AI Assistant

Can be one of the following types:

The Chat Parameters class defines a set of options for tuning the behavior of LLM completion requests in Grooper. These parameters allow users to control the randomness, diversity, and topicality of generated responses, enabling tailored outputs for a wide range of scenarios.

Typical usage involves adjusting one or more parameters to guide the model’s output toward the desired style or content. For example, increasing randomness for creative tasks, or reducing repetition for summarization.

Configuration Guidance

• Adjust either 'Temperature' or 'Top P' to control randomness and diversity, but not both simultaneously.
• Use 'Presence Penalty' and 'Frequency Penalty' together to reduce repetition and encourage broader topic coverage.

Usage in Grooper

Chat Parameters are used throughout Grooper to influence LLM completions for document processing, data extraction, and other AI-driven tasks.

Impact

Proper configuration of ChatParameters can significantly affect the quality, creativity, and reliability of LLM outputs. Experimentation is recommended to find the best settings for your specific use case.

Use these instructions to define the role, personality, and response style of the assistant. A well-crafted set of instructions could include:

• Role and Purpose: Clearly state the primary function of the AI Assistant (e.g., aiding users in understanding travel expenses, providing quick answers to frequently asked questions, etc.).
• Tone and Style: Specify the desired communication style(e.g., friendly, professional, concise) to ensure consistent interaction.
• Response Approach: Outline the preferred methods for answering queries (e.g., direct answers for straightforward inquiries and elaborative responses for complex questions).
• Limitations: Identify areas where the assistant may lack information or cannot operate(e.g., specific business logic or non-available data queries).

Below are some sample instructions.

Your primary role is to guide users through understanding travel expense reports and related queries.

Communicate in a friendly and professional tone, prioritizing clarity and helpfulness.

Provide direct answers to straightforward questions while offering detailed explanations for complex inquiries, ensuring users receive comprehensive knowledge.

Recognize when information is outside your scope and gracefully guide users to alternative resources or suggest ways to obtain the necessary information.

Configure this property to specify which external knowledge sources the assistant can access when responding to user inquiries.

Resources may include:

• Search Index: For semantic and keyword-based document retrieval.
• Database Table: For direct access to structured data in database tables.
• Other Resource Reference: For custom or specialized data sources.

Each resource should be uniquely named. If multiple resources share the same name, validation will prompt you to assign aliases.

Best Practices:

• Include only those resources relevant to the assistant's intended role.
• Regularly review and update the list to ensure accuracy and security.
• Use descriptive names to make resource selection clear in multi-resource environments.

A combination of the following flags:

Use this property to enable or disable core retrieval features that supplement the assistant's custom resources.

Built-in tools include:

• AskUser: Allows the assistant to prompt users for clarification when a query is ambiguous.
• HelpSearch: Enables searching of Grooper help topics and wiki articles.
• LoadSchema: Grants access to schema information for search indexes, web services, and database tables.
• LoadWebPage: Permits loading and retrieval of web page content.

Multiple options can be combined using bitwise flags. The 'All' value enables all built-in retrieval tools.

Typical Usage:

• Enable only those tools that align with your assistant's purpose and user expectations.
• For assistants focused on internal documentation, consider enabling 'HelpSearch' and 'AskUser'.
• For data-centric assistants, 'LoadSchema' may be essential.

Use this property to provide detailed guidance on how the assistant should select, prioritize, and interact with knowledge resources when formulating responses.

Purpose:

• Directs the assistant's retrieval logic, ensuring that information is gathered in a way that aligns with your business rules, compliance requirements, and user expectations.
• Helps the assistant determine when to search, which resources to consult, how to combine results, and when to ask clarifying questions.

Configuration Guidance:

• Clearly specify preferred search strategies (e.g., "Always check search indexes before querying database tables").
• Define thresholds and limits, such as the maximum number of results to retrieve or the acceptable depth for multi-hop retrieval.
• Indicate when the assistant should prompt the user for more information (e.g., "If the user's question is ambiguous, ask for clarification before searching").
• Provide rules for combining or ranking results from multiple resources.

Best Practices:

• Keep instructions concise and focused on actionable guidance.
• Update instructions as your data sources or business needs evolve.
• Test the assistant's behavior with a variety of queries to ensure the Retrieval Plan aligns with your intent.

Example Scenarios:

• "For policy-related questions, search the 'Policies' search index first. If no relevant results are found, check the 'Procedures' index."
• "Limit database queries to 10 results unless the user specifies a different amount."
• "If a user's question could apply to multiple resources, ask which resource they want to use before proceeding."

This property controls the assistant's ability to perform multi-hop retrieval—a process where multiple sequential retrieval steps are executed, with each step refining the query or gathering additional information based on previous results.

How it works:

• When a user submits a complex question, the assistant may need to combine information from several resources (such as Search Index, Database Table, or other sources).
• Each retrieval "hop" allows the assistant to use the output of one retrieval as the input for the next, enabling it to synthesize answers that require reasoning across multiple data points.

Configuration Guidance:

• Set this value to 1 to disable multi-hop retrieval, restricting the assistant to a single retrieval operation per user inquiry.
• Increase the value (up to 8) to allow deeper, more sophisticated retrieval chains for advanced scenarios.

Best Practices:

• Use higher values only when your use cases require combining information from multiple sources or when users frequently ask complex, multi-part questions.
• Monitor performance and response times, as higher values may increase processing time and resource usage.
• Start with the default value and adjust as needed based on user feedback and observed assistant behavior.

Example Scenarios:

• Simple Q&A: For straightforward lookups, a value of 1 is sufficient.
• Cross-Resource Reasoning: For questions like "List all invoices for customers who purchased product X last quarter," a higher value enables the assistant to retrieve and correlate data from multiple tables or indexes.

If not set, all authenticated users may chat with this assistant..

Can be one of the following types:

Use this property to connect the assistant to external communication channels such as Microsoft Teams, Facebook Messenger, Slack, Email, and others.

When configured, the assistant can participate in conversations outside the Grooper UI, broadening its reach and accessibility.

Configuration Steps:

1. Provide the required bot credentials (Bot ID and Bot Password) for Azure Bot Framework integration.
2. Ensure the appropriate channels are enabled in your Azure Bot registration.

Best Practices:

• Use access control to restrict sensitive assistants when exposing them externally.
• Test connectivity and permissions for each channel after configuration.

For more details, see the documentation for Bot Connector.

When enabled, the assistant sends partial responses to the user interface as they are produced by the underlying language model.

Benefits:

• Improves perceived responsiveness, especially for long or complex answers.
• Allows users to begin reading and acting on information before the full response is complete.

Considerations:

• Streaming may not be supported by all client interfaces or LLM providers.
• Disable streaming if you require atomic, all-at-once responses for compliance or workflow reasons.

Can be one of the following values:

An external chat conversation is one which originated via the Bot Connector or the Grooper API rather than the Grooper UI.