WOS Wontopos WOS
Back to site사이트로
Legal법적 고지
Terms of Service이용약관 Privacy Policy개인정보처리방침 Data Processing (DPA)데이터 처리 계약 Service Level (SLA)서비스 수준 협약 Acceptable Use (AUP)이용 정책

Service Level Agreement (SLA)

Last Updated: 2026-06-29 Service Provider: Wontopos, L.L.C.


1. Service Availability (Uptime)

Wontopos targets the following monthly uptime goals, tiered by operational maturity. These are internal targets and do not constitute a contractual guarantee or entitle clients to service credits or compensation.

Period Monthly Uptime Target Max Downtime/Month Notes
Year 1 (through ~2027 Q1) 99.3% ~5 hours Current operational baseline
Year 2 99.5% ~3.6 hours After planned 24/7 alerting
Year 3 99.9% ~43 minutes After planned multi-AZ and on-call rotation

Uptime is calculated as: (Total minutes - Downtime minutes) / Total minutes x 100

"Downtime" means a sustained period of five (5) or more consecutive minutes during which the WOS API is completely unavailable and returns server-side errors (HTTP 5xx) to all requests, as confirmed by our monitoring systems. Slow response times, intermittent or isolated request failures, and errors caused by Client requests (e.g., malformed requests, HTTP 4xx, rate-limit responses) do not constitute Downtime.

Excluded from downtime calculation:

  • Scheduled maintenance (announced 7+ days in advance)
  • Downtime caused by Client's actions or third-party services outside our control (including Client's own LLM providers)
  • Force majeure events

2. API Performance Targets

Metric Target
API response time (average) < 400ms
API response time (p99) < 1000ms
Typical server-side range Varies by data scale and tier

Note: These are targets, not guarantees. Response time targets apply to WOS memory storage and retrieval operations only, and are measured server-side, excluding network transit and the Client's own LLM time. Downstream LLM inference time (Client's own LLM) is excluded from these measurements.

3. Incident Response Times

Severity Definition First Response
Critical (P1) Service completely unavailable Business hours: 4 hours / Off-hours: best-effort, aligned with our alerting roadmap
High (P2) Major feature unavailable or significant degradation 24 hours
Medium (P3) Minor feature impaired Best effort
Low (P4) General inquiries, minor issues Best effort

Business hours: Monday–Friday, 09:00–18:00 KST. Because business hours are KST, US and global customers should expect off-hours handling for incidents reported outside these hours.

Until 24/7 alerting exists, the response times above are good-faith targets, not guarantees.

4. Scheduled Maintenance

  • Advance notice: Minimum 7 days (routine maintenance)
  • Notifications sent via: the status page at status.wontopos.com
  • Emergency maintenance (security-critical): Minimum 24 hours notice where possible

5. Service Credits

The Service is provided without service credits. The uptime targets in Section 1 are internal goals and do not entitle clients to refunds, service credits, or other compensation for any failure to meet them.

6. Support Channels

Plan Support Channel Response SLA
All plans Request form on our website at wontopos.com/contact 1 business day
Enterprise Priority response per separately negotiated enterprise agreement (available as on-call capability is added) Per enterprise agreement

7. Status Page

Real-time service status and incident history: status.wontopos.com

8. Data Backup and Recovery

Backup Type Frequency Retention Window
Account, billing, and usage database (RDS) Daily 7 days 03:19–03:49 UTC
Long-term snapshot Weekly 30 days Scheduled
Vector-store (memory) mirror Daily Latest (same-day) Mirrored to a separate storage server

Database backups are encrypted and retained per the table above. The vector store (customer memory) is mirrored daily to a separate storage server for box-failure recovery; a point-in-time, immutable archive is planned. Audit logs (CloudTrail) are stored with S3 Object Lock (immutable) and retained for 7 years.

Recovery time objective (RTO) and recovery point objective (RPO) are available upon request for Enterprise tier clients.

9. SLA Exclusions

This SLA does not apply to:

  • Beta or preview features
  • Downtime caused by Client's misuse or exceeding rate limits
  • Third-party service failures outside our control (including Client's own LLM providers)
  • Force majeure events

10. Review and Updates

This SLA will be reviewed annually. Material changes require 30 days advance notice.

← Previous이전 Data Processing (DPA)데이터 처리 계약 Next다음 → Acceptable Use (AUP)이용 정책
WOSWontoposWOS
Won + Topos - one place where memory lives.
Product
Why WOS Benchmark Pricing For agents
Developers
Docs API SDKs
Legal
Terms of Service Privacy Policy Data Processing (DPA) SLA Acceptable Use
Start
Get API key Contact Self-host