Using Natural Language Queries in SmartResearch

Written by Uday Kumar Updated

Purpose

This section explains how to use natural language queries in SmartResearch to analyze and report on enterprise data. It provides step-by-step instructions, examples, and best practices for business users.

What Are Natural Language Queries?

Natural language queries allow users to ask questions in plain English (or other supported languages) instead of using technical query languages or complex filters. SmartResearch’s AI-powered engine interprets these questions, translates them into structured data queries, and returns actionable business answers.

Example:

  • “Show all invoices pending approval for more than five days.”

  • “What is our current DSO and how has it changed this quarter?”

  • “List vendors with the highest exception rates last month.”

Procedure

  1. Navigate to the SmartResearch home page.
    • For navigation instructions, please see Navigating to SmartResearch.
  2. At the top of the SmartResearch screen, open the SmartFlow Instance dropdown. Select the instance you want to analyze.
  3. In the main chat or query input field, you can just type your business question in plain English.
    • Example queries:
      • “What are our current days payable outstanding?”
      • “Show top vendors by spend for April.”
      • “List overdue AP aging by department.”
  4. Alternatively, click a suggested query button below the input field (for example, Current DPO, Top Vendors by Spend, Overdue AP Aging) to auto-populate the query.
  5. Select the Submit/Enter button (typically a right-arrow or similar icon) to run the query.
  6. Wait for SmartResearch to process and return results.
  7. Review the results displayed above the input field. 

Note

Note 1: The accuracy and completeness of results depend on the data sources connected for the selected instance.

Note 2: Always review and validate SmartResearch responses before making business decisions.

Note 3: If you do not see the expected results, try rephrasing your question or check your instance/data source configuration.


Figure 1. Entering a Natural Language Query in SmartResearch

Best Practices

  • Be Specific:
    The more specific your question, the more accurate the results.
    Example: “Show overdue invoices for vendor ABC in Q2 2025.”

  • Use Business Terms:
    Use terms familiar to your organization (e.g., “DSO,” “exception rate,” “pending approval”).

  • Follow Up:
    You can ask follow-up questions in the same thread to drill down or clarify results.

  • Check Source Citations:
    Review the data sources and citations provided with each answer for transparency.

Supported Query Types

  • Invoice and payment status

  • Vendor performance and exception rates

  • Financial metrics (e.g., DSO, DPO)

  • Approval bottlenecks

  • Cross-system data aggregation