Ask AI configuration

🚧Documentation Under Construction

We are actively working to improve this documentation. The content you see here may be incomplete, subject to change, or may not fully reflect the current state of the feature. We appreciate your understanding as we continue to enhance our docs.

Ask AI queries your CMS content to summarize and list the existing content you already have on a topic.

Models supported

Brightspot supports OpenAI, Claude, Gemini, and other models. You must obtain credentials for each service from their sites in order to enable AI integrations in Brightspot.

Note

You are not limited to the providers described in these topics. To integrate with a different AI service provider, see Creating custom AI providers using AIChatProvider.

Model	AI Platform	Provider
GPT-4o, GPT-4o mini	OpenAI	OpenAI
Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Haiku, Claude 3 Sonnet	Amazon Bedrock	Anthropic
Gemini 1.5 Pro, Gemini 1.5 Flash	Google Vertex AI	Google
Titan Text Premier, Titan Text Express, Titan Text Lite	Amazon Bedrock	Amazon
Llama 3, Llama 2	Amazon Bedrock	Meta
Cohere Command R, Cohere Command R+	Amazon Bedrock	Cohere
Amazon Nova Micro, Amazon Nova Lite, Amazon Nova Pro	Amazon Bedrock	Amazon

Developer prerequisites

Including Ask AI in a build

The following table lists the dependencies to include in your build configuration.

Artifact	Description
com.psddev:openai	Exposes the OpenAI chat provider and embedding provider.
com.psddev:aws-bedrock	Exposes the Bedrock Anthropic Claude chat provider.
com.psddev:google-vertex-ai	Exposes the Google Vertex AI chat provider.
com.psddev:pinecone	Exposes the Pinecone vector database integration.
com.psddev:ai-chat	Provides the core AI chat framework.
com.psddev:solr-ai	Exposes the Solr vector database integration.
com.psddev:opensearch	Exposes the OpenSearch vector database integration.

Runtime prerequisites

• Developer configuration—None required.

• Ops configuration—

  • If you want to use OpenAI as the embedding provider for Ask AI, Ask AI also needs to be configured with a vector database to store your records. For details, see your provider's documentation, such as Accessing Bedrock for Amazon Bedrock.

• CMS configuration—Configure the site interfacing with the Ask AI provider. For details, see Creating a new Ask AI client for more information.

When you are ready to configure this integration, proceed to Creating a new Ask AI client.

Creating a new Ask AI client

To enable Ask AI on your site, you must first create an Ask AI client.

Note

Before beginning configuration, please open a support ticket to enable Ask AI with Solr as the vector database for your project.

Note

Prior to beginning this process, you must go to the AI provider you are choosing to use and follow their steps to get an API key. You will need this key to complete this configuration.

To configure an Ask AI client:

1. Click > Admin > Sites & Settings.

2. Select the site for which you wish to configure the Ask AI client.

3. Under Integrations, expand the AI cluster.

4. Toggle on Ask AI Enabled.

5. Expand Ask AI Client, and click Create New.

6. Enter a Name for the Ask AI client you are creating.

7. Select a Vector Embedding provider from the list of available providers. This service converts data (such as queries) into high-dimensional vectors. In the module below, click the name of the Vector Embedding provider you selected in step 6, and use the table to complete the fields as needed.

   tip

   Brightspot recommends starting with a smaller provider to see if it works for you before moving onto the larger providers.

   Bedrock Titan Embedding API

   Field	Description
   Credentials	Expand Credentials and select from: - Default—Uses the default credentials that have been set up for your site to access Amazon Bedrock: Titan. - AssumeRole—This option is used to assume a role that has been created for Amazon Bedrock: Titan that may have different access than the default role. You must also provide the ARN (Amazon Resource Name) for this role. - Static—Functions as a separate user with its own set of credentials.
   Region	Enter the AWS region for this Ask AI client.
   Model ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Titan documentation.

   Google Vertex Embedding API

   Field	Description
   Credentials	Select JSON Credentials if it is not already selected.
   JSON Credentials	Enter your JSON credentials to log into Google Vertex AI.
   Scopes	Enter a scope value to log into Google Vertex AI by typing in the proper information and then clicking the add content icon. Repeat this procedure for each scope needed. An authorization scope is an OAuth 2.0 URI string that contains the Google Workspace app name, what kind of data it accesses, and the level of access. Scopes are your app's requests to work with Google Workspace data, including users' Google Account data.
   Project ID	Enter the name of the project you created in Google Cloud.
   Location	Enter the location for the Google Vertex AI project as it was entered in Google Cloud.
   Model	Select the Google Vertex AI model you are using for Ask AI.
   Dimension	Enter a dimension for the vector embedding provider.

   Open AI

   Field	Description
   API Key	Enter the Open AI API key. You must get this key from your account on the Open AI Console. See the Open AI documentation for information about API keys.
   Embedding Model	Enter the name of the embedding model to be used with this configuration. See OpenAI documentation for available models.
   Max Tokens	Enter the maximum number of tokens Open AI should sample. Tokens can be thought of as pieces of words. Before the API processes the request, the input is broken down into tokens. These tokens are not cut up exactly where the words start or end; instead, tokens can include trailing spaces and even sub-words.
   User	Enter a unique identifier that represents your organization, which will help OpenAI to monitor and detect abuse.

8. Select a Search Provider. The selected provider converts your content into vectors for use with Ask AI.

   Note

   To vectorize all of your existing content, please reach out to your Brightspot account representative or project team.

   Amazon Bedrock: Claude

   Field	Description
   Credentials	Expand Credentials and select one of the following options: - Default—Uses the default credentials that have been set up for your site to access Amazon Bedrock: Claude. - AssumeRole—This option is used to assume a role that has been created for Amazon Bedrock: Claude that may have different access than the default role. You must also provide the ARN (Amazon Resource Name) for this role. - Static—Functions as a separate user with its own set of credentials.
   Region	Enter the AWS region for this Ask AI client.
   Model ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Claude.
   Model Version ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Claude documentation.
   Max Tokens To Sample	Enter the maximum number of tokens the model should sample per prompt. A token equates to the basic unit of text that a model learns to understand the user input and prompt.
   Temperature	Enter a value between 0.0 and 1.0. The temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Top K	Enter the Top K value for the model. The model considers a certain number of most-likely candidates for the next token. A lower value narrows the pool and limits the options to more likely outputs, while a higher value expands the pool, giving the model more freedom to consider less likely outputs.
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It is a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.

   Amazon Bedrock: Cohere

   Field	Description
   Credentials	Expand Credentials and select from: - Default—Uses the default credentials that have been set up for your site to access the Ask AI provider. - AssumeRole—Select this option to assume a role that has been created for Amazon Bedrock: Cohere that may have different access than the default role. You must also provide the ARN (Amazon Resource Name) for this role. - Static—Functions as a separate user with its own set of credentials.
   Region	Enter the AWS region for this Ask AI client.
   Model ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Cohere documentation.
   Temperature	Enter a value between 0.0 and 1.0. The temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It is a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.
   Top K	Enter the Top K value for the model. The model considers a certain number of most-likely candidates for the next token. A lower value narrows the pool and limits the options to more likely outputs, while a higher value expands the pool, giving the model more freedom to consider less likely outputs.
   Max Tokens	Enter the maximum number of tokens the model should sample per prompt. A token equates to the basic unit of text that a model learns to understand the user input and prompt.

   Amazon Bedrock: Llama

   Field	Description
   Credentials	Expand Credentials and select from: - Default—Uses the default credentials that have been set up for your site to access Amazon Bedrock: Llama. - AssumeRole—This option is used to assume a role that has been created for Amazon Bedrock: Llama that may have different access than the default role. You must also provide the ARN (Amazon Resource Name) for this role. - Static—Functions as a separate user with its own set of credentials.
   Region	Enter the AWS region for this Ask AI client.
   Model ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Llama documentation.
   Temperature	Enter a value between 0.0 and 1.0. The temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It's a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.
   Max Generation Length	Enter the maximum number of tokens the model should sample per prompt. A token equates to the basic unit of text that a model learns to understand the user input and prompt.

   Amazon Bedrock: Nova

   Field	Description
   Credentials	Expand Credentials and select one of the following options: - Default—Uses the default credentials that have been set up for your site to access Amazon Bedrock: Nova. - AssumeRole—This option is used to assume a role that has been created for Amazon Bedrock: Nova that may have different access than the default role. You must also provide the ARN (Amazon Resource Name) for this role. - Static—Functions as a separate user with its own set of credentials.
   Region	Enter the AWS region for this Ask AI client.
   Model ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Nova.
   Max Tokens To Sample	Enter the maximum number of tokens the model should sample per prompt. A token equates to the basic unit of text that a model learns to understand the user input and prompt.
   Temperature	Enter a value between 0.0 and 1.0. The temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Top K	Enter the Top K value for the model. The model considers a certain number of most-likely candidates for the next token. A lower value narrows the pool and limits the options to more likely outputs, while a higher value expands the pool, giving the model more freedom to consider less likely outputs.
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It is a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.

   Amazon Bedrock: Titan

   Field	Description
   Credentials	Expand Credentials and select from: Default—Uses the default credentials that have been set up for your site to access Amazon Bedrock: Titan. AssumeRole—This option is used to assume a role that has been created for Amazon Bedrock: Titan that may have different access than the default role. You must also provide the ARN (Amazon Resource Name) for this role. Static—Functions as a separate user with its own set of credentials.
   Region	Enter the AWS region for this Ask AI client.
   Model ID	Enter the version ID of the foundation model you are using for this Ask AI client. Learn more about the models available to you by visiting Amazon Bedrock: Titan documentation.
   Max Generation Length	Enter the maximum number of tokens the model should sample per prompt. A token equates to the basic unit of text that a model learns to understand the user input and prompt.
   Temperature	Enter a value between 0.0 and 1.0. The temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It's a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.

   Google Vertex AI

   Note

   Your Google Vertex AI credentials are available on your Google Cloud account. Your credentials consist of your JSON Credentials, Scope, Project ID, Location and Model. These values are entered below.

   Field	Description
   Credentials	Select JSON Credentials if it is not already selected.
   JSON Credentials	Enter your JSON credentials to log into Google Vertex AI.
   Scopes	Enter a scope value to log into Google Vertex AI by typing in the proper information and then clicking . Repeat this procedure for each scope needed. An authorization scope is an OAuth 2.0 URI string that contains the Google Workspace app name, what kind of data it accesses, and the level of access. Scopes are your app's requests to work with Google Workspace data, including users' Google Account data.
   Project ID	Enter the name of the project you created in Google Cloud.
   Location	Enter the location for the Google Vertex AI project as it was entered in Google Cloud.
   Model	Select the Google Vertex AI model you are using for Ask AI.
   Max Tokens	Enter the maximum number of tokens the model should sample per prompt. A token equates to the basic unit of text that a model learns to understand the user input and prompt.
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It is a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.
   Top K	Enter the Top K value for the model. The model considers a certain number of most-likely candidates for the next token. A lower value narrows the pool and limits the options to more likely outputs, while a higher value expands the pool, giving the model more freedom to consider less likely outputs.
   Temperature	Enter a value between 0.0 and 1.0. The temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Presence Penalty	Enter a value between -2.0 - 2.0. This parameter makes the model less likely to repeat words that have already been used in the generated text.
   Frequency Penalty	Enter a value between -2.0 - 2.0. This value discourages the model from repeating the same words multiple times, promoting a richer vocabulary.

   Open AI

   Field	Description
   API Key	Enter the Open AI API key. You must get this key from your account on the Open AI Console. See the Open AI documentation for information about API keys.
   Engine ID	Select the specific model used to power the AI experience.
   Max Tokens	Enter the maximum number of tokens Open AI should sample. Tokens can be thought of as pieces of words. Before the API processes the request, the input is broken down into tokens. These tokens are not cut up exactly where the words start or end; instead, tokens can include trailing spaces and even sub-words.
   Top P	Enter a value between 0-1. Top-p is an inference parameter that controls the model's token choices based on the probability of potential options. It is a float with a minimum and maximum of 0 and 1, respectively, and a default of 1. Choose a lower value to decrease the size of the pool and limit the options to more likely outputs. Choose a higher value to increase the size of the pool and allow the model to consider less likely outputs.
   Temperature	Enter a value between 0.0 and 1.0. Temperature is a parameter that controls the randomness of predictions in a model's text generation. A lower setting makes the model more conservative and deterministic, while a higher setting increases randomness and more creative outputs. Brightspot recommends a temperature of 0.4 (for Ask AI) or 0.6 (for Create with AI).
   Presence Penalty	Enter a value between -2.0 - 2.0. This parameter makes the model less likely to repeat words that have already been used in the generated text.
   Frequency Penalty	Enter a value between -2.0 - 2.0. This value discourages the model from repeating the same words multiple times, promoting a richer vocabulary.

9. Enter the Max Records For Search. This is the maximum number of records the Ask AI client considers when performing a search. Brightspot recommends 3.

10. Enter the Min Score For Search Records. This is a value between 0.0 and 1.0 that signifies the minimum score rating the Ask AI client uses when considering records returned in a search. Brightspot recommends 0.5.

    tip

    For faster response from the model, limit the Max Records For Search to five or under, and the Min Score For Search Records to 0.8.

11. Enter the Max Tokens For Search. The upper limit is dependent on the provider selected. Tokens can be thought of as pieces of words. Before the API processes the request, the input is broken down into tokens. These tokens are not cut up exactly where the words start or end—tokens can include trailing spaces and even sub-words.

12. Under Content Types, click and select the content types you want to expose to the Ask AI client. You may make multiple selections.

13. Under Response Text Formatter, retain None or select Default Formatting. This field is used to format the response from the AI provider, and the default response formatter is a simple text formatter.

14. In System Prompt, enter prompts (if desired) to guide the LLM (large language model) and set the general direction of the conversations.

    

15. Under Max Recent Conversation History for Context, enter the limit of message history to include. Each item in the conversation represents a question from the user and an answer from the model. The default setting is all.

16. Click Save.

Enabling Ask AI

Ask AI gives you the ability to ask questions about your content and data in the CMS and get a response from a generative AI model.

To enable Ask AI:

1. Click > Admin > Sites & Settings.
2. Click Integrations, and expand the AI cluster.
3. Ensure that Ask AI Enabled is toggled on.
4. Select an Ask AI Client. See Creating a new Ask AI client for information on creating an Ask AI client.
5. Click Save.

Note

Brightspot also offers Create with AI, which allows you to use AI to generate content for your site. For information on setting up Create with AI, see Create with AI user guide.

Using Ask AI

Ask AI is an assistant that can index your knowledge bank of content and provide answers to your questions based on that content. It allows editors to use natural language to query Brightspot to get a summary of existing content and a listing of assets used to develop that summary.

Note

In order to use the Ask AI feature, it must be enabled on your site and your user role must have the correct permissions assigned. See the topics below for more information:

• Enabling Ask AI
• Configuring Ask AI permissions for a role

To use Ask AI:

1. In the left navigation, click .

   

2. Enter your question in the Ask a question field.

3. Click Submit.

   Your answer is returned to you, along with a list of all of the assets AI used to compile your answer. You can ask follow up questions as needed.

4. Take other actions as necessary:

   1. Click Restart Chat to clear the widget and start again.

   2. Click > Recent to view the five most recent conversations you have had with Ask AI.

   3. Click > Saved to view any conversations that you have saved.

   4. Click > Save Conversation to save the current conversation with Ask AI.

   5. Click to close Ask AI.

      

tip

Clicking after the AI answer copies the entire answer so you can paste it where needed.

Configuring Ask AI permissions for a role

The ability to use the AI functionality within Brightspot is based on the permissions assigned to an editor's role. Without the appropriate permissions assigned to the role, or the role assigned to the user, the AI icons are not visible.

To configure Ask AI permissions for a role:

1. Click > Admin > Users & Roles.
2. In the Roles widget, locate the role for which you want to enable Ask AI functionality.
3. Click under Additional Permissions and select Ask AI Permission. This enables the Ask AI functionality, which allows users to enter a prompt and get a summary of existing related content.
4. Toggle on Use Ask AI.
5. Click Save.