AutomationEdge AI Studio

AI Studio provides conversational AI capabilities for AutomationEdge, allowing the creation of text and speech-enabled chatbots. It supports three primary project types:

1. NLU (Natural Language Understanding): Used to build machine learning models from sample utterances to identify user intent and extract entities (e.g., city, color, time). In the preview release, this utilizes Microsoft Azure's Conversational Language Understanding.
2. KM (Knowledge Management): Enables searching within unstructured documents (Excel, PDF, Word). The engine returns the specific relevant section of a document. In the preview release, this utilizes Microsoft Azure's Custom Question Answering.
3. Cognibot: A no-code chatbot builder with a dialog designer. Cognibots integrate NLU and KM projects for natural language processing and connect to AutomationEdge for request fulfillment.

After configuring LUIS settings, verify the connection using these two methods:

1. AI Overview Panel: Ensure the panel reflects the correct numbers from your LUIS Conversational app (fresh projects will show 0).
2. Prediction Testing: Navigate to the Prediction sub-menu and test utterances. If the JSON response on the right side of the screen contains the expected Intent and Entity values, the configuration is successful.

In the Cognibot Overview subtab, you can manage basic project metadata including Project Name, Version, Description, Creation date, and Project Type. For Cloud instances, the system displays either a Teams URL (if 'Create Azure Bot' was enabled) or a Messaging Endpoint (if 'Create Azure Bot' was disabled).

Available project operations:

• Edit Project: Modify the project name and description.
• Delete Project: Permanently deletes the project, its dialogs, and extensions. For Cloud instances, this also deletes the deployment.

The NLU project Overview menu provides high-level project details including Project Name, Version, Description, Creation Date/Time, Engine Type, NLU Engine, and Confidence Threshold.

Confidence Threshold: This value determines the minimum confidence level required for a predicted intent to be included in the final output. For example, if set to 80, only results with >80% confidence are displayed; otherwise, the output is empty.

Key Actions:

• LUIS Settings: Interface to add, update, or delete connections to the LUIS Conversation Application.
• Refresh Project: Syncs the project to get the latest details from LUIS.
• Export Project: Exports the project for use in other AI Studio instances.

The KM project Overview sub-menu provides high-level project management and integration settings. It consists of three primary areas:

1. Project Overview: Displays metadata including Project Name, Version, Description, Creation Date/Time, Engine Type, and KM Engine.
2. QnA Maker Settings: Interface to manage (add, update, or delete) the connection settings between AIStudio and a user's QnA Maker Knowledge base.
3. Export Project: Allows users to export project configurations for import into other AIStudio instances.

The Dialog Designer is the primary environment for building chatbots. It consists of several key functional areas:

• Canvas: The visual area for arranging and maintaining the chatbot design flow. Right-clicking the canvas allows users to create new dialogs, create dialogs from workflows, import zip files, and paste copied dialogs.
• Dialogs Panel: Lists all created dialogs. It provides tools to select, rearrange (via the Rearrange Toggle), and perform bulk operations like Export to zip, Copy, Clone, and Delete.
• Dialog Editor: A configuration panel that opens when a dialog on the canvas is clicked. It is used to edit triggers and manage Dialog Elements.
• Skill Selector: A drop-down menu used to switch between different skills when multiple skills exist.

A typical voice flow managed by the Chatbot Web Service (CBWS) follows these steps:

1. Initiation: CBWS initiates an automated call to the end user.
2. Delivery: CBWS delivers a predefined message, which is converted to speech via text-to-speech (TTS) at the user's end.
3. Input Collection: CBWS waits for user interaction via either DTMF (key presses) or speech input.
4. Termination: Once the flow logic is complete, CBWS sends a final message and a command to end the call.

The Group dialog element in Cognibot allows you to cluster multiple dialog elements together to execute them as a collective set based on a specific condition. This is useful for scenarios where different user personas (e.g., Sales vs. Finance) require different input sets or trigger different workflows based on the same condition.

Key constraints:

• You can place one or more dialog elements inside a Group.
• Groups cannot be nested (a Group element cannot be placed inside another Group).