Skip to main content

Create a Client

This guide explains how a platform administrator creates a client on a PaletteAI Inference Launchpad appliance. A client is a named team, business unit, or workload, represented by one or more API tokens. To understand what a client is and how clients, API tokens, and quotas relate, refer to Clients and Quotas.

To generate an API token by itself, without stepping through the full client setup, refer to Generate an API Token.

Prerequisites​

  • A running PaletteAI Inference Launchpad appliance, with the console reachable.
  • Console access with permission to manage clients. Managing clients can require operator access.
  • A workspace to hold the client. To give a team its own workspace, refer to Create a Workspace.

Create a Client​

Create a client through the Add client wizard. The wizard names the client, optionally sets a quota, optionally grants model access, and optionally issues the client's first API token.

  1. In the workspace drop-down menu in the console header, select the workspace that holds the client.

  2. From the left main menu, select Access & Policy. The Clients list opens.

  3. Select Add client. The Add client wizard opens on the Overview step.

  4. On the Overview step, enter a Client name. To give the client its own egress and routing settings instead of its policy's, turn on Override policy settings. Then select Next step. The appliance assigns the client an immutable identifier and registers it as active.

  5. (Optional) On the Quotas step, add usage limits for the client, and then select Next step. For details, refer to Set and Manage Client Quotas.

  6. (Override policy settings only) On the Egress step, turn on Enable external inferencing, and add each external inference endpoint the client can reach. Then select Next step. Without Override policy settings, the client follows its workspace policy's external inferencing settings. For the endpoint table, the Key field, and the Daily limit, refer to Manage a Client's Model Access.

  7. (Optional) On the Routing step, leave the Tier map unchanged to route the client with the appliance's default model routing, or edit the Tier map to route the client's model aliases to specific models. Then select Next step. For details, refer to Manage a Client's Model Access.

  8. On the API tokens step, select Add API Token. In the Add API token dialog, optionally enter a Label, choose an Expires date, and then select Add Token. Expires is required, and the console accepts a date up to 366 days ahead. The token stops working at the end of the day you choose.

  9. Select Create client.

  10. When the console reveals the token, select Copy. The token begins with lpai_. To read the connection steps for a coding assistant without leaving the dialog, select Connect Agent.

warning

The console shows the token only once and stores only a hash of it. Copy it now. If you lose it, revoke the token and create a new one.

A client can hold more than one API token. To add another token to a client later, refer to Generate an API Token.

Verify the Client​

Confirm that the appliance registered the client and its API token.

  1. In the console header, confirm that the workspace drop-down menu shows the workspace that holds the client, and then from the left main menu, select Access & Policy.

  2. On the Clients list, confirm that the new client appears under the name you gave it.

  3. Select the client to open its detail panel, and then select the API tokens section.

  4. Confirm the token you added appears with a state of active.

The client is ready to use once its token is active. To confirm the token works from end to end, connect a coding assistant to the appliance with it, as described in Use Claude Code.

Next Steps​

After you create a client, configure how much it can consume and which models it can reach.