Enabling the Textual Agent

To enable the Textual Agent on a self-hosted instance of Tonic Textual, you must configure the connection to the AI model that you want to use.

Textual supports:

• Any OpenAI-compatible endpoint

• Gemini

• Amazon Bedrock

Selecting the model provider

To identify the provider, set the environment variable CHAT_LLM_PROVIDER.

The available values are:

• openai - Indicates to use an OpenAI-compatible model.

• gemini - Indicates to use a Gemini model.

• bedrock - Indicates to use a model on Amazon Bedrock.

Connecting to an OpenAI-compatible model

If you set CHAT_LLM_PROVIDER to openai, then to configure the connection to an OpenAI-compatible model, set the following environment variables:

• CHAT_MODEL_ENDPOINT - The endpoint URL for the OpenAI service.

• CHAT_MODEL_NAME - The name of the OpenAI-compatible model to use.

• CHAT_API_KEY - The API key to use for authentication.

Connecting to a Gemini model

If you set CHAT_LLM_PROVIDER to gemini, then to configure the connection to a Gemini model, set the following environment variables:

• CHAT_MODEL_ENDPOINT - The endpoint URL for the Gemini service.

• CHAT_MODEL_NAME - The name of the Gemini model to use.

• CHAT_API_KEY - The Gemini API key to use for authentication.

Connecting to a model on Amazon Bedrock

If you set CHAT_LLM_PROVIDER to bedrock, then to configure the connection to a model on Amazon Bedrock, set the following environment variables:

• CHAT_MODEL_NAME - The name of the model to use.

Optionally, also set:

• CHAT_AMAZON_BEDROCK_REGION - The AWS Region where Amazon Bedrock is located. If you do not provide this, then Textual uses the AWS Region that is set as the value of AWS_DEFAULT_REGION.

• CHAT_AMAZON_BEDROCK_MAX_TOKENS - The maximum number of output tokens for Amazon Bedrock responses. The default value is 64000.