Feb 28, 2026AIProductReliabilitySecurityUX
Shipping update - 2026-02-28
This release expands how far back AI-assisted investigations can go and adds stronger organization-level control over AI capabilities. Teams can troubleshoot older issues more reliably while admins get safer rollout controls.
Highlights
- AI-assisted telemetry workflows now support much longer historical lookback windows, moving from one day to up to two weeks.
- Organization admins can now enable or disable AI capabilities at the org level, with clear disabled states shown across AI entry points.
- Time-range query handling is more robust: when relative lookback limits are hit, requests can recover using explicit start/end timestamps.
What this enables
- Teams can investigate incidents and regressions that started days earlier without splitting work into multiple short-range queries.
- Admins can roll out AI usage in a controlled way, including pausing access quickly when policy or readiness changes.
- End users get clearer behavior when AI features are unavailable, reducing confusion and failed interactions.
Technical wins
- Query constraints were updated consistently across logs, traces, metrics, and APM paths, with shared validation aligned to the new two-week window.
- Fallback guidance for ISO timestamp ranges was added to AI tooling flows, improving reliability when lookback-based requests fail.
- AI access control now combines backend route gating with frontend feature checks, so unauthorized paths are blocked server-side while the UI communicates availability accurately.
- Admin-only setting updates were enforced for AI controls, which tightened permission boundaries and reduced accidental configuration changes.