This page describes how JibeONE's AI capabilities work, what data they use, and the commitments we make about them. It is written for buyers and security teams evaluating AI governance.
1. What Jules is
Jules is JibeONE's built-in assistant. It operates inside the platform, on your real records, within your permission model and audit trail. Jules can answer questions about your data, draft content, build views, and prepare actions for a person to review. It is a feature of the operating system, not a separate product that copies your data elsewhere.
2. Model providers
Today, Jules uses OpenAI's Responses API as its underlying model provider. The integration is built through a provider adapter, so the model backend can change over time; when it does, this page is updated. Where a customer deploys on-premises or in a private cloud, AI features are enabled only when the customer configures a provider.
3. No training on your data
We do not use your records or prompts to train foundation models, and our agreement with our AI provider prohibits the provider from using content submitted through the API to train its models. AI requests carry only the context needed to answer the task at hand.
4. Tenant isolation
AI requests are scoped to the requesting user's own instance and permissions. Jules cannot read across tenants, and it cannot see records the requesting user is not authorized to see — AI queries pass through the same role-based access control and row-level data filters as every other read in the platform.
5. Human in the loop
Jules is designed to draft and propose, not to act unilaterally. Actions that create or change records surface for a person to review and approve before they take effect, so a human remains accountable for outcomes.
6. Disable and opt out
AI features can be disabled at the instance level by an administrator. Where features are off, no content is sent to an AI provider.
7. Data residency and logging
The region in which AI requests are processed depends on the configured provider and deployment. Prompt and response activity is logged for security, abuse-prevention, and quality purposes consistent with our
Privacy Policy; logs are retained under the same retention practices as other platform activity.
8. Abuse controls and evaluation
AI endpoints are subject to rate limiting and abuse controls. Prompts that drive AI behavior are managed as versioned platform content rather than being buried in code, so changes are reviewable and controlled. We evaluate AI features for quality and safety before release and as behavior changes.
For AI-specific questions, including our provider terms and no-training commitment, contact
hello@jibe.one. Our AI provider is listed on the
Sub-processors page.
