1. Manual transcription
Operator retypes handwritten Arabic line by line.
Arabic-first document intelligence
The Arabic archive, machine‑readable.
We turn handwritten Arabic manuscripts, letters and records into text you can search, cite and use. On your servers. Your documents never leave.
Programmes and recognition
Inception Program
Technology Fast 50 ME · Impact winner
For Startups
Capital Q pre-accelerator
Activate
For Startups
In the press
CoreTechX OCR system for handwritten Arabic recognition
CoreTechX leads the GCC's transition from AI ethics to quantifiable ROI through transparent AI
CoreTechX launches first end-to-end on-premise AI system for Arabic handwriting
How CoreTechX's OCR system makes Arabic-first document intelligence practical
Unlocking the Arabic archive: CoreTechX's hybrid transformer architecture and the end of the black box
Beyond OCR: CoreTechX unlocks the GCC's locked data assets with sovereign AI
The founders building AI infrastructure for handwritten Arabic
CoreTechX and Arabic-first document intelligenceFox InterviewerHow CoreTechX is building Arabic-first document intelligenceEye of DubaiHow CoreTechX's OCR system turns Arabic documents into strategic assetsHow it works
From a handwritten page to an answer you can trace.
- 1Upload
Scan the page as it is
Faded ink, margins and stamps. No manual clean-up first.
- 2Read
Handwriting becomes text
Line by line, with a confidence score on every word.
- 3Verify
Humans check the hard parts
Low-confidence words are flagged, so experts spend minutes, not days.
- 4Search
Ask the archive
Search by meaning across thousands of pages, cited to the page.
The difference
From manual chaos to machine speed.
One handwritten page, start to answer. Minutes are illustrative.
2. Proofread twice
Cross-checked by a second reviewer.
3. Re-enter into official form
Copy text into institutional templates.
4. Manual tagging
Type in date, author and classification.
5. File naming & storage
Save to shared drives and update the index.
6. Search when needed
Browse folders and filenames by hand.
7. Read for context
Read whole documents to interpret them.
8. Summaries & reports
Write summaries for managers by hand.
9. Repeat next time
Every new question starts from scratch.
1. Upload the scan
Drag and drop into the CoreTechX portal.
2. AI transcribes
Handwritten Arabic to text, with word confidence.
3. Review flagged words
Only low-confidence words go to an editor.
4. Auto-template export
Structured into your Word or PDF template.
5. Search & summaries
Indexed for meaning; answers cite the page.
What we offer
Built for governments, archives and researchers.
National archives, ministries, courts, universities and AI teams.
Handwritten Arabic OCR
Manuscripts, letters, court and land records, including marginal notes and Hijri dates.
Every word, weighed
One dot can change a word. The model weighs each possible reading and shows its confidence, so reviewers only check the close calls.
Automated templates
Owner, date, location, boundaries, price and witnesses pulled straight from the handwriting into your institution's own form, ready for Word or PDF.
Archive search & translation
Ask in Arabic across every collection at once, get summaries, and translate structured content for research.
On-premise & sovereign
Runs fully on your infrastructure. No data leaves your servers. Designed for PDPL and regulated institutions.
Training data for AI teams
Verified handwritten Arabic lines, each image paired with its checked transcription. Rights-cleared and delivered in the format your training pipeline needs.
Engagement models
Start with your own pages.
For institutionsRecommended
Platform & managed digitisation
Run CoreTechX on your own servers, or let our team deliver the whole archive for you. Many institutions combine both.
Customby volume & complexity
- On-premise deployment, no data leaves your servers
- Handwritten Arabic OCR with word confidence
- Managed delivery with human verification by trained linguists
- Search, translation and metadata add-ons
- PDF, Word, JSON outputs and secure APIs
- Timelines and quality targets agreed upfront
For AI teams
Datasets & licensing
For AI teams and research organisations.
Customby volume & complexity
- Licensed handwritten Arabic data
- Human-verified ground truth
- Page- or line-level alignment
- Custom preparation to your spec
FAQ
Questions institutions ask us.
Something else on your mind? We answer within one working day.
Ask on a callWhat documents do you support?
Handwritten and printed Arabic: manuscripts, letters, court and land records, registers and government files, including marginal notes and stamps.
How accurate is it?
It depends on script, ink and paper. Every word gets a confidence score and uncertain words go to human review. Ask for a demo login and test it on your own pages before any contract.
Can it run on our own servers?
Yes. On-premise is our default for government and regulated clients. Your documents never leave your infrastructure.
Is our data used to train your models?
No, not without a separate written agreement.
Which languages and scripts?
Arabic first, in Naskh and Ruq'ah handwriting as well as print. Other scripts on request.
How long does deployment take?
A pilot on your documents typically starts within weeks. Full rollout depends on volume and integration.
Can it connect to our existing systems?
Yes, through secure APIs and exports to Word, PDF and JSON.
Get in touch
Bring us one page. We'll show you the text.
FastestBook a 30-minute call Pick a time with our team. Bring a sample page.
Emailinfo@coretechx.ai
Office111 Al Diwan Street, Riyadh, Saudi Arabia 12726
Arabic-first document intelligence, built for institutions.
Digitise, structure and search handwritten Arabic archives, fully on-premise.
Book a call