Meridian Intel SLA
Uptime, support, and data commitments for Meridian Intel customers. This SLA applies to the Enclave (Enterprise) tier by default and is available to Vector tier customers by request.
Overview
This Service Level Agreement ("SLA") describes the service commitments Mazal Arc provides to Meridian Intel customers. It covers platform availability, support response times, data handling, and the remedies available if we fall short.
Effective date: July 2026. This SLA is reviewed and updated quarterly. Customers are notified 30 days in advance of any material changes.
Platform Availability
We measure availability as the percentage of time the Meridian Intel API and workspace are operational and accepting requests in a given calendar month.
| Tier | Monthly Uptime Target | Max Permitted Downtime |
|---|---|---|
| Enclave (Enterprise) | 99.9% | ~43 minutes / month |
| Vector (Premium) | 99.5% | ~3.6 hours / month |
| Lattice / Segment | 99.0% | ~7.3 hours / month |
| Ephemeral (Free) | Best effort | No SLA |
How we measure
Uptime is measured by automated health checks against the gateway and worker endpoints at 60-second intervals. A failure is recorded when three consecutive checks fail. Scheduled maintenance windows (announced 72 hours in advance) are excluded from uptime calculations.
Maintenance windows
Scheduled maintenance occurs during low-traffic windows (typically Sunday 02:00–06:00 UTC). Emergency maintenance may occur at any time but will be communicated via email within 15 minutes of initiation.
Support Response Times
Response times are measured from when a support request is received at support@meridianintel.app to when a team member acknowledges the issue.
| Severity | Definition | Enclave Response | Vector Response |
|---|---|---|---|
| Critical | Platform completely unavailable or data integrity at risk | 1 hour | 4 hours |
| High | Core functionality degraded — queries failing, ingestion broken, auth down | 4 hours | 12 hours |
| Medium | Non-critical feature issue — history not loading, UI rendering problem | 1 business day | 2 business days |
| Low | General inquiry, feature request, documentation question | 2 business days | 3 business days |
Support channels
- Email: support@meridianintel.app — all tiers
- In-app bug report: Auto-filled with tier, field, and document context — all tiers
- Dedicated onboarding: Enclave tier includes a named contact and onboarding session
Data Commitments
Retention
| Data Type | Retention | Notes |
|---|---|---|
| Ephemeral session data | Session only | Cleared on page close. No server-side persistence. |
| Document chunks (Segment+) | 30 days after last access | Auto-expired. User can delete anytime from History. |
| Document chunks (Vector/Enclave) | Perpetual | Stored until user deletes. AES-256-GCM encrypted at rest. |
| Query history | 30 days | Auto-purged. User can delete individual queries. |
| Feedback ratings | Perpetual | Used for quality improvement. No document content stored in feedback. |
Encryption
All document chunks are encrypted with AES-256-GCM at the application level before storage. Encryption keys are managed server-side and are not accessible through the database or storage layer.
Isolation
- Enclave tier: Per-user isolated vector store collections. Complete separation of document data between users.
- All tiers: Strict user-ID and document-ID scoping on every query. Cross-document contamination prevention verified with automated tests.
Deletion
Users can delete individual documents and queries from the History sidebar at any time. Deleted data is removed from the vector store, FTS5 index, and SQLite databases. Deletion is permanent and irreversible.
Incident Management
Notification
In the event of a service disruption affecting availability or data integrity, we will notify affected customers by email within:
- 15 minutes for Critical severity incidents
- 1 hour for High severity incidents
- Next business day for Medium/Low severity incidents
Post-incident review
For Critical and High severity incidents, a post-incident report will be provided within 5 business days. The report includes root cause analysis, timeline, impact scope, and preventive measures.
Escalation
If a support request is not acknowledged within the response time for its severity level, customers may escalate by emailing faith@meridianintel.app directly.
Service Credits
If monthly uptime falls below the committed target, affected customers receive service credits applied to the next billing cycle.
| Monthly Uptime | Service Credit |
|---|---|
| 99.0% – 99.9% | 10% of monthly fee |
| 95.0% – 98.9% | 25% of monthly fee |
| Below 95.0% | 50% of monthly fee |
Credits must be requested within 30 days of the affected month. Credits do not apply to the Ephemeral (Free) tier. Maximum credit in any month is 50% of that month's fee.
Exclusions
This SLA does not apply to downtime caused by:
- Scheduled maintenance announced 72+ hours in advance
- Force majeure events (natural disasters, acts of war, government actions)
- Third-party service outages beyond our control (Google Cloud, OpenAI, AWS)
- Customer-side network, hardware, or browser issues
- Abuse, misuse, or violation of the Terms of Service
- Features explicitly labeled as beta, preview, or experimental
Tier Coverage
| Commitment | Ephemeral | Segment | Lattice | Vector | Enclave |
|---|---|---|---|---|---|
| Uptime target | — | 99.0% | 99.0% | 99.5% | 99.9% |
| Support response | Best effort | 3 days | 2 days | See table | See table |
| Service credits | — | — | — | ✓ | ✓ |
| Incident reports | — | — | — | Summary | Full RCA |
| Dedicated contact | — | — | — | — | ✓ |
| Custom SLA terms | — | — | — | — | ✓ |
Custom SLA: Enclave customers may negotiate custom uptime targets, response times, and credit terms as part of their enterprise agreement. Contact sales@meridianintel.app to discuss.