The latest articles on DEV Community by 이상윤 (@baccaso). https://dev.to/baccaso https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4037674%2F1bcb3587-5f37-4991-9733-78d074d46283.jpeg https://dev.to/baccaso en 이상윤 Mon, 20 Jul 2026 08:17:30 +0000 https://dev.to/baccaso/building-a-desktop-app-for-repetitive-excel-and-csv-work-4i26 https://dev.to/baccaso/building-a-desktop-app-for-repetitive-excel-and-csv-work-4i26 <p>I’ve been building Fuga, a Windows desktop app for repetitive Excel and CSV work.</p> <p>The basic idea is simple: open a spreadsheet, type what you want done, and the app creates the needed tables, charts, or summaries.</p> <p>I started working on it because a lot of spreadsheet work is not especially hard, but it is slow to repeat.</p> <p>Things like:</p> <ul> <li>grouping rows</li> <li>cleaning imported CSV or Excel data</li> <li>making summary tables</li> <li>creating charts</li> <li>checking which rows match a condition</li> <li>rebuilding the same analysis across similar files</li> </ul> <p>These are not always “programming problems”, but they often feel like work that should be easier to delegate.</p> <h2> Why a desktop app? </h2> <p>A lot of spreadsheet tools are web-first, but I wanted Fuga to feel closer to working with local files.</p> <p>The app imports Excel and CSV files, keeps generated results on a canvas, and shows intermediate tables and charts instead of hiding everything inside a chat response.</p> <p>That part matters to me because AI-generated work can be hard to trust if the result is just a final answer. For spreadsheet work, I want to see what changed.</p> <h2> What the workflow looks like </h2> <p>A typical workflow is:</p> <ol> <li>Import an Excel or CSV file.</li> <li>Ask for a task in plain English.</li> <li>Let the app generate tables, charts, or summaries.</li> <li>Inspect the generated results.</li> <li>Roll back the AI change if needed.</li> </ol> <p>For example, instead of manually building a summary table and chart, you can ask the app to analyze the imported data and create the result directly on the canvas.</p> <h2> Keeping intermediate results visible </h2> <p>One of the design decisions I keep coming back to is visibility.</p> <p>If the app creates a derived table, that table stays visible. If it creates a chart, the chart stays visible. If it needs intermediate data to answer a question, that can stay visible too.</p> <p>This makes the workflow feel less like “trust this black box answer” and more like a sequence of spreadsheet operations that can be inspected.</p> <h2> Current state </h2> <p>Fuga currently supports:</p> <ul> <li>Excel and CSV import</li> <li>plain-English spreadsheet tasks</li> <li>generated tables and charts</li> <li>visible intermediate results</li> <li>AI change inspection</li> <li>rollback</li> <li>a Windows desktop installer</li> </ul> <p>It is still early, and I’m mainly trying to understand which spreadsheet tasks people would actually want to automate.</p> <h2> What I’m looking for </h2> <p>I’d love feedback from people who spend a lot of time working with spreadsheets.</p> <p>In particular:</p> <ul> <li>What repetitive spreadsheet tasks do you run into often?</li> <li>Would you trust an AI tool more if it showed intermediate tables and charts?</li> <li>What would make this kind of workflow useful in a real work setting?</li> </ul> <p>Project link:</p> <p><a href="https://fuga-site.pages.dev/?lang=en" rel="noopener noreferrer">https://fuga-site.pages.dev/?lang=en</a></p> ai scala database programming