AI Connection
You can use the AI connection objects and step templates in your workflows. Each connection stores the configuration and credentials needed to authenticate with a specific provider, making it easy to switch between providers or models without modifying your workflow.
When a job reaches an AI connection step in a workflow, RICOH ProcessDirector uses the associated AI connection to communicate with the configured AI provider. The step sends a prompt along with an optional input file, such as a document, image, or data extract, to an AI model. The AI model processes the request and returns a response, which the step can store either as a job property or as a file in the job spool directory. This flexibility lets you use AI for a variety of purposes in your workflows. For example, you might use an AI step to:
- Analyze document content
- Extract data from unstructured text
- Generate summaries
- Classify documents
- Validate information against business rules
System objects:
This feature adds these object types:
- AI Connection
- An AI Connection object defines the settings that RICOH ProcessDirector use to communicate with an AI service provider. When you create an AI connection, you specify the credential that authenticates with the AI provider and select which AI models are available. Steps that reference this connection can then send requests to the AI model and receive AI-generated responses.
- Amazon Bedrock credential
- An Amazon Bedrock credential object stores the Amazon Web Services (AWS) authentication information that RICOH ProcessDirector uses to connect to Amazon Bedrock services. The credential defines which AWS account and IAM role ProcessDirector assumes when making requests to Amazon Bedrock AI models.
Step template
The feature adds a step template that sends AI prompts and receive responses to these prompts during job processing.
- ConnectToAI
- The ConnectToAI step template sends data to an AI service and stores the response for use by steps in the workflow. The step saves the response as either a job property or a file and continues processing in the workflow. Use this step to analyze document content, generate text, extract information, classify jobs, or make intelligent routing decisions based on AI-powered processing.