Service Level Agreement · v1.0
What we commit to. What you’re entitled to. How we operate.
Breach disclosure
72h
Section 01 · Security
These commitments match the public posture at /security and the live attack log at /awais/incidents.
Why 72 hours on disclosure
Section 02 · Data
Your data belongs to you. The platform’s job is to never be the reason you cannot leave.
Account-closure return is not extraction
Section 03 · AI & model
Gladius runs a third-party AI model provider as the primary model tier. Some things about that are guaranteed; some things are not, and we do not pretend otherwise.
Model uptime
Inherits the upstream provider’s SLA. When the provider is up, we are up against their number; when it is down, we route through the fallback below.
Model fallback (automatic)
When the primary model is unavailable, we silently fall back to the next tier, a lower-cost tier, with no customer action required. The fall is logged and surfaced on /status.
Model accuracy
We do not guarantee AI output quality. High-stakes outputs (desking, F&I, and customer communication) route through human review. An operator does not delegate the close to a chatbot, and neither does our platform.
Section 04 · Exclusions
Honest carve-outs. None of these are loopholes. They are things genuinely outside our control or outside the scope of a standard contract.
Section 05 · SLA changes
Current version
Direct contact
Live platform telemetry lives at /status. Severity definitions live at /support. Security posture lives at /security. Changes to this SLA are published 30 days in advance at /changelog.