How to OCR and Merge Scanned Tax Slips Locally

Short answer: Put the scanned slips in one folder, run on-device OCR to turn the images into searchable text, let a local AI classify each slip, then review and export one merged PDF. With PDF Insight the OCR and the classification both run on your own machine — no scan is ever uploaded.

Clients hand you scans and phone photos, not tidy digital PDFs. A scanned T4 is just an image until something reads the text off it. Doing that in the cloud means uploading a document full of SINs and income figures. Here's how to OCR and merge scanned slips entirely on your own machine.

Why keep OCR local

• Privacy. Scanned slips contain SINs and full financials. On-device OCR means the image and its text never leave your machine — no third-party processor, no upload, nothing to breach.
• No connection needed. Tax season often runs on locked-down or air-gapped machines. Local OCR keeps working offline.
• Compliance comfort. Not sending client documents to a cloud OCR service sidesteps the PIPEDA concerns that come with third-party processing.

Step by step

1. Collect the scans into one folder Scanned PDFs and photographed slips both work. Flat, well-lit, full-page images OCR best.
2. Run on-device OCR Add the folder to PDF Insight. It runs OCR locally on each scanned page, turning images into searchable text with no internet connection.
3. Let the local AI classify A local LLM reads the OCR text and identifies each slip — T4, RL-1, T5, T4A, and so on — on your own machine.
4. Review and merge Check the labels and order, fix any misread, and export one merged PDF per client. Nothing is uploaded.

OCR accuracy depends on scan quality, and slip box numbers change yearly. Always review the classifications and verify slip details against current CRA and Revenu Québec guidance before filing.

Tips for better OCR results

• Scan at 300 DPI or photograph the full slip flat, in good light.
• Crop out backgrounds; a slip filling the frame reads better than one in the corner.
• Re-scan skewed or shadowed pages rather than fighting a bad image.
• Use the review step — it's faster to confirm a label than to hunt for a misfiled slip later.

Performance

PDF Insight organizes a real 11-document bundle, scans included, in about 100 seconds on a 16GB Mac. For crunch time there's an optional, opt-in paid cloud speed lane — clearly separated from the local-private default, where nothing leaves the machine.