor both score filters and custom filters together, so you can act on any subset in one click.
- Item selection + bulk actions: tick individual items (or ยซ select all pending on this page ยป) and approve/reject the exact selection from a dedicated action bar independent from the filtered-bulk path.
- Grid / List layout toggle: switch between the card grid and a denser row-based list, with your preference persisted per browser. The list layout stacks scores vertically and gives more room to the preview.
- Accumulating pagination: ยซ Load more ยป now extends the visible list instead of replacing it, with an accurate remaining-items count. A ยซ N more pending not loaded ยป hint surfaces when pagination is holding items back from your selection.
- Card polish: SCORES / PREVIEW section titles, custom filter chips, better truncation of long score values, markdown rendering with proper line breaks, and a more visible ยซ Clear filters ยป affordance.
โก Plenty of bug fixes and performance improvements across the studio.