Service Level Agreement Overview
This page is a commercial SLA overview for OneAI customers. A binding SLA requires a signed enterprise order form or master services agreement.
Scope
OneAI provides commercial API access for model routing, task intelligence, usage visibility, and Agent OS handoff records. External execution is performed by customer tools, OneClaw, OpenClaw, bots, or human operators.
Availability target
Production customers may request a written SLA. Default public service availability is provided on a commercially reasonable-efforts basis until a signed enterprise agreement is active.
Support response
Enterprise response targets can be defined by contract. Critical production incidents should include requestId, endpoint, provider, model, timestamp, and business impact.
Incident classification
Critical incidents affect production API availability or customer billing/security data. High incidents degrade key workflows. Normal incidents include configuration, provider, or task-specific support.
Customer responsibilities
Customers must keep API keys server-side, configure allowed usage, monitor spend, avoid abusive traffic, and provide accurate incident details for support.
Exclusions
Upstream provider outages, customer key misconfiguration, customer network failures, third-party executor failures, and misuse outside approved plan policy are excluded unless otherwise agreed.
Credits
Service credits, if any, require an enterprise agreement and are not automatically granted by the public self-serve plan.
What OneAI can measure for SLA review
Health endpoint, request success, error rate, latency, and provider availability.
Model catalog sync, provider configured state, pricing coverage, and per-model health checks.
Handoff contract creation, approval, proof callback, result callback, and proof review state.
API key events, customer usage, billing events, failed requests, and audit logs.