Skip to main content

Aragora Service Level Agreement (SLA)

This Service Level Agreement ("SLA") describes the service availability commitments and support terms for Aragora platform services.

1. Definitions​

  • "Monthly Uptime Percentage": Total minutes in a month minus minutes of Downtime, divided by total minutes in a month.
  • "Downtime": Period where the Aragora API returns 5xx errors for more than 5 consecutive minutes, excluding scheduled maintenance.
  • "Scheduled Maintenance": Planned maintenance announced at least 72 hours in advance.
  • "Credits": Service credits applied to future invoices as compensation for SLA breaches.

2. Service Tiers​

FeatureStandardProfessionalEnterprise
Uptime SLA99.5%99.9%99.95%
Support HoursBusiness hoursExtended (16x5)24x7x365
Response Time (P1)4 hours1 hour15 minutes
Response Time (P2)8 hours4 hours1 hour
Response Time (P3)2 business days1 business day4 hours
Dedicated SupportNoSharedNamed CSM
Custom IntegrationsNoLimitedFull
SLA CreditsNoYesYes
Error Budget DashboardNoYesYes

3. Uptime Commitment​

3.1 Monthly Uptime Targets​

TierMonthly UptimePermitted Downtime
Standard99.5%3 hours 36 minutes
Professional99.9%43 minutes
Enterprise99.95%21 minutes

3.2 Service Credit Schedule​

If Aragora fails to meet the Monthly Uptime Percentage, eligible customers receive service credits:

Monthly UptimeCredit (% of Monthly Fee)
99.0% - 99.9%10%
95.0% - 99.0%25%
90.0% - 95.0%50%
Below 90.0%100%

Maximum Credit: Total service credits cannot exceed 100% of monthly fees.

4. Performance Targets​

4.1 API Latency​

Endpoint CategoryP50P95P99
Debate Creation50ms200ms500ms
Debate Retrieval10ms50ms100ms
Document Processing100ms300ms1000ms
WebSocket Messages5ms20ms50ms

4.2 Debate Completion​

Debate TypeP50P95
Standard (3 agents)8s20s
Extended (5+ agents)15s45s
Complex (10+ rounds)30s90s

Note: Debate completion times depend on LLM provider response times and are best-effort targets.

4.3 Webhook Delivery​

MetricTarget
Delivery Latency (P99)< 5 seconds
Delivery Success Rate99.9%
Retry Attempts3 (with exponential backoff)

4.4 Error Rates​

ServiceTargetMaximum
API Error Rate< 0.5%< 1.0%
WebSocket Error Rate< 1.0%< 2.0%
Background Job Failure< 0.1%< 0.5%

5. Support Response Times​

5.1 Priority Definitions​

PriorityDescriptionExamples
P1 - CriticalComplete service outage, data loss riskAPI unavailable, database corruption
P2 - HighMajor feature unavailable, severe degradationDebates failing, auth broken
P3 - MediumFeature impaired, workaround availableSlow performance, UI issues
P4 - LowMinor issue, cosmetic, feature requestDocumentation, enhancement

5.2 Response and Resolution Targets​

PriorityFirst ResponseStatus UpdateTarget Resolution
P1Per tier SLAEvery 30 min4 hours
P2Per tier SLAEvery 2 hours8 hours
P3Per tier SLADaily5 business days
P42 business daysWeeklyBest effort

6. Maintenance Windows​

6.1 Scheduled Maintenance​

  • Notification: At least 72 hours advance notice
  • Preferred Window: Sundays 02:00-06:00 UTC
  • Frequency: No more than once per month for major maintenance
  • Duration: Maximum 4 hours per maintenance window

6.2 Emergency Maintenance​

  • Notification: As soon as reasonably possible
  • Scope: Security patches, critical fixes only
  • Post-Incident: Root cause analysis within 48 hours

7. Exclusions​

The SLA does not apply to:

  1. Force Majeure: Events beyond reasonable control
  2. Customer Actions: Misconfiguration, abuse, or unauthorized access
  3. Third-Party Services: LLM provider outages, cloud infrastructure failures
  4. Beta Features: Features explicitly marked as beta or preview
  5. Free Tier: SLA applies only to paid subscriptions
  6. API Abuse: Requests exceeding rate limits or terms of service

8. Claiming Credits​

8.1 Process​

  1. Submit claim within 30 days of incident
  2. Include: Account ID, incident date/time, description
  3. Submit to: [email protected] or via support portal

8.2 Verification​

  • Aragora will verify claims against monitoring data
  • Response within 10 business days
  • Credits applied to next invoice cycle

9. Monitoring and Reporting​

9.1 Status Page​

Real-time service status available at: status.aragora.ai

9.2 Monthly Reports (Enterprise)​

  • Uptime metrics and trend analysis
  • Incident summaries
  • Performance against SLO targets
  • Error budget consumption

9.3 Prometheus SLO Recording Rules​

Enterprise customers receive access to SLO metrics:

# API Availability
- record: slo:api:availability:ratio
expr: sum(rate(http_requests_total{status!~"5.."}[5m])) / sum(rate(http_requests_total[5m]))

# API Latency P99
- record: slo:api:latency:p99
expr: histogram_quantile(0.99, sum(rate(http_request_duration_seconds_bucket[5m])) by (le))

# Webhook Delivery Latency P99
- record: slo:webhook:delivery:latency:p99
expr: histogram_quantile(0.99, sum(rate(webhook_delivery_seconds_bucket[5m])) by (le))

# Debate Completion P95
- record: slo:debate:completion:p95
expr: histogram_quantile(0.95, sum(rate(debate_completion_seconds_bucket[5m])) by (le))

10. Contact Information​

ChannelContactAvailability
Support Portalsupport.aragora.ai24/7
Email[email protected]Per tier
Phone (Enterprise)Dedicated line24/7
Slack (Enterprise)Dedicated channel24/7

11. SLA Revision​

This SLA may be updated with 30 days notice. Current version is always available at docs.aragora.ai/sla.


Effective Date: January 28, 2026
Version: 1.0.0
Last Updated: January 28, 2026