UTC --:--:--

Goable — Service Level Agreement (SLA)

Version: 2026-06-26. Targets below are the published commitments; as more post-launch uptime data accumulates they may be revised upward.

This SLA forms part of the Terms of Service and applies to paid plans (Starter / Pro / Scale). The Free plan is best-effort with no uptime commitment, no service credits, and no support response targets, see clause 9 below.

1. Uptime commitment

PlanMonthly Uptime targetCovered endpoints
Starter99.5%Core scoring (/v1/score, /v1/score/series, /v1/score/multi)
Pro99.9%Core scoring + /v1/recommend-spot + /v1/decision + /v1/score/explain-counterfactual (each covered only while declared GA)
Scale99.9%All endpoints declared GA under your plan

"Monthly Uptime" means the percentage of minutes in a calendar month, measured in UTC, in which Goable's monitoring shows the applicable GA endpoint was available to accept and process valid, authenticated requests. A response is treated as unavailable when Goable returns a 5xx, fails to respond, or exceeds the published timeout threshold due to Goable-controlled systems. Documented exclusions in clause 2 apply.

Measurement source: Goable's server-side monitoring, evaluated against the /v1/health/ready readiness probe and genuine tenant requests under normal load. Timezone: UTC. Scheduled maintenance: Goable targets no more than 4 hours of scheduled maintenance per calendar month, announced at least 48 hours in advance (see clause 6); this target is reviewed periodically.

Only endpoints declared GA are covered. Endpoints in Preview or design-partner access, or that are planned and not yet available, are excluded from these targets (see clause 10).

2. Exclusions from Monthly Uptime

The following are excluded from the Monthly Uptime calculation:

  • Scheduled maintenance announced at least 48 hours in advance via the public status strip and the tenant contactEmail.
  • Emergency security patches applied to mitigate an active vulnerability; we minimise these windows and disclose them promptly once started.
  • Force majeure events as defined in the Terms of Service.
  • Customer misconfiguration: invalid API keys, malformed request payloads (4xx response codes), exhaustion of plan rate limits or daily quotas, exhaustion of the per-tenant LLM token budget.
  • Third-party provider outages, handled by severity of impact:
    • Degraded-with-flagged-output: where a provider degradation is absorbed and the API returns a clearly-flagged degraded-confidence response (with a confidenceDetail.driftFlag or degraded_mode indicator), the endpoint is treated as available and this does not count against uptime.
    • Single-provider error absorbed by fallback: where one upstream provider errors but the request is served through the provider fallback chain, the endpoint is treated as available.
    • Total loss: only where an upstream failure genuinely prevents Goable from returning any valid response (typically a 503 with an X-Goable-Degraded header) is the exclusion applied. Degradations that Goable's own systems could have absorbed are not excluded as third-party outages.
  • Preview / design-partner / planned endpoints excluded per clause 10.

3. Service credits

If the Monthly Uptime for your plan falls below the target, you may request a service credit on a future invoice:

PlanTier 1Tier 2Tier 3
Starter<99.5% — ≥99.0% → 10% credit<99.0% — ≥95.0% → 25% credit<95.0% → 50% credit
Pro<99.9% — ≥99.0% → 10% credit<99.0% — ≥95.0% → 25% credit<95.0% → 50% credit
Scale<99.9% — ≥99.0% → 15% credit<99.0% — ≥95.0% → 30% credit<95.0% → 50% credit

The credit applies to the monthly subscription fee of the affected plan (not to metered overage charges) and is the sole and exclusive remedy for SLA breach.

To claim a credit: send a request to support@goable.io within 30 days of the end of the affected month, including the time windows you observed downtime and any correlation id we may need to investigate. We respond within 10 business days; approved credits are applied to the next invoice.

4. Support

PlanChannelTarget first-response timeHours
FreeCommunity / docs onlyN/AN/A
StarterEmail (support@goable.io)Best-effort, next business dayMon–Fri 09:00–18:00 CET
ProEmail (support@goable.io)1 business day for general issues, 4 business hours for production-impact issuesMon–Fri 09:00–18:00 CET
ScaleEmail + named contact4 business hours for general issues, 1 business hour for production-impact issuesMon–Fri 09:00–18:00 CET (after-hours on agreed escalation paths)

"Production-impact" means a 5xx error rate above 1% sustained for more than 5 minutes against your tenant key, or a security-class incident affecting your tenant.

5. Severity classification

When you open a support ticket, classify the issue with one of:

  • Sev 1 — service unavailable for your tenant; cannot score; no workaround.
  • Sev 2 — degraded performance or partial feature failure; workaround exists.
  • Sev 3 — non-impacting issue, documentation question, feature request.

Scale-plan response times are committed per severity; Pro best-effort follows the same ladder.

6. Maintenance windows

Routine maintenance is performed in low-traffic windows (typically 02:00–04:00 CET on weekdays or weekends) and announced via the public status strip and tenant email at least 48 hours in advance. Goable targets no more than 4 hours of scheduled maintenance per calendar month.

Emergency security patches may be applied without prior notice; we disclose them promptly once started via the status strip.

7. Data durability and backups

  • Primary database: Neon Postgres with automated point-in-time recovery and regular logical backups; backups stored encrypted with separate IAM access.
  • Audit log: 1-year retention by default per the Terms; extendable under signed DPA addendum.
  • Recovery targets: Goable maintains documented internal Recovery Point and Recovery Time targets for the calibration, outcomes, and tenant-account datasets, reviewed periodically. These are operational targets rather than hard guarantees.

8. Status and incident communication

  • Status strip: https://goable.io (the homepage carries a live status strip).
  • Public status page: when a more detailed public status page ships, it will be at https://status.goable.io.
  • Incident updates: posted to the status strip promptly on detection; emailed to tenant contacts on incidents that affect their traffic; post-mortem published within 7 business days of resolution for any incident that breached the SLA.

9. Free plan — no SLA

The Free plan is provided on a best-effort basis with no uptime commitment, no service credits, and no support response targets. We will of course do our best to maintain it, but if continuous availability is critical to your use case, please upgrade to Starter or above.

10. Product status and endpoint availability

Goable uses one status vocabulary across its documentation and this SLA:

  • GA — generally available; contractually available and SLA-covered where included in your plan.
  • Preview — available for testing, excluded from the SLA, and not intended for critical decisions.
  • Design partner — limited contractual access under a specific arrangement, not general availability.
  • Planned — not available; not to be relied on as a current capability.

Only GA endpoints count toward the uptime targets in clause 1. The following surfaces are not GA today:

  • /v1/score/difficulty — Pro+, Preview.
  • /v1/projections/adaptation-report — Scale-only, Planned pending CORDEX / R2 data-ingest activation.
  • /v1/underwriting/quote|bind|evaluate — Scale-only, Design partner, awaiting first paid integration.

We will update this list as endpoints graduate to GA.

11. Contact

Support: support@goable.io Status / incidents: see the homepage status strip and the support@ channel Legal / contractual notices: contact@fabio-carucci.com


Last updated: 2026-06-26. This document is published at https://goable.io/legal/sla.