Skip to main content

How Wirevox is organized

Wirevox is workspace-scoped. A workspace is the security, ownership, and billing boundary for its agents and operational data.

Platform hierarchy​

Workspace
├── Members and permissions
├── Agents
│   ├── Draft configuration
│   ├── Published configuration
│   ├── Channel, model, and voice
│   ├── Attached knowledge bases
│   └── Attached functions
├── Shared resources
│   ├── Knowledge bases
│   ├── Reusable functions
│   └── Integration connections
├── Delivery
│   ├── Phone numbers
│   └── Chat deployments
├── Operations
│   ├── Calls and chat conversations
│   ├── Contacts
│   └── Logs and analytics
└── Billing
    ├── Included credits
    ├── Purchased credits
    └── Usage ledger

Workspaces are hard boundaries​

Resources from one workspace are not automatically available in another. Before Wirevox reads or changes a resource, the platform verifies that the current user can access the active workspace and has a permission accepted by that operation.

Switching the active workspace changes the scope of agents, knowledge, integrations, contacts, analytics, and billing information shown in the application.

See Workspaces, members, and permissions for workspace types, switching, the enforced role matrix, secure invitations, agency subaccounts, leaving, and deletion.

Agents reference shared resources​

Knowledge bases, functions, and integration connections are managed as workspace resources. An agent can reference compatible resources from the same workspace.

This separation allows a reusable function or knowledge base to support more than one agent without duplicating the underlying resource. Removing or changing a shared resource can therefore affect multiple agents and should be reviewed carefully.

Connection, function, and prompt instructions are separate​

For an integrated action to work correctly, three layers may be required:

  1. Connection: the workspace authorizes the external provider.
  2. Reusable function: the workspace selects which supported provider actions can be used.
  3. Agent configuration: the function is attached to a compatible agent and its system prompt contains exact operational instructions.

Completing one layer does not automatically complete the others.

Conversation records are operational data​

Calls and chat conversations are records created while agents run. They can contain status, duration, messages or transcripts, function activity, outcomes, and errors, depending on the channel and runtime path.

These records are not agent instructions. Wirevox treats transcript and message content as untrusted user data when it is later reviewed by Vox or another automated system.

Current state versus historical usage​

The current credit balance is a live workspace billing value. Historical credits used are cumulative records and must not be subtracted from an already net current balance.

When reviewing an agent, also distinguish between its current draft, its published version, and historical conversation records created by older versions.