> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://help.sigmacomputing.com/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://help.sigmacomputing.com/_mcp/server. # Configure an external AI provider > Set up external AI providers like OpenAI, Anthropic, Gemini, or Amazon Bedrock to power Sigma organization AI features. When you [configure AI features for your organization](/docs/configure-ai-features-for-your-organization), you can set up an external AI provider to use AI models provided by a third party, such as OpenAI, Anthropic, Gemini, or Amazon Bedrock. For the models that Sigma uses with each provider, see [Supported AI models](/docs/supported-ai-models). The use of AI features is subject to the following [disclaimer](/docs/notice-for-enabling-ai-enabled-features-in-sigma). You can integrate Sigma with the following external AI providers: * [OpenAI](#add-openai-as-an-ai-provider) * [Azure OpenAI Foundry](#add-azure-openai-as-an-ai-provider) * [Gemini through Google](#add-google-gemini-as-an-ai-provider) * [Anthropic](#add-anthropic-as-an-ai-provider) (Beta) * [Amazon Bedrock](#add-amazon-bedrock-as-an-ai-provider) Sigma recommends a token limit of at least 100,000 tokens per minute (TPM) for an external AI provider. For any AI provider, you can optionally [configure a separate AI provider to use for embeddings](#configure-an-embeddings-provider), which Sigma uses to index your data sources for semantic search. Some providers, such as Anthropic, require one. After setting up an AI provider, you can [change the provider or remove an existing setup](#manage-an-external-ai-provider). ## User requirements * You must be assigned the **Admin** [account type](/docs/create-and-manage-account-types). ## Add OpenAI as an AI provider If you have an OpenAI API key, set up Sigma to use the supported OpenAI model for all AI-powered features. ### Retrieve OpenAI credentials To integrate with OpenAI directly, you must obtain, manage, and secure an OpenAI API key. You can retrieve the API key in the API keys section of the OpenAI developer platform. For instructions, see [Where do I find my OpenAI API key?](https://help.openai.com/en/articles/4936850-where-do-i-find-my-openapi-api-key) in the OpenAI documentation. Sigma uses the most capable supported OpenAI model available in your account. For the models that Sigma uses, see [Supported AI models](/docs/supported-ai-models). ### Add OpenAI as an AI provider to Sigma To add OpenAI as an external AI provider for your organization: 1. Go to **Administration** > **AI settings**: 1. In the Sigma header, click your user avatar to open the user menu. 2. Select **Administration** to open the **Administration** portal. 3. In the side panel, select **AI settings**. 2. In the **AI provider** section, for **Provider hosting**, select the **External models** option. 3. For **AI provider**, select **OpenAI**. 4. In the **API key** field, enter the OpenAI API key. 5. Click **Save** to authenticate. After you save, the **API key** field appears blank and Sigma does not display the key. After the integration is successfully authenticated, AI functionality is available for your organization. ## Add Azure OpenAI as an AI provider If you access OpenAI models through Azure OpenAI in Microsoft Foundry, set up Sigma to use the supported models for AI-powered features. ### Set up Azure OpenAI models To use OpenAI through Azure OpenAI in Microsoft Foundry Models, you must have the following: * An Azure subscription with access to Azure OpenAI. * Both a reasoning model and an embeddings model deployed in Microsoft Foundry. For the models to deploy, see [Supported AI models](/docs/supported-ai-models). Both models are required to use AI features in Sigma. If needed, review the [prerequisites](https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/how-to/deploy-foundry-models?view=foundry-classic#prerequisites) listed in the Deploy Microsoft Foundry Models in the Foundry portal page of the Microsoft Azure documentation. To integrate Azure OpenAI in Microsoft Foundry Models with Sigma, you must provide a deployment ID, endpoint URL, and API key for each required model. You can retrieve the required information from Microsoft Foundry. See [Manage models](https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/how-to/deploy-foundry-models?view=foundry-classic#manage-models) in the Microsoft Azure documentation. 1. Log in to Microsoft Foundry. 2. Select your project. 3. On the **Models + endpoints** page, retrieve the relevant details: | Details | Where to find | | ---------- | ---------------------------------------------------------------------------------- | | Model name | Listed as the **Model name** for a given resource. | | Endpoint | Next to the resource name, click **Get endpoint** and copy the **Endpoint** value. | | Key | Next to the resource name, click **Get endpoint** and copy the **Key** value. | In Sigma, use the values as follows: * **Model name** as the **Deployment ID** * **Endpoint** as the **Endpoint URL** * **Key** as the **API Key** For more details, see [Deploy Microsoft Foundry Models in the Foundry portal](https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/how-to/deploy-foundry-models?view=foundry-classic) in the Microsoft Azure documentation. ### Add Azure OpenAI as an AI provider in Sigma To add Azure OpenAI as an external AI provider for your organization: 1. Go to **Administration** > **AI settings**: 1. In the Sigma header, click your user avatar to open the user menu. 2. Select **Administration** to open the **Administration** portal. 3. In the side panel, select **AI settings**. 2. In the **AI provider** section, for **Provider hosting**, select the **External models** option. 3. For **AI provider**, select **Azure OpenAI**. 4. For each model, provide the [required details](#set-up-azure-openai-models): * For **Deployment ID**, enter the model name. * For **Endpoint URL**, enter the endpoint. * For **API key**, enter the key for the endpoint. If the language model and embeddings model are in the same project, the endpoint URL and key are the same for both models. 5. Click **Save** to authenticate. After you save, Sigma does not display the values of your credentials. If Sigma displays a "Failed to validate" error, you might have restricted access to your Azure instance using an IP allowlist. To fix this issue: 1. Retrieve the IP addresses used by Sigma. See [Add Sigma IPs to the allowlist](/docs/connect-to-data-sources#add-sigma-ips-to-the-allowlist). 2. Update your allowlist in the Azure portal. See [Grant access from an internet IP range](https://learn.microsoft.com/en-us/azure/ai-services/cognitive-services-virtual-networks?tabs=portal#grant-access-from-an-internet-ip-range) in the Microsoft Azure documentation. After the integration is successfully authenticated, AI functionality is available for your organization. ## Add Google Gemini as an AI provider If you use Google Gemini, set up Sigma to use the supported model for AI-powered features. Sigma uses the [global endpoint](https://docs.cloud.google.com/gemini-enterprise-agent-platform/resources/locations) to access the supported model. If you have regional restrictions for AI data processing, consider using a different AI provider. ### Google Gemini credentials To use Google Gemini, you must obtain, manage, and secure a Gemini API key. You can retrieve the API key in the [API keys](https://aistudio.google.com/api-keys) section of Google AI Studio. For the model that Sigma uses, see [Supported AI models](/docs/supported-ai-models). When using AI features in Sigma, consider the rate limits set by Gemini. For more details, see [Rate limits](https://ai.google.dev/gemini-api/docs/rate-limits) in the Gemini API documentation. ### Add Gemini as an AI provider to Sigma To add Google Gemini as an external AI provider for your organization: 1. Go to **Administration** > **AI settings**: 1. In the Sigma header, click your user avatar to open the user menu. 2. Select **Administration** to open the **Administration** portal. 3. In the side panel, select **AI settings**. 2. In the **AI provider** section, for **Provider hosting**, select the **External models** option. 3. For **AI provider**, select **Gemini**. 4. For **API key**, enter the Gemini API key. 5. Click **Save** to authenticate. After you save, the **API key** field appears blank and Sigma does not display the key. After the integration is successfully authenticated, AI functionality is available for your organization. ## Add Anthropic as an AI provider This documentation describes one or more public beta features that are in development. Beta features are subject to quick, iterative changes; therefore the current user experience in the Sigma service can differ from the information provided in this page. This page should not be considered official published documentation until Sigma removes this notice and the beta flag on the corresponding feature(s) in the Sigma service. For the full beta feature disclaimer, see [Beta features](/docs/sigma-product-releases#beta-features). If you have an Anthropic API key, set up Sigma to use the supported Anthropic model for AI-powered features. Anthropic does not provide an embeddings model that Sigma can use for [semantic search](/docs/search-in-sigma). After you configure Anthropic as your AI provider, you must also configure a [separate AI provider](/docs/configure-ai-features-for-your-organization#set-up-an-ai-provider) to use for embeddings. ### Retrieve Anthropic credentials To integrate with Anthropic, you must obtain, manage, and secure an Anthropic API key. You can retrieve the API key in the API keys section of the Anthropic console. For instructions, see [Authentication](https://platform.claude.com/docs/en/manage-claude/authentication) in the Anthropic documentation. For the model that Sigma uses, see [Supported AI models](/docs/supported-ai-models). ### Add Anthropic as an AI provider to Sigma To add Anthropic as an external AI provider for your organization: 1. Go to **Administration** > **AI settings**: 1. In the Sigma header, click your user avatar to open the user menu. 2. Select **Administration** to open the **Administration** portal. 3. In the side panel, select **AI settings**. 2. In the **AI provider** section, for **Provider hosting**, select **External models**. 3. For **AI provider**, select **Anthropic**. 4. In the **API key** field, enter the Anthropic API key. 5. Click **Save** to authenticate. After you save, the **API key** field appears blank and Sigma does not display the key. 6. Configure a separate embeddings provider. Anthropic does not provide an embeddings model that Sigma can use, so you must complete this step. See [Configure an embeddings provider](#configure-an-embeddings-provider). After both the AI provider and embeddings provider are successfully configured, AI functionality is available for your organization. ## Add Amazon Bedrock as an AI provider You can use Amazon Bedrock as your AI provider and power Sigma AI features. You can use Amazon Bedrock models directly by setting up a role for Sigma to use in Amazon Web Services (AWS). To set up Amazon Bedrock as an AI provider, complete the following steps in your AWS console and Sigma: 1. Review the [Amazon requirements](#amazon-requirements). 2. [Step 1: Create a role in AWS](#step-1-create-a-role-in-aws). 3. [Step 2: Add Amazon Bedrock as an AI provider in Sigma](#step-2-add-amazon-bedrock-as-an-ai-provider-in-sigma). 4. [Step 3: Set up a trust relationship in AWS](#step-3-set-up-a-trust-relationship-in-aws). 5. [Step 4: Test the AI provider connection](#step-4-test-the-ai-provider-connection). ### Amazon requirements You must have access to Amazon Bedrock and have permissions to create roles in the AWS console. To use Anthropic models with Amazon Bedrock, you must submit a one-time request with a use case for model access. See [Request access to models](https://docs.aws.amazon.com/bedrock/latest/userguide/model-access.html) in the Amazon Bedrock documentation. When you set up Amazon Bedrock as your AI provider, Sigma accesses the following types of models using the Amazon Bedrock runtime API: * [Anthropic](https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html#model-regions-anthropic) foundation models * [Amazon Titan Text Embeddings models](https://docs.aws.amazon.com/bedrock/latest/userguide/titan-embedding-models.html) For the specific models that Sigma uses for inference and embeddings, including planned models, see [Supported AI models](/docs/supported-ai-models). For details about all Anthropic models, see [Anthropic](https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html#model-regions-anthropic) in the Regional availability page in the Amazon Bedrock documentation. If a supported model is not available in your AWS region, set up [Cross-region inference](https://docs.aws.amazon.com/bedrock/latest/userguide/cross-region-inference.html) to access it from your geographic region. If data residency is not a requirement, set up global inference instead. ### Step 1: Create a role in AWS To use your Amazon Bedrock models with Sigma, you must create a role to allow a Sigma user for your organization to access Amazon Bedrock model resources. In your AWS console, follow the steps to [Create a role to give permissions to an IAM user](https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_create_for-user.html) in the AWS Identity and Access Management User Guide: * Create the role for **This account**. * Skip adding permissions. You add permissions to the role in [Step 3: Set up a trust relationship in AWS](#step-3-set-up-a-trust-relationship-in-aws). * Do not require MFA. Sigma uses an API to assume the role and cannot use multi-factor authentication to sign in. * When prompted, create a policy. Set up an IAM policy for the role that allows the [required actions](#required-iam-policy-actions) and [resources](#required-iam-policy-resources), or refer to the example: [Example IAM policy](#example-iam-policy). For guidance creating IAM policies, see [Define custom IAM permissions with customer managed policies](https://docs.aws.amazon.com/IAM/latest/UserGuide/access_policies_create.html) in the AWS Identity and Access Management User Guide. For the specific action and resource options available for Amazon Bedrock, see [Actions, resources, and condition keys for Amazon Bedrock](https://docs.aws.amazon.com/service-authorization/latest/reference/list_amazonbedrock.html) in the AWS Service Authorization Reference. * Choose an identifiable role name. For example, `sigma_bedrock` to indicate that this is a role used by Sigma to access Amazon Bedrock. #### Required IAM policy actions The IAM policy that you set up must allow the following actions: * `bedrock:InvokeModel` * `bedrock:InvokeModelWithResponseStream` For full details on those IAM actions, see [Permissions Reference for Amazon Bedrock](https://aws.permissions.cloud/iam/bedrock) in the Amazon Bedrock documentation. #### Required IAM policy resources The IAM policy that you set up must specify the following resources: * [Anthropic](https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html#model-regions-anthropic) foundation models * [Amazon Titan Text Embeddings models](https://docs.aws.amazon.com/bedrock/latest/userguide/titan-embedding-models.html) * An inference profile that provides access to Anthropic models. For details about inference profiles, see [Set up a model invocation resource using inference profiles](https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles.html) in the Amazon Bedrock User Guide. To track model usage from Sigma, you can create an application inference profile and reference that inference profile resource in your IAM policy. See [Create an application inference profile](https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles-create.html) in the Amazon Bedrock User Guide. For example, for an inference option for an EU region with data residency requirements, specify the following resources: * `arn:aws:bedrock:eu-central-1::foundation-model/anthropic.*` * `arn:aws:bedrock:eu-central-1:*:inference-profile/eu.anthropic*` * `arn:aws:bedrock:eu-central-1::foundation-model/amazon.titan-embed-text-*` Sigma recommends using a wildcard to refer to the model resources to allow Sigma to upgrade models without requiring you to update your IAM policy in AWS. #### Example IAM policy If your AWS region is in the United States, you can copy the following IAM policy JSON: ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": [ "bedrock:InvokeModel", "bedrock:InvokeModelWithResponseStream" ], "Resource": [ "arn:aws:bedrock:us-*::foundation-model/anthropic.*", "arn:aws:bedrock:us-*:*:inference-profile/us.anthropic.*", "arn:aws:bedrock:us-*::foundation-model/amazon.titan-embed-text-*" ] } ] } ``` ### Step 2: Add Amazon Bedrock as an AI provider in Sigma To add Amazon Bedrock as an external AI provider for your organization: 1. Go to **Administration** > **AI settings**: 1. In the Sigma header, click your user avatar to open the user menu. 2. Select **Administration** to open the **Administration** portal. 3. In the side panel, select **AI settings**. 2. In the **AI provider** section, for **Provider hosting**, select the **External models** option. 3. For **AI provider**, select **Amazon Bedrock**. 4. For **AWS IAM role ARN**, enter the Amazon Resource Name (ARN) of the role to which you assigned the IAM policy with access to invoke Amazon Bedrock models. 5. For **AWS region**, specify the region of the models that you want to access. The region you specify must match the region associated with the model resource paths in the IAM policy. For example, `eu-central-1`. 6. Click **Save**. 7. After you save, Sigma displays an **Amazon Bedrock external ID**. Copy the external ID and store it somewhere secure. After storing the external ID, click **Close**. 8. Review the AI provider details and copy the **AWS IAM user ARN** for your Sigma organization. You need the **Amazon Bedrock external ID** and **AWS IAM user ARN** in [Step 3: Set up a trust relationship in AWS](#step-3-set-up-a-trust-relationship-in-aws). ### Step 3: Set up a trust relationship in AWS In the AWS console, use the external ID and IAM user ARN associated with your Sigma organization to set up a trust relationship. The trust relationship allows the Sigma IAM user to assume the role associated with the Amazon Bedrock IAM policy. Follow the steps to [Update a role trust policy](https://docs.aws.amazon.com/IAM/latest/UserGuide/id_roles_update-role-trust-policy.html) in the AWS Identity and Access Management User Guide. For the trust policy, use a policy like the following example, replacing `` with the **AWS IAM user ARN** and `` with the **Amazon Bedrock external ID** values copied in [Step 2: Add Amazon Bedrock as an AI provider in Sigma](#step-2-add-amazon-bedrock-as-an-ai-provider-in-sigma): ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Principal": { "AWS": "" }, "Action": "sts:AssumeRole", "Condition": { "StringEquals": { "sts:ExternalId": "" } } } ] } ``` ### Step 4: Test the AI provider connection After setting up a trust relationship in AWS to allow Sigma to use Amazon Bedrock models, test the AI provider connection: 1. Go to **Administration** > **AI settings**: 1. In the Sigma header, click your user avatar to open the user menu. 2. Select **Administration** to open the **Administration** portal. 3. In the side panel, select **AI settings**. 2. In the **AI provider** section, click **Test connection**. If the connection is successful, you see a message like "Connection verified: Sigma successfully assumed your IAM role and reached Amazon Bedrock." If you see an error, review your IAM policy settings or contact [Sigma Support](/docs/submit-a-support-request). ## Configure an embeddings provider Sigma uses an embeddings model to index your data sources for [semantic search](/docs/search-in-sigma). By default, Sigma uses your AI provider to generate embeddings. If your AI provider does not support embeddings, or you want to use a different provider for embeddings, you can configure a separate embeddings provider. Anthropic does not provide an embeddings model that Sigma can use. If you use Anthropic as your AI provider, you must configure a separate embeddings provider. To configure a separate embeddings provider: 1. Go to **Administration** > **AI settings**. 2. In the **Embeddings provider** section, turn off the **Use the configured AI provider** toggle. If your AI provider does not support embeddings, this option is unavailable and you must configure a separate embeddings provider. 3. For **Provider hosting**, select one of the following, then configure your selected provider: * **External models**: For **AI provider**, select **OpenAI** or **Gemini**, then enter the **API key**. * **Data warehouse hosted model**: In the **Connection** dropdown menu, select a connection. For BigQuery connections, you can also select a **Region**. For connection and model requirements, see [Configure a warehouse-hosted model as AI provider](/docs/configure-warehouse-ai-model-integration). 4. Click **Save**. When you add or change the embeddings provider, Sigma reindexes searchable sources in the background. ## Manage an external AI provider You can edit the external AI provider integration at any time — for example, to update your credentials or switch to a different AI provider: 1. Go to **Administration** > **AI settings**. 2. In the **AI provider** section, click **Edit**. 3. Make the desired changes. For example, change the AI provider or add a new API key. 4. Click **Save**. ### Remove an AI integration To disable AI functionality within Sigma, you can remove the AI provider: 1. Go to **Administration** > **AI settings**. 2. In the **AI provider** section, click **Edit**, then click **Remove**. 3. When prompted to confirm, click **Remove**. 4. Click **Save**. After the integration is successfully removed, the **AI provider** section displays a **Save** button, and AI functionality is unavailable for your organization. ## Related resources * [Explain charts with AI](/docs/explain-visualizations-with-ai) * [Use AI with formulas](/docs/use-ai-with-formulas) * [Ask natural language queries with Sigma Assistant](/docs/ask-natural-language-queries-with-assistant) * [Supported AI models](/docs/supported-ai-models)