Service Level Agreement

Novex Systems LLC is a US company serving customers worldwide. The availability commitment and the service credits below apply to Cosmos. The data-protection statements apply to every paid plan. The support targets apply to all plans as targets, and are not contractual on any plan.

Novex targets 99.5% monthly availability. We have met that on both the web application and the API in each of the last two complete months (July and August 2026).

What “available” means. The Novex web application responds to a request and its health endpoint reports healthy.

How we measure it. Availability for a calendar month (UTC) is successful checks divided by total checks. The historical figures above come from our own monitor at status.nov3x.com. Since 9 June 2026 an external monitor (UptimeRobot) has also probed the public site and API from outside our infrastructure at a fixed interval; its record is public at https://stats.uptimerobot.com/Kc6JBIW3PL and, for July and August 2026, it too shows above 99.5% for both. A month at 99.5% allows roughly three and a half hours of downtime — 3 h 36 m in a 30-day month, 3 h 43 m in a 31-day month.

Recovery point. We archive the database transaction log continuously, at most five minutes behind. In a total loss of the primary database we would expect to lose no more than five minutes of committed data.

Recovery time. We hold an encrypted nightly backup off-site and we rehearse restoring it. Our most recent rehearsal restored a full copy in 15 minutes. We target restoring service within 4 hours of deciding to restore; the rehearsal time is the database alone and excludes diagnosis.

Retention. Nightly backups are encrypted at rest and held off-site. Point-in-time recovery is available for the most recent 7 days.

• Planned maintenance we expect to interrupt service. We announce it on the status page at least 48 hours ahead and schedule it in a period of low usage. Routine deployments are not planned maintenance: they are applied without interrupting service and are not announced.
• Emergency security patching, announced as soon as we reasonably can.
• Problems caused by something the customer controls — their network, their browser, their own integration, credentials they revoked, or usage above their plan’s published limits.
• Failures of the content-delivery and DNS network in front of Novex, including its edge, DNS or certificate services. We route all traffic through it and cannot serve around it.
• Failures of third-party services we depend on but do not run — WhatsApp/Meta, our email delivery provider, payment providers, and any identity provider the customer connects. Availability here is measured on the Novex application itself; a delayed message or a failed payment at one of those providers is not application downtime.
• Force majeure, including a datacentre or upstream network failure at our hosting provider.

If availability in a calendar month falls below 99.5%, the customer may request a service credit against the following month:

Requested in writing within 30 days of the month in question, applied to future invoices rather than paid in cash, capped at 50% of one month’s fee. Service credits are the sole remedy for missed availability.

We accept support requests 24 hours a day, every day — the inbox is always open; it is not a staffed 24-hour desk. There is no business-hours window and no single time zone: our customers are worldwide and our targets are measured in elapsed hours, not working hours.

We aim to reply within 24 hours. Cosmos customers also have a named Customer Success Manager. These are targets, not contractual commitments, and no service credit attaches to them. We will publish them as commitments once our support system records the time of first reply.

Beam (free) is supported through our documentation and status page.

For a total outage — Novex unreachable for all of a customer’s users — we treat it as our highest priority, post to the status page as soon as we have confirmed it, and keep that page updated until it is resolved.