Assurance center
One health view for the parts of AxioRank that run in the background, your integrations, the ML lane, data residency, and the async job queue, so a silent degradation never hides.
Most of AxioRank's value sits on the hot path, where a failure is loud: a denied call, a raised alert. But the supporting machinery runs in the background, where a failure can be quiet. A Slack channel stops delivering, the ML lane runs out of quota, a webhook receiver starts timing out. The assurance center is the single view that surfaces these before they become a gap in your coverage.
Where to find it
The assurance center reads live from your workspace, summarized into a health
banner on the dashboard and inline callouts on the pages each signal affects
(channels, alerts, policies). A single needsAttention count tells you, at a
glance, whether anything wants a look.
What it checks
Integration health
Every outbound delivery path reports its current state (ok, degraded, or down):
- Notification channels: Slack and email destinations for alerts and digests.
- Log streams: SIEM forwarders (Splunk, Datadog, Elastic, S3).
- Webhook endpoints: your own receivers for AxioRank events.
- SOAR connections: Jira, ServiceNow, and signed-webhook response actions.
A path that has started failing is flagged so you can fix it before the next alert silently fails to land.
ML lane status
The detection-intelligence lane is optional and gated, so the assurance center reports exactly where it stands: whether your plan is entitled (Team and above), whether the workspace has opted in to model egress, and whether the endpoint is configured. When all three are true, it also reports the last 24 hours of health: how many assessments ran, completed, or came back unavailable, so a degradation (errors plus unavailable responses) is visible rather than inferred.
Data residency
If your workspace selects an EU data region, the assurance center confirms the region is actually being honored: an in-region inference endpoint must be configured, or the lane would fall back to the default region. The check makes a residency promise verifiable instead of assumed.
Dead-letter queue
Async jobs (webhook deliveries, stream forwards, assessments) that exhaust their
retries land in a dead-letter queue. The assurance center reports the pending
count so a backlog never goes unnoticed. You can inspect and replay a dead-lettered
job once the underlying cause is fixed (POST /api/dead-letter/[id]/replay).
Reading the report
needsAttention sums the conditions that warrant action: an integration that is
degraded or down, a data region that is not being honored, an ML lane that is
entitled and opted in but degrading, or a non-empty dead-letter queue. Zero means
the background machinery is healthy. A non-zero count points you straight at the
signal that needs attention.
Next steps
- Detection intelligence: the ML lane this monitors.
- Audit export: the log streams this monitors.
- Response engine: the SOAR and webhook paths this monitors.
- Data residency: the residency promise this verifies.