On This Page

French AI company specializing in handwriting recognition technology, one of many intelligent document processing vendors offering the iink SDK for developer and OEM integration alongside consumer apps for note-taking and mathematical computation.

Overview

MyScript develops digital ink and handwriting recognition technology for two distinct audiences: developers and OEMs who embed recognition capabilities via the iink SDK, and end users who access the technology through consumer apps. The company supports 66 languages across its consumer products and has expanded language coverage with each SDK release.

The product line divides into two layers. The SDK layer provides the recognition engine as an embeddable component, OS-agnostic and designed for integration into hardware, applications, and cloud services. The consumer layer delivers that engine through MyScript Notes (formerly Nebo) and MyScript Math, both available on iOS, iPadOS, macOS, and Android.

The January 2026 release of iink SDK 4.3 marks the most significant architectural change in the product's recent history. MyScript replaced its per-language model architecture with a unified encoder-decoder attention model that handles 7 Latin languages simultaneously without requiring explicit language selection. The practical results, measured against MyScript's internal benchmarks on large note-taking datasets under real-world conditions, are concrete: 20% fewer English recognition errors, an 83% reduction in resource file size (from over 100 MB to a single 18 MB file), and 30% faster batch processing for raw content recognizer use cases. For OEM partners deploying on constrained hardware, the resource reduction alone changes the economics of embedding the engine.

SDK 4.2 added Chinese handwriting generation using structural modeling, extending the generation capability beyond recognition. SDK 3.1 introduced the RawContent Recognizer and content classification APIs, giving developers finer control over how ink input is categorized before conversion. Qt's official 2026 documentation confirms MyScript SDK as one of two commercial handwriting recognition engines supported by Qt Virtual Keyboard, alongside Cerence Handwriting, covering 72 languages in the standard integration. A dedicated partnership contact (qt@myscript.com) indicates a formal arrangement rather than a community integration.

MyScript's disclosed roadmap includes extending the unified multilingual engine to additional Latin languages beyond the current 7, and developing a dedicated resource for CJK (Chinese, Japanese, Korean) scripts. A CJK resource would extend the addressable market into Asian enterprise document workflows, where handwritten form processing remains common.

How MyScript Handwriting Recognition SDK and Digital Ink Processes Documents

MyScript's processing pipeline centers on digital ink rather than scanned document OCR. The iink engine captures stylus or pen input as a sequence of strokes, applies real-time recognition to convert those strokes into digital content, and outputs structured text, math notation, or diagram representations. This distinguishes it from OCR-based platforms like ABBYY that process static images of existing documents: MyScript operates on live ink at the point of creation.

The SDK 4.3 unified engine handles passive language detection across 7 Latin languages in a single model pass. Previously, developers needed to manage language selection logic and carry separate resource files per language. The new architecture removes both requirements. The 30% batch processing improvement applies specifically to raw content recognizer use cases, which cover document migration, indexing, and large-scale import workflows. These are standard intelligent document processing (IDP) scenarios, and a 30% throughput gain directly affects processing cost and latency at scale.

For mathematical content, the engine parses handwritten equations including variables, fractions, inverse trigonometry, and matrices, then solves them and renders step-by-step solutions. Diagrams drawn by hand convert into precise geometric shapes that remain editable when exported to applications such as PowerPoint. The scratch-to-erase gesture replaces the need to switch between writing and correction tools, keeping the interaction model close to physical writing. Pen pressure now adjusts eraser size dynamically, a minor addition in SDK 4.3 that improves the physical writing analogy.

The SDK exposes this pipeline through APIs documented at developer.myscript.com, covering recognition, ink rendering, and content classification. The RawContent Recognizer introduced in SDK 3.1 allows developers to access ink data before full recognition is applied, enabling custom classification logic on top of the base engine. For document classification use cases where handwritten input needs routing before conversion, such as those handled by platforms like 4semantics, this API layer provides the necessary hook. The legacy per-language engine remains fully supported for integrations using subset knowledge resources, custom lexicons, or text recognition candidates.

The platform does not publish independent benchmark accuracy figures. Recognition quality is attested through user reviews and the company's own internal benchmarks.

Use Cases

Education and academic work

Students deploy MyScript Notes for lecture notes, mathematical problem sets, and study organization. The app handles handwritten math equations and exports them as LaTeX, addressing a specific friction point in academic workflows: moving between handwritten working and typeset submissions. The quiz generation feature converts handwritten notes into study questions without manual reformatting. MyScript Math's step-by-step solver supports learning by exposing the reasoning behind each solution rather than returning only a final answer.

Professional note-taking and meeting documentation

Knowledge workers use MyScript Notes for meeting notes, planning documents, and annotated PDFs. The ability to share annotated content with teams and export to standard formats reduces the gap between handwritten capture and collaborative workflows. Users in sales and project management roles cite the organizational structure and search capability across handwritten content as the primary productivity benefit. The infinite canvas board mode supports brainstorming and mind mapping sessions where fixed-page layouts create friction.

Developer and OEM integration

Hardware manufacturers and application developers embed the iink SDK to add handwriting recognition to stylus-enabled devices and productivity software. The OS-agnostic architecture and the unified multilingual engine in SDK 4.3 reduce integration complexity for products targeting international markets: one 18 MB file replaces what previously required over 100 MB in per-language resources for equivalent Latin script coverage. The Chinese handwriting generation capability in SDK 4.2 extends the addressable OEM market into East Asian device categories. The Qt Virtual Keyboard integration places MyScript in competition with Cerence Handwriting for embedded device deployments, a market distinct from enterprise IDP but relevant to vendors building handwriting capture into hardware products. No specific OEM customer names are disclosed in available sources.

Technical Specifications

Feature Specification
Document types Handwritten notes, mathematical equations, diagrams, annotated PDFs
Input formats Digital ink (stylus/pen input), PDF (annotation), images
Output formats Typed text, LaTeX (math), editable shapes (diagrams), PDF, standard document formats
Processing pipeline Real-time stroke recognition, passive language detection, content classification, structured output conversion
Recognition engine Unified encoder-decoder attention model (SDK 4.3); legacy per-language engine retained for custom lexicon integrations
API/Integration iink SDK (OS-agnostic); REST APIs at developer.myscript.com; Qt Virtual Keyboard commercial integration (72 languages)
Deployment options Mobile (iOS 17.0+, Android), macOS 14.0+ with Apple M1, SDK for embedded/cloud
Languages supported 66 languages in consumer apps; 7 Latin languages simultaneously in iink SDK 4.3 unified engine; 72 languages in Qt integration; Chinese generation in SDK 4.2
Resource footprint 18 MB single multilingual file (SDK 4.3); down from 100+ MB in per-language architecture
Batch processing 30% faster for raw content recognizer use cases vs prior architecture (MyScript internal benchmarks)
English accuracy 20% fewer recognition errors vs prior architecture (MyScript internal benchmarks, large note-taking datasets)
Certifications Not disclosed
Pricing MyScript Notes free with in-app purchases ($1.99–$24.99 iOS); SDK pricing not disclosed

Resources

Company Information

Massy, France. Founded 2001. Independent company. No disclosed funding rounds or parent company identified in available sources. Products distributed globally through consumer app stores and direct SDK licensing to developers and OEMs.