Per-Collaborator AI Agent Permission

🔍 Problem Statement

As a Pelcro account owner, I have no way to control whether the AI Agent is enabled on my account or which of my collaborators can use it, which prevents me from managing data access and consumption at both the account and per-user level.

💡 User Stories

1. Account-level enablement and consent. As an account owner, I want to enable the AI Agent for my account through a one-time consent prompt, so the feature is unavailable to anyone on my team until I explicitly turn it on.

2. Per-collaborator toggle. As an account owner, once the AI Agent is enabled for my account, I want to enable or disable it for each individual collaborator from Settings > Account Collaborators, so I control who on my team can use the agent.

3. Permissions toggle idea > preline-pro-templates/dashboard/account-notifications.html

4. Visible but gated entry point. As a collaborator without AI Agent permission, I want to still see the AI Agent in the UI but be told to contact my account owner when I try to chat, so the path to enablement is obvious.

🎯 Definition of Done (DoD)

1. Upgrade the collaborator form to preline.

2. Only the account owner can enable the AI Agent at the account level; collaborators have no access to this control.

3. On the account owner's first AI Agent access, a consent prompt appears: "Do you want Pelcro's AI Agent to access your data?" — the agent stays disabled for the entire account until accepted.

4. Once enabled at the account level, AI Agent access is controlled per collaborator and defaults to enabled for the account owner and disabled for all other collaborators.

5. The account owner can toggle AI Agent access per collaborator from Settings > Account Collaborators.

6. This toggle is visible for all but cannot be edited except by owner.

7. A collaborator without permission still sees the AI Agent panel, but any message they send returns: "Please ask your account owner to enable this feature for you."

8. when a collaborator already has an open AI Agent session and their access gets revoked mid-session, this message should be returned in this scenario as well.

9. A collaborator with permission uses the agent normally.

10. Owners who refuse to give consent initially should be able to re-enable it later.

11. Toggling a collaborator's access takes effect on their next interaction without requiring re-login.