Building LoanReady: An AI-Powered Bank File Processor
Built a working web application using AI coding tools to automate financial document preparation for CAs.
Kartik Sodhi
Associate - Private Capital at Finex Hong Kong Limited
From their time as
Associate - Private Capital
Finex Hong Kong Limited • 2025
Overview
Kartik built LoanReady independently, outside of his day job, using a combination of Claude Code, Gemini, and OpenAI Codex. The product is a web application designed for chartered accountants: a CA can upload raw bank files for their clients, and the tool automatically organizes and formats the data into a presentation-ready output suitable for submission to banks.
The Story
Kartik built LoanReady independently, outside of his day job, using a combination of Claude Code, Gemini, and OpenAI Codex. The product is a web application designed for chartered accountants: a CA can upload raw bank files for their clients, and the tool automatically organizes and formats the data into a presentation-ready output suitable for submission to banks.
The problem it solves is a real one in the CA workflow: bank files from clients arrive in inconsistent formats, and preparing them for bank submission is a manual, time-consuming process. LoanReady removes that step.
Building it required Kartik to work across the full stack using AI-assisted development, from the file ingestion logic to the output formatting layer. He used the AI coding tools not just for boilerplate but to move through unfamiliar technical territory faster than he could have with conventional learning.
He is also working on a second tool: a private company intelligence tracker that pulls data from GST filings and other public sources to generate timely updates on unlisted companies. That project is still in development.