I built an offline document indexer, and Ollama taught me two things I did not expect

Описание и применение

Инструмент локально сканирует, распознаёт и классифицирует сканированные PDF‑файлы, создавая Markdown‑индексы и отчёты, не загружая данные в облако.

Задача

A folder of hundreds of scanned, image‑only documents was unsearchable and could not be uploaded to any service, making manual opening the only way to find information.

Результат

The tool enables local indexing and classification of scanned documents, producing searchable Markdown outputs and AI‑ready instructions, all without uploading data to external services.

Рабочий процесс

• Point tool at a folder
• Scan files and detect pages without text layer
• OCR those pages
• Slice oversized PDFs
• Read every page
• Classify each page into item with type, date, author, summary
• Write four Markdown files (index, timeline, review report, AI instructions) to an output folder
• Leave originals untouched

Доказательства

You point it at a folder. It scans, runs OCR only on the pages that have no text layer...

Источник

Classification runs on an open model through Ollama, on the machine itself.

Источник