> ## Documentation Index > Fetch the complete documentation index at: https://docs.prisme.ai/llms.txt > Use this file to discover all available pages before exploring further. # One organization or several? > The structure decision most teams hesitate on, and where groups fit. Before you open the platform to teams, you decide how to structure it. This is the question most organizations hesitate on: a single organization with groups, or several organizations? One org with groups gives a shared catalog, asset reuse, groups for members and permissions, and lighter administration. Multiple organizations give hard isolation of data, quotas, billing, and models, per-subsidiary compliance, and independent catalogs. Choose multiple orgs only for hard separation. ## How to decide Keep a shared foundation. **Groups** organize members and permissions, and [membership rules](/products/ai-governance/identity-access) auto-assign roles as people join. Best when teams collaborate and reuse each other's assets. Separate everything: data, quotas, billing, and available models. Best for subsidiaries or business units with their own compliance perimeter and no cross-sharing. Decide on a few concrete criteria: * **Data segregation**: must data never cross a boundary? That points to multiple orgs. * **Billing and chargeback**: do you need fully separate cost attribution? Multiple orgs make it clean. * **Compliance perimeter**: different regulatory scopes per entity? Multiple orgs. * **Reuse and collaboration**: do teams share agents, skills, and tools? One org with groups keeps that simple. **Recommendation:** choose multiple organizations only when you need hard separation of data, billing, or compliance. Otherwise, one organization with groups keeps reuse and administration simple. The two levels of governance (central and per-org) apply either way, see [Governe](/products/ai-governance/overview). ## Next steps The teams and rights per level Open to all or curate, without chaos