// scope
This SLA applies to managed services delivered under a signed Master Services Agreement that references it. The tier, uptime target, and support window for your engagement are stated in your SoW. Where your SoW specifies a higher tier than the defaults below, the SoW controls.
Last updated: August 16, 2026
1. Scope
This Service Level Agreement (the "SLA") describes the service-level commitments for managed services we deliver under a signed Master Services Agreement (MSA) and Statement of Work (SoW). For each service, the SoW names the tier: Standard, Business, or Sovereign, and that tier sets the targets below.
2. Definitions
- "Service" means the specific managed workload named in the SoW (for example, a private LLM endpoint, a hosted web application, an inference cluster).
- "Uptime" means the percentage of total minutes in a calendar month that the Service is reachable and responding to valid requests with a successful status code, measured as described in Section 6.
- "Downtime" means the difference between 100% and Uptime for the Service during a calendar month.
- "Incident" means an unplanned disruption to a Service, or a material degradation of a Service, that we confirm upon investigation.
- "Service Credit" means a dollar credit issued against the next invoice, calculated under Section 4.
- "Business Hours" means 08:00-18:00 Pacific Time, Monday through Friday, excluding U.S. federal holidays.
- "Sev" means severity, classified per Section 5.
3. Uptime commitments
| Tier | Monthly Uptime | Max Downtime / 30 days | Maintenance window |
|---|---|---|---|
| Standard | 99.5% | 3 h 36 min | Sun 02:00-06:00 PT, with 5 business days notice |
| Business | 99.9% | 43 min | Sun 02:00-04:00 PT, with 5 business days notice; emergency maintenance permitted with 24h notice |
| Sovereign | 99.95% | 21 min | Coordinated, with 14 days notice; emergency maintenance requires joint approval |
Sovereign tier additionally includes a hot standby, quarterly failover test, and a 1-hour Recovery Time Objective (RTO) with a 15-minute Recovery Point Objective (RPO) for in-VLAN services.
4. Service credits
When Uptime falls below the committed level for a calendar month, we issue a Service Credit on the next invoice as follows:
| Uptime achieved | Standard credit | Business credit | Sovereign credit |
|---|---|---|---|
| ≥ target | - | - | - |
| Target − 0.5 pts | 5% | 10% | 15% |
| Target − 1.0 pts | 10% | 20% | 30% |
| Target − 2.0 pts | 25% | 40% | 60% |
| < Target − 2.0 pts | 50% | 75% | 100% |
"Credit %" is a percentage of the monthly fee for the affected Service. Credits are your sole and exclusive remedy for failure to meet an uptime target, except where a different remedy is required by applicable law.
5. Support and response times
| Severity | Definition | Standard | Business | Sovereign |
|---|---|---|---|---|
| Sev-1 | Production down; no workaround. | Response 4 business hours · Mitigation 1 business day | Response 1 hour · Mitigation 4 hours | Response 15 min · Mitigation 1 hour · 24/7 |
| Sev-2 | Production degraded; workaround exists. | Response 1 business day · Mitigation 2 business days | Response 4 business hours · Mitigation 1 business day | Response 1 hour · Mitigation 4 hours · 24/7 |
| Sev-3 | Non-production or minor issue. | Response 2 business days | Response 1 business day | Response 4 business hours |
| Sev-4 | Question, how-to, feature request. | Response 3 business days | Response 2 business days | Response 1 business day |
Response time is the maximum time between a ticket being opened with a valid severity and an engineer acknowledging it. Mitigation is the maximum time until the Service is restored, the workaround is in place, or the affected component is no longer user-impacting.
6. How we measure uptime
We measure Uptime from synthetic checks and synthetic user transactions originating from at least two geographically distinct monitoring regions, every 60 seconds. A check must succeed (HTTP 2xx or 3xx within 5 seconds for the synthetic request, or a successful OpenAI-compatible /v1/models response within 5 seconds for inference services) for the Service to be counted as up.
We publish a per-Service status page at status.empowered.guru (or, for Sovereign customers, a dedicated status URL). The status page is the source of truth for incidents.
7. What is excluded
Downtime does not include any period during which:
- The Service is down due to your Client Materials, your configuration, your network, or your breach of the MSA / AUP.
- You have requested a scheduled or emergency maintenance window.
- A force majeure event (per our Terms) prevents delivery.
- Downtime is caused by a third-party service outside our reasonable control, except where we have contractually committed to a third-party dependency in the SoW (for example, a specific model registry).
- The Service was intentionally suspended under the AUP or MSA.
- Betas, previews, or features marked "best-effort" in the SoW.
8. How to claim a credit
Email billing@empowered.guru within 30 days of the end of the affected month with: (a) the Service name, (b) the affected month, (c) the times you observed the issue, and (d) any supporting evidence (request IDs, screenshots). We acknowledge within 5 business days, validate against our monitoring data, and either issue the credit on your next invoice or explain in writing why the claim is denied under Section 7.
9. Changes
We may update this SLA. Changes that narrow commitments will not apply to an active SoW during its current term; they apply only to the next renewal unless the SoW is amended. The current version is always at this URL.
10. Contact
Questions: support@empowered.guru. For incident reports: security@empowered.guru or our responsible disclosure page.