AI Agents Could Replace Spreadsheets
Spreadsheets made lightweight programming accessible to almost anyone, but agents could shift some of that work into custom software with logic that's easier to inspect.
Developed from a conversation between Pete Winn and Andy David
Excel became one of the world’s most widely used programming tools without most people treating it as a programming language. Its grid is a remarkably useful primitive. People can add values across rows and columns, track lists, build financial models, analyse data and create charts without first learning conventional software development.
That flexibility becomes a liability when a workbook grows or several people work on it. Version control is difficult, and a formula error can sit invisibly inside a cell while producing results that look plausible. Inserting rows or copying a formula into the wrong range can leave an entire model subtly wrong, forcing reviewers to trace calculations cell by cell to find where the mistake entered.
AI agents make a different approach practical for some of this work. Instead of constructing a general-purpose workbook, someone could ask an agent to build a small program tailored to the task and share it with the team. The program’s code could be reviewed directly, checked by a second agent and equipped with tests that expose failures. A Jupyter notebook could still show the formulas, working and charts, while making the underlying logic more explicit than calculations hidden across spreadsheet cells.