Skip to content
SLT OCR

Documentation

From upload to workbook.

Six steps in the interface, and one HTTP API if you would rather automate it.

Getting started

  1. Create an account

    Register with an e-mail; trial credits land immediately.

  2. Create a project

    A project binds a set of documents to one extraction model.

  3. Upload documents

    PDF, JPEG or PNG. Held in memory only.

  4. Run extraction

    One file at a time, with visible progress and resumable on failure.

  5. Review the fields

    Every value shows its source and confidence before it is accepted.

  6. Export the workbook

    A Keys XLSX with the same columns as the delivery pipeline.

Available models

IdentifierModelCost / pageStatus
slt-nid-v1

Egyptian National ID

Reads name, national number, address and issue data from scanned Egyptian ID pages, including handwritten legacy cards.

5Enabled
slt-page-cls-v1

Contract page classifier

Labels every page of a scanned contract bundle and emits a section map with bookmarks for review.

5Enabled
slt-generic-ocr-v1

Generic Arabic / English OCR

Full-page text recovery for mixed Arabic and English documents. Not yet enabled on this deployment.

10Not enabled

HTTP API

Full interactive reference (OpenAPI)

Authentication is the same cookie session the application uses. No bearer token is sent in a header.

MethodPathPurpose
GET/api/app/modelsList extraction models
GET/api/app/subscriptionCredits, usage and engine status
GET/api/app/projectsThe account's projects
POST/api/app/projectsCreate a project
GET/api/app/files?projectId=Files in a project
POST/api/app/filesUpload files (multipart)
POST/api/app/files/bulk-extractQueue files for extraction
POST/api/app/files/{id}/extractExtract one file
GET/api/app/files/{id}/previewField values and provenance
POST/api/app/files/exportDownload the Keys workbook
GET/api/app/audit-logsActivity log

Document retention

Uploaded files are held in memory for fifteen minutes and then discarded. If the window closes before extraction, the record reads source expired and the document must be re-uploaded. What persists is the extraction result, never the document.

Integration questions? Write to info@smartleadtech.com or see the terms.