Turn your data into an analytics and LLM-ready environment.

Hawaii Mortgage Lending Analytics

Explore Home Mortgage Disclosure Act (HMDA) data for Hawaii. This chatbot answers aggregate questions about mortgage applications, approvals, denials, loan purpose, applicant income, and lender activity from the 2022-2024 Modified LAR dataset.

Download the source HMDA data from the FFIEC Data Browser.

MCP-connected HMDA dataset

How The Chatbot Works

User Browser

You ask a question, request a chart, or export an aggregate result from this website.

LLM Reasoning

The LLM interprets the question, creates analysis logic, and keeps responses tied to the data.

MCP Connector

MCP bridges the LLM to controlled tools and the prepared Hawaii HMDA dataset.

Analytics Output

The chatbot returns answers, SQL details, tables, charts, and dashboard-ready summaries.

Example: "What is the approval rate by year?"
Example: "Create a line chart of denial rate by year."
Example: "Which loan purposes have the highest application volume?"

Recommended Questions

Responses are aggregate-only and read from the MCP-connected HMDA dataset.
Executed SQL and calculation details 

        

        

          

            Ask
            
Best for common HMDA metrics. Uses the app's approved query templates first, then returns a grounded aggregate answer, table, and SQL details.

          

          

            Explore With LLM SQL
            
Best for open-ended cuts of the data. The LLM writes a read-only aggregate SQL query, the app validates it, then runs it against the HMDA dataset.

          

        

      


      

      

        
How Customer Usage Improves The Chatbot

        

          

            

              Usage Feedback Loop
              Better questions -> better signals -> better answers
            

          

          

            Customers ask HMDA questions
            
Real customer questions show which lending metrics, geographies, and comparisons matter most.

          

          

            Capture usage signals
            
The chatbot records the question, intent, selected filters, and answer feedback.

          

          

            Find gaps and patterns
            
Repeated questions and negative feedback reveal missing definitions, templates, or UX guidance.

          

          

            Improve chatbot behavior
            
Those insights become better SQL templates, prompts, chart choices, and metric explanations.

          

          

            Ship better answers and UX
            
Customers get clearer responses, safer guardrails, and more useful dashboard-ready outputs.

          

          

            Customers trust it more and ask better questions
            
Better answers create more useful usage data for the next improvement cycle.

          

        

        

          

            Usage signals
            Questions, intent category, filters, feedback
          

          

            Patterns to review
            Unanswered questions, wrong metrics, unclear answers, repeated prompts
          

          

            Behavior improvements
            Better SQL templates, prompt guidance, chart choices, definitions
          

          

            Customer impact
            Clearer answers, safer guardrails, better dashboards