Work with the existing desktop application
A Python script running on a Raspberry Pi with Windows used Pywinauto to poll the messenger every 30 seconds and extract new listing messages.
Workflow automation
Turn repeated manual work into a reliable routine.
Tools built with real estate agents to move information from messenger conversations into a shared working record.
Message-to-spreadsheet workflow
01KakaoTalkNew listing messages
02PythonExtract & deduplicate
03Google SheetsA shared record
Start with the work people repeat.
Alongside my degree, I worked part-time at DwiorA Real Estate in Seoul. Working directly with agents showed me the repetitive tasks that were worth automating.
One tool turned listings shared in KakaoTalk group chats into structured rows in Google Sheets. Other work included an Electron and Puppeteer tool for property adverts, and a Chrome extension for structured property notes.
Understand the work before changing it.
Agents shared useful property details in messenger conversations, then manually copied price, deposit, area, and other fields into a tracking sheet. Keeping that record up to date was repetitive and error-prone.
The choices behind the product.
Structure the information before sharing it
Regular expressions extracted the property fields. Duplicate checks and formatting rules helped keep the Google Sheet consistent.
Learning through the work.
The workflow saved agents hours of manual work daily. The structured-notes extension was adopted across the office.
Working beside the users made it possible to notice small, repeated tasks and build tools that fitted naturally into the day. That experience later informed Estait and JipXL.