Let’s be honest: most companies are sitting on a mountain of scanned documents, image-based PDFs, handwritten forms, and old files that haven’t seen the light of day in years.

Sure, they’re stored somewhere. But usable? Interactive? Actionable? Not even close.

That’s exactly what we’re changing with ARender.

From OCR to ICR + AI: Smarter Understanding of Your Documents

We’re developing a powerful new module based on ICR — Intelligent Character Recognition — enhanced with AI to go far beyond traditional OCR.

The mission is clear: make image-based, unstructured documents selectable and usable across your business workflows.

So what’s different?

• It reads both handwritten and numeric text — even messy notes or stylized form fields
• It handles layout noise (folds, stamps, low contrast) and poor scan quality
• It understands context, not just visual patterns
• It’s pluggable with any AI/ML backend, including on-premise models for full control
• And it integrates seamlessly into ARender, enhancing the viewing experience without changing your workflow
  

Built for Reality, Not Just Demos

We’ve been working closely with one of France’s major insurance providers (we can’t name names — but trust us, you know them).

Together, we:

• Validated multiple AI/ICR approaches on real documents from production
• Built a fully functional proof-of-concept, directly integrated into ARender
• Proved it could handle not only scanned documents, but also mobile phone photos — even skewed, shadowed, or partially folded

Because in the real world, users don’t always have a scanner. Sometimes it’s just a smartphone and a document on a desk.
We built for that reality.

Early prototype in action — a mobile photo of a handwritten form, made selectable with one click

Real-World Input Types We Handle

Our prototype already supports a wide range of real-world document formats, including:

• Scanned PDFs and TIFFs (single- or multi-page)
• Handwritten or typed forms
• Numeric fields in printed or written formats
• Faxes or degraded legacy documents
• Photos taken with mobile devices — imperfect, angled, or shadowed
• Mixed-content files (handwriting + stamps + printed text)

All of these can now be explored and selected as if they were born digital.

Why This Matters

With ICR, previously “dead” zones of a document — like handwritten notes, checkboxes, or photographed pages — become selectable and copyable.

• Users can highlight and extract text directly within ARender
• You can integrate the detection via REST API into your own tools
• No need to switch apps or send documents elsewhere for processing
  

🔍 And while the detected content isn’t searchable yet, that’s a possible next step — once we ensure performance and scalability.

What’s Next: Looking for Real-World Testers

We now have a fully functional proof of concept — running inside ARender, tested on real insurance documents, and already able to make handwritten or numeric content selectable with a single click.

The next step? Put it in the hands of real users.

We’re looking to integrate this early ICR capability with interested customers and prospects — those dealing with scanned forms, photos of paperwork, or legacy image-based content — and who want to help shape how intelligent text detection fits into their workflows.

• You’ll get early access to the ICR module
• You’ll help guide its evolution with real use cases
• And your feedback will directly influence the final product

Want in? Let’s talk. We’re ready when you are.