Nuclearn Book a demo
Platform Documents and OCR

Your documents, handwriting included,
searchable and citable.

Documents come into the Nuclearn platform once: uploaded, or synced by Nuclearn Connect from the document systems your plant already uses. Our OCR reads scans and handwriting. Each document keeps its revision history, with the current one marked, and the products search and cite the same library.

Handwriting

It reads the handwriting, too.

Decades of work orders, procedure drafts, log sheets and field notes were filled in by hand. Our OCR reads handwritten print and cursive along with the printed form around it, so a 1977 work request turns up in a search beside this year's, and an answer can cite it.

Top half of a 1977 Three Mile Island work request approval form, work request 21000. Handwritten in the form: unit 1, priority 11a, utility, system air handling, component AH-F4B, and the request: "B" control building fan filter unit, delta P 2.4, filters need changing. Across the top, in another hand: see C0689, cancel.
Text read by OCR
  1. See C0689
  2. Cancel
  3. WORK REQUEST APPROVAL TMI Nuclear Station
  4. Unit No. 1
  5. Work Request No. 21000
  6. Work Request Procedure (A.P. 1016 Sect. 6.0) T231 should be used as a guide in filling out this form.
  7. Priority 11a
  8. W.O./Account No. T803/530-5 NPRD Form Req'd. No
  9. Utility
  10. Items 1 through 5 completed by originator
  11. 1. System: Air Handling
  12. 2. Component (name & number) AH-F4B
  13. 3. Describe malfunction or modification and recommended corrective action.
  14. "B" control Bldg Fan Filter Unit ΔP 2.4 Filters need changing
What our OCR read from two public records in NRC ADAMS: a 1977 Three Mile Island work request (ML20064K185, page 43) and a page of a 1978 maintenance procedure, written by hand, from a Unit 2 work package (ML20136C699, page 7). The text is shown as the OCR returned it, unedited, and each box is where on the page it read that passage.
Scans

Then it's searchable and citable.

Each PDF is checked when it arrives. Pages without a text layer are read with OCR, and so are PDFs whose text layer is there but unreadable. What OCR reads is indexed with everything else, so search finds it and answers cite it.

  • Catches scanned pages inside an otherwise digital PDF.
  • Citations and highlights land on the scanned page itself.
  • Each document shows its stage while it's read, run through OCR and indexed.
A library search for monthly operating report contents finds a 1974 Atomic Energy Commission regulatory guide that exists only as a scan. It opens at page 3, with the text read from the scan listed beside the page image.
Screenshots show public NRC documents in a demo project.
Coming next · Metadata

Metadata worked out for you.

In development now, and not yet in a release. The platform will read each document for what a document controller records, and check each value against the document before it uses it, so nobody has to key in an index.

Number, title, revision, date

The document number, title, revision and effective date, read off the document itself, from its cover block, masthead or header, scans included.

Checked against the page

Each value comes with the exact words it was read from. If those words aren't in the document, the value isn't used, so a file name or a guess can't become a document number.

Your systems' record first

When a document arrives from your document system with its number and revision, those values come first. A second copy of the same file counts as a copy, not a new revision.

Revisions

Know which revision you're reading.

Each document keeps its revision history, with the current revision marked. Earlier revisions stay in the record, marked superseded, and stay out of search results unless a project pins one.

  • Opening a superseded revision says which one is current.
  • Citations name the page and the revision.

Coming next: put a collection under document control and each new revision is found and linked to the document it revises, put in order by revision and effective date, with drafts and withdrawn revisions kept but never made current, and matches it isn't sure of queued for a person to confirm.

Regulatory Guide 1.192 with its Revision History open: Rev 5 marked current, Rev 4 and an earlier revision marked superseded, each with created and superseded dates and buttons to view or download it.
Screenshots show public NRC documents in a demo project.
What comes in

The files plants already have.

Office files, PDFs and scans

PDF, Word, Excel, PowerPoint, CSV, text, HTML and Markdown, with scanned and handwritten pages read by OCR.

Forms, read into fields

Document Intelligence matches a form to its template and extracts its fields, such as text, dates, numbers, checkboxes and signatures, for you to check.

From your systems, kept private

Nuclearn Connect syncs from your document systems, file shares, Maximo and Aras Innovator. A synced collection starts private, and search only returns what each person may read. Integrations →

Request a meeting

Bring the documents your team searches most.

We'll show you a library of public NRC documents, scans and handwriting included, and how your own would come in.

Request a meeting