AI Agent Chat Window

Overview

The Shibumi AI Agent intelligently reviews your program to help you progress initiatives or opportunities, get ahead of risks, and make prioritization decisions. The Agent responds to natural language prompts by assessing your relevant program data.

The AI Agent evaluates information configured within the solution to generate contextual insights. In addition to analyzing attribute data, the Agent understands the broader structure of your program, including how work items are organized, connected, and managed. By combining program data with structural context and solution logic, the Agent can provide comprehensive responses about both program performance and the methodology used to manage it.

When the AI Agent identifies recommended changes based on a user prompt, users can review, modify, and accept attribute value updates directly within the chat window without ever leaving the conversation. Governance and data integrity remain fully intact, with the Agent only surfacing attributes it has been granted visibility to and all existing editability rules enforced, ensuring users can only update values they have the appropriate rights to change.

Enable AI Chat for Templates

In addition to the default Program Structure and Logic, the AI Agent can be configured to review specific attribute values from select templates. Once configured, the Agent will respond to questions with the specified attribute values returned.

1. As an App Admin, select ‘Apps’ from the Sidebar.
2. Choose an App name to be brought to the App Detail page.
3. Select the ‘AI’ Tab.
4. To define a new template for the AI Agent, select the ‘+ Configure Type’ button.
5. From the AI Agent Enablement dialog, choose a template type. The Name attribute will be included by default.
6. Use the checkbox to the left of the attribute name to enable an attribute (association, checkbox, date, metric, number, pick list, text, or role). The chosen attribute’s definition will be automatically included. Optionally, use the selector on the right to filter available attributes by their type. Follow the same process to add more attributes if desired.
7. Toggle on the ‘Enable for AI Create’ to allow the Agent to create instances of the template type directly from the Chat window. The Name attribute will automatically be ‘Required for Create’. Optionally define additional attributes that are Required for Create. Note: Attributes must be ‘Enabled’ for the AI Agent before they can be ‘Required to Create’.
8. Save and Publish.

To edit a previously configured template for use by the AI Agent, click the edit icon (pencil) for an existing template to add or remove attributes.

Note: Admin Only attributes are not available for AI Enablement.

AI Agent Chat Window Key Features

• Easy Access & Continuous Chatting. The Agent can be accessed from any page within the solution and maintains a history of previous chat logs for easy reference.
• Recommendation and Updating. The Agent can provide recommendations to attribute values it has access to based on a user prompt. These recommendation can then be applied directly within the chat window.
• Security. Shibumi recognizes that security is critical for clients. The Agent follows the Shibumi security model and will never respond with information that the user does not already have permission to access.
• Structural Intelligence.  Understands the structure of the program, including template relationships, business rules, and calculations. This allows the Agent to interpret how work items are connected, who is responsible for them, and how they influence one another across the solution.
• Reasoning. The Agent explains, in plain language, the steps taken to arrive at its conclusions and includes supporting data/visualizations when available.