---
title: "Configure custom AI providers"
canonical: "https://support.appfire.com/space/RS/3484352515/Configure%20custom%20AI%20providers"
format: markdown
---
> Macro (aura-html)

Appfire QueueZero includes a platform-managed connection to Anthropic models. You can also connect QueueZero to an LLM provider managed by your organization.

Configure the organization provider from **Organization settings** > **Organization LLM provider**.

> 📝 The provider you save becomes the default for QueueZero spaces in the organization. Individual spaces can use a different provider when space-level provider overrides are allowed. See [Space settings](https://appfire.atlassian.net/wiki/spaces/RS/pages/3393486849) for more information.

QueueZero supports the following organization-managed providers:

- Anthropic
- Google Vertex AI
- AWS Bedrock

The following sections details how to configure each of the LLM providers. 

> ℹ️ Custom LLM connections, or **Bring your own LLM**, are available only on certain plans. Choosing your own AI model is also a plan-dependent feature. See [Billing and plans](https://appfire.atlassian.net/wiki/spaces/RS/pages/3293184199) for more information.

## Using Anthropic

To add your Anthropic key, select **Anthropic** from the **Provider **pulldown menu (Figure 1, below), then add your [API key](https://platform.claude.com/login?returnTo=%2Fsettings%2Fkeys). Your key must start with `sk-ant-`. Click **Save**. 

Your key will automatically be verified and if it is valid, you will see the success message `API key configured`. 

To update your API key, click **Update key**, enter a new key and click **Save**. 

![Add your Anthropic key to QueueZero](media://ac9785f4-2b71-4d6c-a73e-b7c24ddfbae8)

## Using Google Vertex

To connect your Vertex AI account, select **Vertex AI** from the **Provider **pulldown menu (Figure 2, below) and click **Configure Vertex**. Add the following: 

- **Project ID**
- **Project number**
- **Region**
- **WIF pool ID**
- **WIF provider ID**
- **Service account email**

> ℹ️ ### Pin model versions
> ℹ️ 
> ℹ️ By default, QueueZero uses the supported model versions available through your Vertex AI configuration.
> ℹ️ 
> ℹ️ To use specific Claude model versions, expand **Advanced** and configure the versions QueueZero should use for the supported Sonnet, Haiku, or Opus models.

Click **Test connection** to verify your configuration. Click **Save** to complete the configuration and set Vertex as your organization’s default model. 

## Using AWS Bedrock

To connect your AWS Bedrock account, select **AWS Bedrock** from the **Provider **pulldown menu (Figure 2, below) and click **Configure AWS Bedrock**. 

Before saving the connection, you must configure AWS so QueueZero can assume a role in your account and access the required Bedrock models.

### Configure AWS access

The QueueZero configuration screen provides the AWS policy information required for the connection.

Use the values shown in the product to configure:

- the IAM role trust policy
- the permissions policy
- access to the required Claude models in Amazon Bedrock

Use the **Copy** actions in QueueZero when creating the policies in AWS. This ensures you use the values generated for your environment.

In Amazon Bedrock, also make sure the Claude models required by your QueueZero configuration are enabled under **Model access**.

### Enter your AWS configuration

Back in QueueZero, enter:

![Add AWS Bedrock to QueueZero](media://019b1972-48fc-4c64-9c3f-5712bd7f0ab4)

- **AWS region –** The AWS region QueueZero will use for Claude requests. Select a region where the required Anthropic Claude models and prompt caching are available through Bedrock.
- **AWS account ID** – Your 12-digit AWS account ID.
- **Role name** – The name of the IAM role QueueZero will assume in your AWS account. Enter the role name only; QueueZero constructs the full role ARN.
- **External ID** – A shared value used to protect the cross-account role from confused-deputy attacks. You must use the same value in the role’s `sts:ExternalId` trust-policy condition.

### Pin model versions

If you need QueueZero to use specific Claude model versions, expand **Advanced** and configure the supported Sonnet, Haiku, or Opus versions.

### Test and save the connection

Select **Test connection** to verify that QueueZero can assume the configured IAM role and access the required Bedrock models.

When the test succeeds, click **Save**.

Saving the provider makes AWS Bedrock the default LLM provider for the organization.

### Edit an existing Bedrock connection

When you edit a saved Bedrock configuration, the **External ID** field may appear empty. QueueZero does not display the saved External ID again.

If you are changing the configuration, enter the External ID required by your AWS trust policy before saving.