Platform availability
99.5%
Standard
≤ 43.8 hrs / year permitted
99.9%
Enterprise
≤ 8.7 hrs / year permitted
99.95%
Gov Highside
≤ 4.4 hrs / year permitted
Uptime is measured as the percentage of minutes the platform API is available in a calendar month, excluding scheduled maintenance (communicated ≥ 72 hours in advance) and events outside OBEL's reasonable control (see exclusions below).
Severity definitions
How we classify incidents.
When you open a support ticket, you nominate an initial severity. Our team will confirm or adjust within the initial response. The final classification governs the applicable response commitment.
The OBEL platform is completely unavailable or the security pipeline (PII scrubber, ARGUS-i™ classifier, sovereign gate) is not functioning. All users in the affected organisation are unable to submit requests, or requests are passing through without security enforcement.
A significant feature is unavailable or severely degraded. Core workflows are impacted for a material number of users but the security pipeline remains operational.
A non-critical feature is degraded, a workaround exists, or a cosmetic/UX issue is causing meaningful friction for users.
A minor bug, documentation request, feature enquiry, or general question with no operational impact.
Response time SLAs
Guaranteed response times by plan.
How to reach us
Support channels and escalation path.
Email support
Open a support ticket via email to support@ninthlabs.ai. Include your organisation ID, severity level, and a description of the issue. All plans.
Dedicated Slack channel
Enterprise and Gov Highside customers receive a private Slack Connect channel for direct async communication with the support and engineering team.
Security issues
Security vulnerabilities and potential breaches must be reported immediately to security@ninthlabs.ai — not via standard support channels. P1 SLA applies.
Engineering escalation
Gov Highside contracts include a named escalation path directly to the OBEL engineering team for deployment issues, compliance queries, and critical incidents.
SLA exclusions
SLA commitments do not apply to: scheduled maintenance (communicated ≥ 72 hours in advance); incidents caused by the customer's own actions or third-party integrations outside OBEL's control; third-party model provider outages (OpenAI, Anthropic, Google, etc.) where OBEL's infrastructure is unaffected; force majeure events including but not limited to natural disasters, government actions, and internet infrastructure failures. Uptime calculations exclude the above and are measured over a rolling calendar month.
SLA credits for Enterprise and Gov Highside: where OBEL fails to meet the stated availability target, affected customers may request a service credit equivalent to 10× the pro-rated daily fee for each hour of downtime beyond the target. Maximum credit per month is 30% of the monthly fee. Credits are applied to the next invoice — they are not paid out in cash. Contact support@ninthlabs.ai within 14 days of the incident to claim.