Core Positions
7 key contract terms with defined positions and flexibility levels
| Term | Standard Position | Rationale | Flexibility | Escalation |
|---|---|---|---|---|
| Accuracy & Performance | Commercially reasonable efforts to maintain accuracy; no absolute guarantees on AI output precision | AI outputs are probabilistic by nature. Absolute accuracy warranties create unlimited liability exposure. | Low | VP Legal |
| Data Usage & Model Training | Aggregated, de-identified data may be used for model improvement; customer data never shared with third parties | Model improvement benefits all customers. Opt-out provisions weaken the collective value proposition. | Medium | Chief Privacy Officer |
| Human-in-the-Loop Controls | Configurable HITL controls available; customer responsible for determining appropriate oversight level | The platform provides the tools; the customer must define governance based on their risk appetite and regulatory requirements. | High | None |
| Intellectual Property | Customer owns their input data and custom configurations. The provider retains ownership of models, algorithms, and platform IP. | Clear IP boundaries protect both parties. AI outputs generated by the models are licensed, not transferred. | Low | General Counsel |
| Liability & Indemnification | Mutual limitation of liability with standard cap. AI-specific indemnification limited to the provider's negligence or willful misconduct. | Unlimited AI liability is commercially unviable. Risk should be allocated proportionally to each party's control. | Low | General Counsel |
| Regulatory Compliance | Compliance-enabling features are provided. The customer is responsible for jurisdiction-specific compliance determinations. | The platform operates globally. Individual customer compliance obligations vary dramatically by jurisdiction and industry. | Medium | VP Legal |
| Service Level Agreements | Standard SLA framework with uptime commitments. AI-specific performance SLAs measured by availability, not output quality. | AI output quality depends on input data quality and customer configuration. SLAs should measure what the provider controls. | Medium | VP Engineering |
Accuracy & Performance
Commercially reasonable efforts to maintain accuracy; no absolute guarantees on AI output precision
AI outputs are probabilistic by nature. Absolute accuracy warranties create unlimited liability exposure.
Escalation: VP Legal
Data Usage & Model Training
Aggregated, de-identified data may be used for model improvement; customer data never shared with third parties
Model improvement benefits all customers. Opt-out provisions weaken the collective value proposition.
Escalation: Chief Privacy Officer
Human-in-the-Loop Controls
Configurable HITL controls available; customer responsible for determining appropriate oversight level
The platform provides the tools; the customer must define governance based on their risk appetite and regulatory requirements.
Escalation: None
Intellectual Property
Customer owns their input data and custom configurations. The provider retains ownership of models, algorithms, and platform IP.
Clear IP boundaries protect both parties. AI outputs generated by the models are licensed, not transferred.
Escalation: General Counsel
Liability & Indemnification
Mutual limitation of liability with standard cap. AI-specific indemnification limited to the provider's negligence or willful misconduct.
Unlimited AI liability is commercially unviable. Risk should be allocated proportionally to each party's control.
Escalation: General Counsel
Regulatory Compliance
Compliance-enabling features are provided. The customer is responsible for jurisdiction-specific compliance determinations.
The platform operates globally. Individual customer compliance obligations vary dramatically by jurisdiction and industry.
Escalation: VP Legal
Service Level Agreements
Standard SLA framework with uptime commitments. AI-specific performance SLAs measured by availability, not output quality.
AI output quality depends on input data quality and customer configuration. SLAs should measure what the provider controls.
Escalation: VP Engineering