AI Safety & Governance

When the first approver is asleep,
the request climbs the ladder, on its own

Approval Escalation is the time-boxed chain behind every approval in Approval Queue. Each tier has a TTL. If the tier does not respond in time, the request escalates to the next tier automatically. The chain is configurable per service tier and per action class. No more "we paged the on-call but they were on a flight, so the action just sat there for two hours."

Get Started Talk to Sales
app.novaaiops.com / approval-escalation
● LIVE
5
Default chain depth
Tunable
TTL per tier
Auto
no missed approvals
Logged
every escalation
Five Default Tiers

Owner, on-call, lead, secondary, admin

The default chain has five tiers. The action's owner team is tier 1. Their on-call is tier 2. Team lead is tier 3. Secondary on-call is tier 4. Platform admin is tier 5. Each tier has its own TTL: short for low-risk classes, longer for time-sensitive incidents. Tiers can be added, removed, or reordered per action class.

  • Tier 1 · owner: the team that owns the resource; handles most approvals
  • Tier 2 · on-call: team's on-call rotation if owner did not respond in TTL
  • Tier 3 · team lead: lead engineer; usually only paged for ambiguous cases
  • Tier 4 · secondary: secondary on-call from another team for cross-cutting actions
  • Tier 5 · platform admin: last resort; sub-minute TTL means platform admin is paged within a few minutes if everything else stalled
app.novaaiops.com / approval-escalation · tiers
Per-Class Tuning

Different actions, different urgency

A routine config change can wait an hour. A destructive prod operation cannot. Each action class has its own escalation chain. Routine: 60-minute TTL per tier. Destructive (DROP, TRUNCATE, force-push): 2-minute TTL with simultaneous notify to all five tiers. The class-specific tuning is what makes the system feel calibrated, not annoying.

  • Routine class · 60m TTL: config edits, log queries, read-only reports, long fuses
  • Destructive class · 2m TTL: sub-minute escalation; first responder wins
  • Time-sensitive · simultaneous notify: page all tiers at once instead of sequentially when the incident is hot
app.novaaiops.com / approval-escalation · per-class
Quiet Hours Aware

Skip the sleeping engineer, do not wake the whole company

Each engineer can mark quiet hours (vacation, parental leave, conference). The chain skips quiet-hour engineers and goes to the next available tier. The skip is logged so audit can see why a request escalated faster than usual. Nobody gets paged at 3am when they marked themselves unreachable.

  • Quiet hours respected: engineers in quiet hours are skipped; chain advances to the next tier
  • Logged skip reason: every skip is recorded with the engineer's quiet-hours config so audit sees the math
  • Backup-tier coverage: when a quiet-hour engineer is the only person at their tier, the chain pre-promotes the next available tier
app.novaaiops.com / approval-escalation · quiet
Final Tier Behavior

When everyone is silent, the agent does not act

If every tier exhausts without an approval, the request does NOT escalate to "just do it anyway." The agent falls back to suggest mode and writes a runbook for the next available human to execute by hand. Default-deny is the right posture; auto-approval after a chain timeout would defeat the whole point of the gate.

  • Default-deny on full chain timeout: no auto-approval ever; the gate fails closed
  • Fallback to suggest mode: agent writes a runbook documenting the proposed change; a human runs it later
  • Notify and audit: failed-chain events are surfaced to platform admin so the chain config can be tuned (longer TTL, more depth)
app.novaaiops.com / approval-escalation · timeout
Video walkthrough coming soon

Subscribe to Nova AI Ops on YouTube for demos, tutorials, and feature deep-dives.

A chain that catches the rare cases without paging everyone

Most approvals land at tier 1. The chain exists for the 5% that do not. Configurable, audited, never silent.

Get Started Request a Demo