Chat Summary Widget

Overview

The Chat Summary Widget enables users to review, edit, and track structured conversations captured during interactions with an AI agent. It primarily consists of question-and-answer pairs from assessments, customer interactions, or guided workflows.

Key Features:

✅ Real-Time Progress Tracking – Displays the percentage of assessment completion. ✅ Editable Responses – Users can modify captured answers if the edit configuration is enabled. ✅ Historical Review – Enables users to revisit previous conversations and assessments. ✅ Customizable Workflow Integration – Supports data retrieval and submission workflows for structured chat summaries. ✅ Multi-Format Question Support – Handles diverse question types including text, numbers, dates, addresses, currency, and selections.

🔹 Example Use-Case: A loan assessment AI uses the widget to capture borrower responses, allowing loan officers to review and update the applicant’s financial details before processing.

Configurations

Widget Setup

The Chat Summary Widget must be configured at the agent level within the Widgets Tab of the Agent Configuration.

Workflow Integration

The widget requires two key workflows: 1️⃣ Data Workflow – Fetches the chat summary and displays it in the widget. 2️⃣ Submit Workflow – Saves and updates the chat summary when users make edits.

---

📌 Data Workflow

The Data Workflow retrieves and processes the chat summary when an event triggers a refresh. It receives input data via the Emit Event Node and must return the output in a structured format.

How Input is Passed to the Data Workflow

• When a workflow triggers the "Refresh Conversation Summary" event, the Emit Event Node passes the data field as input.

• This input can contain assessment questions, responses, and metadata.

• The only required field is "disableEdit" (controls whether responses are editable).

Example Input Format

{
   "conversationId": "9bb7d598-316d-476f-95ea-56ccf80e2591",
   "disableEdit": false,
   "questions": [
      {
        "questionId": "needed_loan_amount",
        "title": "Loan Amount Needed",
        "text": "What is the needed loan amount?",
        "options": [],
        "type": "number"
      }
   ],
   "values": {
       "needed_loan_amount": 1000000
   },
   "progressPercentage": 40
}

---

Expected Output Format from Data Workflow

The workflow processes the input and returns a formatted summary in JSON format.

{
 "widgetTitle": "Chat Summary",
 "progress": 40,
 "sections": [
   {
     "header": "Assessment Details",
     "questions": [
       {
         "_id": "needed_loan_amount",
         "type": "number",
         "title": "Loan Amount Needed",
         "questionId": "needed_loan_amount",
         "text": "What is the needed loan amount?",
         "options": []
       }
    ],
     "values": {
        "needed_loan_amount": 1000000
     }
   }
 ]
}

---

📌 Submit Workflow

• If editing is enabled, a Submit button appears at the bottom of the summary.

• When a user modifies responses and clicks submit, the Submit Workflow is triggered.

• The workflow processes the updated data and stores it for future retrieval.

• If required, the workflow can store the updated responses in a database or table.

---

How It Works:

📌 Step 1: A user interacts with the agent, and their responses are stored in the chat summary. 📌 Step 2: The Emit Event Node triggers the Refresh Conversation Summary event. 📌 Step 3: The Data Workflow retrieves the chat summary and displays it in the widget. 📌 Step 4: If editing is enabled, the user updates responses and submits changes. 📌 Step 5: The Submit Workflow processes and saves the modified data.

🔹 Example Scenario: A customer support chatbot logs an issue, allowing agents to review, update, and track case resolution details using this widget.

Applications of the Widget

The Chat Summary Widget is designed for scenarios where structured question-and-answer interactions need to be captured, reviewed, and modified. Below are some key applications of this widget:

---

1️⃣ Loan or Financial Assessments

✔ Helps track and manage loan application details collected through AI-driven conversations. ✔ Allows borrowers to update financial information if needed before final submission. ✔ Provides a real-time assessment progress indicator for easy tracking.

🔹 Example: A loan underwriting system uses the widget to store financial details, such as borrower income, liabilities, and requested loan amount. Loan officers can review and edit the information before approval.

---

2️⃣ Customer Support Summaries

✔ Captures customer interactions, making it easier to track issue resolution. ✔ Allows agents to update ticket details based on the latest customer responses. ✔ Ensures that conversation history is structured and easily accessible.

🔹 Example: A customer service chatbot summarizes previous inquiries, troubleshooting steps, and resolutions, allowing agents to review, edit, and finalize case updates.

---

3️⃣ Project or Task Reviews

✔ Helps teams track ongoing tasks, ensuring all necessary details are captured. ✔ Allows for real-time progress tracking, ensuring smooth project execution. ✔ Supports editable responses, so teams can update task statuses or add missing information.

🔹 Example: A project management AI assistant records team progress updates, enabling managers to review and adjust project timelines directly within the widget.

---

4️⃣ Survey Responses & Audits

✔ Stores survey answers or audit responses for easy reference. ✔ Allows users to modify responses, ensuring accurate data collection. ✔ Provides a structured view of collected answers, making data analysis easier.

🔹 Example: A company compliance bot captures audit responses, allowing auditors to review and correct information before submission.

---

5️⃣ Medical or Health Assessments

✔ Maintains health-related Q&A interactions, helping doctors track patient responses. ✔ Allows patients to review and update symptoms if new information arises. ✔ Supports various medical data formats, including dates, numerical values, and free-text responses.

🔹 Example: A telehealth chatbot captures patient symptoms and medical history, allowing doctors to review, edit, and refine diagnosis details.

Limitations of the Widget

While the Chat Summary Widget enhances structured data collection and review, it has certain limitations that developers should be aware of when implementing it in workflows.

1️⃣ Always Displays Data in Form View

✔ The widget currently only supports form-based summaries. ✔ It does not support a bullet-point summary format for chat responses.

🔹 Developer Note: If a more compact summary format is required, consider designing a custom workflow that converts structured data into bullet-point insights before displaying it.

---

2️⃣ Data Workflow Output Must Follow a Strict Format

✔ The Data Workflow output must match the expected JSON structure. ✔ Any deviation in the format may prevent the widget from displaying the summary correctly.

🔹 Developer Note: Always validate workflow outputs to ensure they conform to the required input structure before integrating with the widget.

---

3️⃣ Requires an Event Trigger for Updates

✔ The widget does not automatically refresh when new responses are captured. ✔ The "Refresh Conversation Summary" event must be explicitly emitted using the Emit Event Node to update the displayed summary.

🔹 Developer Note: Ensure the Emit Event Node is correctly configured in the workflow to trigger real-time updates when new data is available.