Maintenance Services and SLAs

Version 1.2 · Last updated 2026-07-27 · legal@sparred.ai

This page is the operational complement to our General Terms and Conditions. It describes how Prestance Lab keeps the Solution running, the service levels we commit to, and the channels Customers use to reach us when something breaks. It applies to every paid Order Form and is referenced from the GTCs as the then-current Maintenance Services description and SLA.

1. Scope of Maintenance Services

The Maintenance Services cover the production Solution as defined in the Order Form. They are provided in a good, diligent and professional manner in accordance with industry standards, as committed in clause 2.1 of the GTCs.

Included

Hosting and operation of the SaaS backend — API, voice agent worker, and manager dashboard.
Hosting and operation of the supporting databases and object storage.
Application of routine security patches, dependency updates, and bug fixes.
Monitoring and incident response on all of the above.

Out of scope

Customer-side hardware, network, browser, or identity-provider issues — see GTC §2.3 and the technical prerequisites listed in our Documentation.
Customisation, integration work, and training of Customer staff — these are Professional Services and are scoped separately in the Order Form.
Inference-provider outages cascading from Google, Fireworks AI, Supabase, LiveKit, or Fly.io infrastructure — see §6 below for our proactive-notification commitment.

2. Service availability target

Target: 99.5% monthly availability of the production Solution, measured at the application edge (Fly.io cdg Paris region).
Calculation window: calendar month (UTC).
Excluded from the calculation: planned maintenance windows announced at least 48 hours in advance; force-majeure events per GTC §14.4; upstream sub-processor outages as described in §6.
Availability reports are provided on request to the Customer's designated representative.

3. Planned maintenance window

Default window: Sundays 02:00–06:00 CET.
Notice: at least 48 hours, sent by email to the Customer's designated technical contact.
Emergency maintenance — security patches addressing an active exploit risk — may occur outside the window with as much notice as practicable.

4. Support tiers and response targets

Severities, response targets, and channels are summarised below. Targets are measured in business hours unless stated otherwise.

Severity	Definition	First response	Restoration target	Channel
P1 — Critical	Production Solution unavailable for all Authorized Users, or a confirmed data-security incident.	`4 business hours`	`24 hours` target restoration	Email support@sparred.ai with `[P1]` in the subject line.
P2 — High	Production Solution degraded for one or more Authorized Users — for example unacceptable voice-drill quality, coaching-pipeline lag > `24h`, or a partly broken manager dashboard.	`1 business day`	`5 business days` target restoration	Email support@sparred.ai.
P3 — Standard	Cosmetic issue, single-user inconvenience, or documentation request.	`5 business days`	Resolved in the next minor release.	Email support@sparred.ai.

Business hours are Monday–Friday 09:00–18:00 CET, excluding French public holidays.

Enterprise tier

Customers on the Enterprise tier, as identified in the Order Form, are assigned a named Customer Success Manager with a direct contact channel published at onboarding.

5. Reporting an incident

Send an email to support@sparred.ai from the Customer's designated representative.
Include: workspace identifier, Authorized User identifier, approximate first-occurrence time, reproduction steps if known, and a screenshot or transcript excerpt where relevant.
Sensitive content: do not include real customer personal data in the report. Reference the affected drill by drill ID instead.
Acknowledgement: Prestance Lab will acknowledge receipt within the first-response window listed in §4.

6. Upstream sub-processor outages

The Solution depends on a small set of sub-processors listed in our Data Processing Agreement. An outage at Google, Fireworks AI, Supabase, LiveKit, or Fly.io may cause Solution availability to drop below the 99.5% target. When this happens:

Prestance Lab will notify the Customer's designated representative within 1 hour of detection.
We will share status-page links and escalation channels where available.
Time during a confirmed upstream outage is excluded from the availability calculation in §2.

7. Evaluation Licenses

Per GTC §2.4, the SLAs in this document do not apply to Evaluation Licenses. Prestance Lab will nonetheless make commercially reasonable efforts to keep evaluation environments running and to acknowledge support requests within 5 business days. The disclaimer in GTC §7 — that the Solution is provided on a best-effort basis and that the Internet is not error-free — applies in full to evaluation use.

8. Updates and changes to these terms

Per GTC §2.7, Prestance Lab may update the contents of this page from time to time. Material changes will be notified to active Customers at least 30 days in advance, unless the change is required by applicable law or to address a material security risk. The "Last updated" date in the header reflects the most recent revision.

9. Contact

Support: support@sparred.ai
Status escalation: include [P1] in the subject line for production-down incidents.
Account and commercial matters: via the Customer's designated Prestance Lab contact.