RedactorBuddy removes names, accounts, and personal identifiers from financial documents on your own machine — so you can use external AI on the numbers without exposing the people behind them.
Each layer is more thorough than the last — and all of them run entirely on your hardware.
Pulls text and tabular data out of your document while keeping every number, category, and total in place.
PDF · CSV · XLSX · TXT · JSON · MDThree stacked passes — fast regex, then Microsoft Presidio, then an optional local Ollama model — catch what the layer before missed.
regex → Presidio → OllamaScores residual PII risk and refuses to export if it's too high. You review, then ship a sanitized summary to any external LLM.
fail-closed · audit ledgerPreserves numbers, categories, and financial calculations so downstream analysis stays accurate.
If residual PII risk crosses your threshold, export is blocked. Safety is the default, not an afterthought.
Every redaction is logged on-device for traceability — useful when someone asks what was removed and when.
PDF, CSV, XLSX, TXT, JSON, and Markdown in; a clean, shareable sanitized summary out.
The optional LLM cleanup pass runs through local Ollama — no API keys, no cloud, no data egress.
Drop files into a monitored folder and let them process automatically — ideal for steady daily document volume.
Raw financial documents must never leave the local machine. Everything RedactorBuddy does is built around keeping that true — and being honest about its limits.
The local redaction core is free forever. Paid tiers unlock automation, audit export, and commercial use rights.
Free to download. Runs on Windows. Your documents stay exactly where they are.
// regex + Presidio + Ollama · spaCy en_core_web_lg · fully offline