> ## Documentation Index
> Fetch the complete documentation index at: https://docs.odella.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Manager responsibilities

> Understand an AI Employee manager's authority, credential responsibilities, limits, and handover requirements.

An AI Employee's manager is the person accountable for how that worker is configured and used. In an organization account, assign the manager in **Settings → Governance**.

This is a permission-bearing relationship, not just a name on the worker's profile. Choose an active account member who can oversee the worker's responsibilities and the connections needed for unattended work.

## What the manager can do

The manager receives employee-level administration for that worker, including access outside its ordinary team or role audience. This authority covers the following areas, subject to applicable product restrictions such as release-managed configuration.

| Responsibility | Manager controls |
| - | - |
| **Use and supervise the worker** | Chat, request work, give feedback, and review available activity. |
| **Identity and instructions** | Manage the worker's profile, role, and durable working guidance. |
| **Capabilities** | Configure skills, tools, models, memory, and supporting resources, subject to those resources' own permissions. |
| **Availability** | Manage working hours and related work settings. |
| **Communication and safety** | Configure channels, inbound access, outbound policy, and approval rules. |
| **Recurring work** | Create, update, and remove scheduled or event-based work. |
| **Governance** | Manage the worker's audience, editing assignments, conversation-sharing setting, and manager assignment. |
| **Lifecycle** | Terminate the worker through the supported controls. |

Selected editors can receive permission to change particular settings areas. They do not become the accountable manager, and an editor assignment does not by itself grant access to the worker.

See [Governance](https://docs.odella.ai/workforce/governance.md) for the access and editing controls, and [Approvals](https://docs.odella.ai/workforce/approvals.md) for reviewing proposed actions. Manager status does not mean every action automatically pauses for approval: configure the policy deliberately.

## Why the manager matters for credentials

When a verified member asks for work interactively, that member supplies the user scope. The manager does not replace that identity merely because the member lacks a connection.

For governed background work, the manager supplies the user scope when automation is authorized. Admitted external contacts without a member identity can also use the manager as the accountable credential subject.

This has two practical consequences:

1. **User-scoped integrations:** The manager needs the appropriate connection and provider permissions for unattended work that uses it.
2. **Vault API keys:** The manager must explicitly share the required keys with the worker. Assigning a manager does not automatically share their Vault.

Account-scoped integrations use the account connection instead. The integration's defined scope determines the connection lookup; the worker cannot change it by asking to act as someone else.

> [!NOTE]
>
> Selecting the manager's credential scope is not the same as lending the
> manager's full permissions. Background work needs its own active
> authorization, and external work remains limited by its channel admission.
> Credentials alone do not authorize unrelated actions.

See [Credentials and user scope](https://docs.odella.ai/platform/credentials-and-scope.md) and [Vault and protected credentials](https://docs.odella.ai/platform/vault.md).

## What manager status does not grant

Managing one AI Employee does not automatically grant:

* Organization-wide administration, billing management, or authority over other workers.
* Access to every workflow, document, knowledge source, or connected application.
* The ability to inspect another person's Vault or retrieve reusable secret values through the protected Vault interface.
* Broader provider permissions than those granted to the connected identity or key.
* Permission to bypass tool restrictions, outbound-recipient rules, or required approvals.
* Unrestricted access to other members' conversations. Conversation use and visibility follow their own policies.

Configure conversation sharing separately, and do not use it as a substitute for a privacy or access review.

## Change managers safely

Follow [Hand over an AI Employee](https://docs.odella.ai/platform/manager-handover-tutorial.md) for a worked transfer, connection preparation, access-removal checks, and a handover acceptance checklist.

A manager change is a security handover, not just an organizational update. Changing the manager revokes affected employee automation grants; do not assume existing unattended work will continue without review and renewed authorization.

**1. Review the worker's responsibilities**

Inventory scheduled actions, event-based work, connected channels, sensitive
tools, and pending approvals.

**2. Prepare the replacement**

Confirm the new manager is an active account member. Connect the user-scoped
applications they need and explicitly share any required Vault keys with the
worker.

**3. Transfer the assignment**

Save the new manager in Governance. Review the worker's audience, selected
editors, and conversation-sharing choice as part of the handover.

**4. Review and reauthorize unattended work**

Review affected automations, restore authorization through the supported
controls, and test representative work before relying on it.

**5. Remove obsolete access**

Remove old Vault grants and unnecessary connections as appropriate. Revoke
provider credentials separately when required by your offboarding process.

If the manager leaves or the worker shows that reassignment is required, ask a person with manager-assignment permission to resolve it. Departure handling revokes affected automation authority; a missing manager should not be treated as permission to use another member's credentials.

Continue with [Organization administration](https://docs.odella.ai/platform/organization.md) and [Roles and permissions](https://docs.odella.ai/platform/roles-and-permissions.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.