Local-first Excel AI rewriter
Let privacy-conscious Excel users edit formulas and data with plain English
- Problem
- Many people can't recall Excel formulas or write VBA, but uploading entire spreadsheets to cloud AI feels risky. Most AI spreadsheet tools require sending the original file, so compliance-sensitive users avoid them.
- Solution
- A desktop app or browser extension that parses Excel locally and sends only column names, data types, and a sampled summary to the LLM. It returns formulas or transformation steps, applies them locally, shows a diff for every change, and only saves after confirmation. Files never leave the machine.
- Users
- Finance, ops, and HR professionals who live in Excel but avoid complex formulas, especially those in companies with data-compliance requirements.
- MVP
- Import xlsx/csv and parse locally; generate formulas or transformation steps from plain-language requests; show before/after diffs for confirmation; write changes back locally; export an operation log.
- Revenue
- Free tier plus one-time purchase or annual subscription (~$39/year); on-premise team licenses per seat.
- Why now
- Cheap models like DeepSeek make summarize-then-generate cheap, while corporate fear of data egress is peaking. An indie developer can ship an MVP with existing spreadsheet parsing libraries and no model training.