We have identified a growing demand among our clients for free text redaction. The solution must not only deliver precision, but also operate effectively in local languages, accommodate domain-specific terminology, and address organization-specific requirements.

In response we have developed a new tool for accurate free text redaction. The BONSAI text application can be fine-tuned to the achieve a high level of accuracy tailored to your specific vocabulary and language. The initial fine-tuning has been carried out for the Finnish language within the healthcare domain.

Sensitive or confidential free-text documents are becoming a real bottleneck for AI, analytics, and collaboration – especially when redaction must be accurate, fast, and reliable. This challenge is further amplified when working with specialized terminology and non-English languages such as Finnish, where precision is essential.

Organizations want to move faster with documents and AI, but three business-critical challenges keep getting in the way:

• Manual redaction is time consuming
  Teams still spend valuable hours reviewing and masking sensitive information manually, slowing down document sharing, analysis, and internal workflows.
• Free-text redaction in Finnish – and other non-English languages – remains challenging
  Most redaction tools are developed with English-centric language models, which are not well-suited for handling documents in languages such as Finnish that contain nuanced personal, confidential, and context-specific information.
• Security and compliance concerns block progress
  Organizations cannot risk breaching data protection regulations (such as GDPR and the AI Act) or exposing confidential business information by sending documents containing personal or confidential information to external AI services or third-party processors.

For these reasons, we have introduced BONSAI text, a new application designed to enable secure and accurate redaction of sensitive information in free-text documents.

With BONSAI text, organizations can:

• Protect sensitive information before sharing, analyzing, or using documents in AI-assisted workflows
• Redact multiple PDF, TXT, and DOCX files in one task, reducing manual effort
• Choose the redaction style that fits their use case – black bars for strict masking or entity-type labels for downstream workflows
• Improve precision with customer-specific blacklists and whitelists, tailored to internal terminology and policies
• Keep documents fully in their own environment, with no reliance on external AI providers, generative AI, or large language models

The result is faster document handling, lower privacy risk, and more confidence in using sensitive text data safely.

BONSAI text is continuously evolving to meet the specific needs of our customers:

• The initial version has been fine-tuned for Finnish-language use cases. Our next objective is to enable organizations globally to protect sensitive free-text data with the same level of accuracy, control, and security.
• The application is first available as Azure Managed Application, offering easy deployment for organizations operating within the Azure ecosystem. Additional cloud deployment options will follow.

Book a demo to see how BONSAI text can support your document workflows and privacy goals. More information regarding the application is available on our webpage.