Service Level Agreement

1. Purpose and Scope

This Service Level Agreement ("SLA") defines the operational commitments MHLE ("Provider") makes to its Enterprise customers ("Customer") regarding the availability, performance, and support of the MHLE platform — a multi-perspective AI-powered knowledge synthesis and cognitive learning environment. This SLA applies exclusively to Enterprise-tier subscribers and governs the production environment only. Development, staging, and sandbox environments are expressly excluded.

2. Service Availability

2.1 Uptime Commitment

MHLE commits to a monthly uptime target of 99.5% for the core platform, measured across all production services including the analysis engine, knowledge graph, synthesis pipeline, and user-facing web application.

2.2 Measurement Method

• Monitoring is conducted via automated synthetic health checks at 60-second intervals against the production environment.
• Downtime is defined as any period during which the core platform is unavailable or returns HTTP 5xx errors for more than three consecutive checks (3 minutes).
• Measurements are logged and retained for 90 days, and monthly availability reports are available to Enterprise customers upon request.

2.3 Scheduled Maintenance

Scheduled maintenance windows are excluded from uptime calculations provided:

1. The Customer receives 72 hours' advance notice via the registered account email and in-app notification banner.
2. Maintenance is performed during the standard window: Saturdays 02:00–06:00 UTC.
3. Total scheduled maintenance does not exceed 8 hours per calendar month.

Emergency maintenance required to address active security incidents or critical failures may be performed outside the standard window with best-effort notification.

2.4 Exclusions

The following events are excluded from uptime SLA calculations:

• Force majeure events (natural disasters, government actions, widespread internet outages).
• Downtime caused by Customer's own actions, third-party integrations not managed by MHLE, or misuse of the API.
• Degradation attributable to upstream AI provider outages (OpenAI, Anthropic, Google, Perplexity) where MHLE failover mechanisms have been engaged.
• Scheduled maintenance performed within the agreed windows.
• Performance degradation during free or trial tier usage.
• Beta features explicitly labeled as pre-release.

3. Performance Standards

MHLE commits to the following response-time targets under normal load conditions (defined as ≤80% of provisioned capacity). All targets are measured from the moment a valid request is received by the MHLE application server to when a complete, usable response is returned to the client.

3.1 Standard Analysis (Multi-Headed Analysis)

Applies to: single-note multi-headed analysis across Financial, Ethical, Engineering, and Skeptic lenses.

3.2 Background Graph Analysis

Applies to: relationship detection, knowledge graph construction, and graph update operations triggered in the background.

3.3 Synthesis Paper Generation

Applies to: full academic synthesis paper generation from knowledge graphs.

3.4 Performance Degradation Notification

If any performance metric exceeds its P99 threshold for a sustained period of 15 minutes or more, MHLE will treat the event as a P2 incident and follow the incident response process defined in Section 4.

4. Incident Response

4.1 Severity Classification

4.2 Response Time SLAs

Business hours are defined as 09:00–18:00 Eastern Time (ET), Monday–Friday, excluding U.S. federal holidays.

4.3 Communications Protocol

1. Detection: Incidents are detected via automated monitoring or Customer report. P1 incidents automatically page the on-call engineering team.
2. Acknowledgement: Upon initial response, Customer receives confirmation of incident acknowledgement, assigned severity, and the name of the incident owner.
3. Updates: Status updates are published to the MHLE Status Page and communicated directly to affected Enterprise Customer administrators via email at the cadence defined in Section 4.2.
4. Resolution: Upon resolution, Customer receives an incident closure notice including a brief root-cause summary.
5. Post-Incident Review (PIR): For all P1 incidents, a written PIR is delivered to the Enterprise Customer within 5 business days of resolution, including root cause analysis, impact assessment, and remediation actions taken.

4.4 Incident Reporting Channel

Enterprise Customers must report suspected incidents through the designated Enterprise support channel (priority support email or ticketing portal). Incidents reported through general community channels may not receive priority handling.

5. Data Recovery

5.1 Recovery Objectives

5.2 Backup Cadence

5.3 Restoration Procedure

1. Incident Declaration: A data loss or corruption event is declared by the MHLE on-call team or escalated by an Enterprise Customer.
2. Scope Assessment: Engineering determines the extent and time boundary of the affected data.
3. Backup Selection: The most recent clean backup at or before the incident boundary is identified from the backup registry.
4. Restoration Execution: Database is restored to an isolated environment for validation before promotion to production.
5. Validation: Data integrity checks are run against restored state. Customer data access is verified.
6. Promotion & Notification: Restored environment is promoted to production. Enterprise Customer administrator is notified of restoration completion and confirmed RPO achieved.
7. Documentation: Full incident and recovery log is retained for audit purposes.

6. Support Tiers

6.1 Enterprise Support

6.2 Out-of-Scope Support

The following are not covered under this SLA:

• Issues caused by Customer's own infrastructure, browser, or network environment.
• Custom feature development or professional services (governed by separate SOW).
• Training and onboarding (available as a paid add-on).

7. Service Credits and Remedies

7.1 Credit Schedule

If MHLE fails to meet the monthly uptime commitment defined in Section 2.1, the Customer is eligible for a service credit applied to the following month's invoice:

7.2 Credit Claim Process

1. Customer must submit a credit claim within 15 calendar days of the end of the affected month.
2. Claims must be submitted in writing to the designated Enterprise support channel, referencing the specific incident(s).
3. MHLE will review the claim against internal monitoring data and respond within 10 business days.
4. Approved credits are non-transferable, have no cash value, and apply solely to future MHLE service fees.

7.3 Limitations

Service credits constitute the Customer's sole and exclusive remedy for SLA failures. Aggregate credits in any calendar month shall not exceed 30% of the monthly recurring fee for that month. Credits do not apply to downtime excluded under Section 2.4.

8. SLA Review and Amendments

This SLA is reviewed annually or upon a material change to the MHLE platform architecture. MHLE will provide Enterprise Customers a minimum of 30 days' written notice before any reduction in service commitments takes effect. Customers who do not accept amended terms must notify MHLE in writing within the notice period; continued use of the service constitutes acceptance.

Contact

• Enterprise & SLA inquiries: enterprise@mhle.app

Document Version 1.0 — Effective April 1, 2026 — © 2026 MHLE (Cognitive Engine, Inc.). This document is confidential and intended solely for MHLE Enterprise customers.