Improvements
- Dashboards
- Stat and gauge tiles can now go much smaller without losing legibility, so you can fit more on a single dashboard
- Reduced whitespace across the dashboard for better density
- Discover Services: Performance History loads noticeably faster, even on busy services
- Discover Applications (RUM)
- Web Vitals (LCP, CLS, INP, FCP) come with attribution — e.g., a poor LCP now tells you which element on the page was the largest contentful paint, and a CLS regression points at the specific element that shifted. Learn more
- Pages that block the main thread are tracked automatically, with the worst offenders surfaced per session
- Sessions now capture a much richer view of how a page actually loads — from the moment the request starts to the moment it becomes interactive
- Each session records the user's network conditions — connection type (4g/3g/2g), bandwidth, round-trip time, and whether Data Saver is on — so you can tell apart real backend slowness from a flaky network
- If your app already knows the user's country, city, or organization, you can pass that in at SDK init time instead of relying on inferred location. Learn more
Fixes
- Dashboards: Dashboard search from the global search bar returned only a few matches even when the dashboards page found many more
- Discover Services
- Performance graphs broke when maximized, making it hard to inspect a chart closely
- Filtering operations in the Outgoing Calls and All Operations tabs didn't update the graph above the list
- Agents: The coding agent onboarding page showed a placeholder endpoint instead of your organization's actual write URL
- Settings: Some users in multi-org setups saw a misleading "contact admin" error when logging in, even though they were already invited
- General: Charts didn't show the date on the x-axis when the selected range spanned more than a day, making it hard to tell which day a point belonged to