This one is personal. RightClickOCR exists because opposing counsel produced thousands of pages of scanned documents, ⌘F found nothing in any of them, and every cloud OCR service wanted me to upload privileged and confidential client material to servers I knew nothing about. What follows is the guide I wish had existed then — written by a practicing attorney, for the people who actually handle productions.
Why productions arrive unsearchable
Even in 2026, an enormous amount of discovery is produced as image-only PDFs: paper files run through a scanner, native documents “printed” to image on purpose, fax-era records, exhibit binders scanned at the courthouse. Sometimes it’s workflow laziness; sometimes the producing party simply has no obligation to hand you text. Either way, what lands in your matter folder is photographs of pages — no text inside the files, which is why ⌘F comes up empty.
The cost is concrete. Every keyword search becomes a page-turn. Finding each mention of a name across a production means reading the production. Deposition prep balloons. And things get missed — not because anyone was careless, but because human eyes at page 3,100 are not a search engine.
The cloud OCR problem is an ethics problem
The obvious fix — upload the PDFs to a free OCR website — is where lawyers need to stop and think. Client confidences are involved. Depending on the matter, so are protective orders with explicit handling restrictions, privileged materials inadvertently produced, and documents designated confidential or attorneys’-eyes-only.
Uploading those files hands copies to a third-party service under terms you likely haven’t reviewed, with retention you can’t verify, on infrastructure in jurisdictions you can’t name. Lawyers have a duty of confidentiality and a duty of technological competence in most U.S. jurisdictions — and “the website said it was secure” is not the analysis either duty contemplates. Attorney-client privilege doesn’t have a “we use industry-standard encryption” exception.
That doesn’t mean cloud tools are never permissible — firms make considered vendor decisions all the time, with contracts and diligence. It means a free web upload form isn’t a considered vendor decision. The cleanest answer is to keep the documents on hardware you control. (The general version of this argument: OCR without the cloud.)
What the alternatives look like
- Full e-discovery platforms (hosted review databases) OCR everything as part of processing. For document-heavy matters with review teams, that’s the right tool — and a per-gigabyte, per-month cost structure that makes no sense for the small and mid-size matters most lawyers actually have.
- Adobe Acrobat Pro can OCR and batch. It works; it’s also a subscription suite, and the batch workflow takes setting up. If the firm already licenses it, it’s a fair choice — see our roundup.
- Preview’s Live Text lets you copy text off a page on screen but writes nothing into the file — the production stays unsearchable, and it can’t touch a folder of documents.
A clean on-device workflow for a scanned production
Here’s the workflow RightClickOCR was built around. Everything happens on your Mac; nothing crosses a network.
- Preserve the production as received. Keep the original production folder intact — filenames, structure, everything as produced. You may need to reference it exactly as it arrived.
- Right-click the production folder in Finder and choose Make Searchable (OCR). Every PDF inside is processed in one batch.
- Each document gets a sibling copy — “Name (Searchable).pdf” — with an invisible text layer under the page images. Critically, the originals are never modified: the files as produced remain byte-for-byte untouched, which is exactly what you want for anything that might become an exhibit or the subject of a dispute about what was produced.
- Work from the searchable copies. ⌘F inside any document; Spotlight across the whole matter folder for a name, a date, a phrase (how Spotlight indexing works). Crooked scans are deskewed automatically, and handwritten notes — the margin scribbles that turn out to matter — are recognized too; accuracy varies with legibility, so verify key passages against the page image.
The OCR runs on Apple’s Vision framework, entirely on-device. The app is sandboxed: macOS itself restricts it to folders you approve, and it collects nothing — no telemetry, no analytics, no account. For documents under a protective order, the operative fact is simple: they never leave the machine.
Practical notes from actual use
- Batch first, then read. OCR the entire production the day it arrives. Searchability changes how you read — you triage by keyword before you read linearly.
- Remember OCR is an aid, not evidence. Quote from the page image, and treat search as a way to find the page, not as proof a word does or doesn’t appear — recognition on a bad scan is imperfect.
- Paralegals need this more than you do. The people asked “can you find the page where…” ten times a day benefit most. The workflow is a right-click in the Finder they already use; there’s nothing to train.
- One-time setup: macOS requires flipping the extension toggle in System Settings once before the menu item appears — the 60-second walkthrough.
Unsearchable discovery is a tax on small firms — the big shops pay platforms to make it go away, and everyone else pays in evenings. It doesn’t have to be that way for the price of a court filing fee.