Overview
IDA is a structured finance chatbot that enables users to ask questions across both data and documents from a single interface. The Data side is used for exploring tables and visualizations, while the Document side focuses on searching, comparing, and understanding textual content.
Data Side Tabs
The Data side is designed for numerical analysis and visual exploration. Users can ask questions that return tabular or graphical results and refine them using follow-up queries.
Table – For questions that return structured tabular data
Visualization – For questions that return charts or graphs
Follow-ups – To refine or extend the most recent question
Document Side Tabs
The Document side allows users to search, analyze, and compare document content.
Table Search – View search results in a structured table format
Document RAG – Ask contextual questions about documents
Summaries – Generate a summary of the dashboard or document
Clause Comparison – Compare two clauses side by side
The Summaries tab provides a quick overview of the dashboard content.
Other Tabs
IDA also includes the following supporting tabs:
History – Revisit previously asked questions
Export / Download – Save results for offline use
Demo Questions
Data Module:
What is the top 5 property state-wise distribution and its percentage?
Follow-up: Visualize it as a pie chart
What are the top 10 largest loans?
Document Module:
Present the complete Applicable Margin table, outlining all relevant categories and associated values
What is the Adjusted Pool Balance?
When is the closing date?
What are the conditions precedent to the closing date?
Confidence and Citations
Each response includes a confidence score and page-level citations. Selecting a citation highlights the exact referenced line within the source document.
New Features
IDA includes the following enhanced capabilities:
Clause Comparison – Compare two clauses side by side for quick analysis
Document Intelligence – Displays source context and page numbers along with answers
Feedback
Users can provide feedback on each response to help improve accuracy and future performance.
Getting Started
Process Workflow: