Commands, Reports, and Follow-Through
This week tightened three parts of Humm that people actually touch: commands, scheduled reports, and thread follow-ups. You should see more context around saved commands, fewer report workflow failures, and less UI weirdness when you ask Humm to keep going.
Commands With Real History
Saved commands now show more of their own story. You can see who created a command, how often it has been used, which version is current, and what changed over time.
Reports That Connect Better
Reports and commands are linked more cleanly now. You can jump from a report row straight into the related command, and report tool output now uses the same left-column and side-panel layout as threads, which makes inspection less awkward.
Scheduling That Works In One Pass
Creating a scheduled command is less fiddly. You can add subscribers before the first save, and the scheduling dialog now stays usable on shorter screens instead of pushing the footer off the page.
Follow-Ups That Don’t Lose Their Place
Follow-up questions in threads now hold onto their optimistic state more reliably while the backend catches up. In practice, that means the reasoning block appears immediately and the thread is less likely to snap back to an older state while Humm is still working.
Improvements
- Analyst users no longer see organization settings they cannot actually use.
- Follow-up suggestion chips now come from backend suggestions only, which should make them feel less random.
Bug Fixes
- Fixed scheduled report runs that could fail or finalize incorrectly.
- Fixed report comparison event dispatch that could duplicate finalization work behind the scenes.
- Fixed thread follow-up polling races that could kill optimistic UI state until refresh.
- Fixed report tool-panel sizing so the main content and side panel behave more like the standard thread layout.
- Fixed command scheduling modal sizing on short viewports.