Version: 2026-08-08 v2.0
AI outputs may be inaccurate, incomplete, outdated, biased, misleading, non-unique or unsuitable. Users must review material outputs before relying on them.
Customer administrators are responsible for configuring permissions, approval thresholds, financial limits, autonomous boundaries, human-review requirements and lawful use.
Julia must not be used as the sole decision-maker for legal, financial, accounting, tax, employment, recruitment, disciplinary, medical, insurance, credit, safety-critical or other high-impact matters where qualified human review is required.
Users must not configure the Services for unlawful, prohibited or unauthorised high-risk AI uses, including prohibited biometric, emotion-inference, social-scoring, exploitative or unlawful automated decision practices.
Where law requires disclosure that a user is interacting with AI or receiving AI-generated content, Customer is responsible for providing that disclosure to its own users and recipients.
Momentum may restrict, pause or block activity that reasonably appears unlawful, prohibited, abusive or materially unsafe. Such controls may be incomplete and do not replace Customer's own oversight duties.
Julia may automatically select prompts, providers, models and supported model parameters for a task. Selection may consider capability, task fit, risk, quality, cost, latency, availability and jurisdiction.
Higher-cost or stronger models are not automatically preferred where a lower-cost model is reasonably suitable for the task.
High-impact or regulated tasks remain subject to applicable human-review requirements regardless of model selection.