Faster setup, cleaner workflows
This week focused on reducing friction in the places people actually feel it: connecting tools, waiting on catalog updates, and working through threads. Humm also got better at adapting to different roles and model preferences without adding more setup work.
Jira auth that behaves
You can now connect Jira through a direct Atlassian OAuth flow in the product, instead of relying on a more fragile setup path behind the scenes. We also tightened the Jira connection and query behavior, so re-auth, project-scoped searches, and comment search are more reliable once you're connected.
Catalog updates that stay live
Long-running data catalog updates now keep the UI in sync while the work is happening, rather than leaving you to refresh and guess when a job finished. That applies across generated descriptions and catalog edits, which makes larger source cleanup work much less clunky.
More roles, more model range
Humm now has built-in role prompts for marketing specialists, data analysts, operations leaders, operations specialists, and HR users. We also expanded the supported model list, so teams have more room to choose the right speed and reasoning profile for a given task.
Improvements
- Billing timeframe selections now persist as you move between billing views.
- Reports has a more useful empty state instead of a dead-end screen.
- Data catalog column counts now display correctly in the source catalog UI.
- Gong transcript access is constrained to a single transcript at a time, which makes the behavior clearer and more predictable.
Bug Fixes
- Fixed Jira Cloud introspection issues that could block or confuse setup.
- Fixed Jira JQL project scoping issues around
ORDER BYqueries. - Fixed Jira comment search to use the correct search endpoint.
- Fixed MCP connection status handling so auth state is reported more accurately in integrations UI.