Your documents deserve more than bolt-on AI

Our AI Assistant is a powerful tool integrated into our document SDKs (Web and Mobile) and low-code and workflow products. It’s designed to revolutionize how you work with documents. Its core capabilities include:

• Intelligent Q&A — Search your document content by asking natural language questions and get precise answers with references to the source.
• Summarization — Quickly generate concise summaries of entire documents or specific selected text passages.
• Translation — Enhance your users’ experiences across multiple languages.
• Contextual actions — Select any text within a document to instantly perform AI-driven actions like “summarize this" or “translate this," or ask specific questions about the selection.‍
• AI-powered redaction — Automate the identification and staging of sensitive information (e.g. PII, specific keywords) for redaction, with a human-in-the-loop review process before finalization.

• Multi-document analysis (roadmap) — While our current focus is deep interaction with single documents, robust multi-document support is a key priority. This will include features like cross-document search, Q&A across entire document sets (knowledge bases/case files), and eventually, deriving complex insights from large repositories.‍
• AI-enhanced document comparison — We launched an AI-powered text comparison feature. This goes beyond simple “diff” views, allowing you to ask the AI to summarize key changes between versions based on specific criteria (e.g. “What changed in the payment terms?”).

We employ several strategies:

• Source referencing — AI-generated answers and summaries are linked back to the specific text in the document, allowing users to easily verify the information.
• Human-in-the-loop — For critical processes like redaction, the AI suggests changes, which are then reviewed and approved by a human user before being applied.
• Prompt engineering — We provide guidance and build our systems with effective prompt engineering to improve the consistency and relevance of AI responses.
• ‍User feedback — We actively gather user feedback to continuously refine the AI’s performance and the usability of its features.

Nutrient offers flexible integration and deployment:

• SDKs — AI Assistant is available through our Web SDK and Mobile SDKs (for both iOS and Android), enabling seamless embedding into your applications.
• Simplified deployment — We’ve streamlined the process; for instance, direct connection to AI services is possible for mobile and web without needing a separate Document Engine container. 
• ‍API and headless mode (roadmap) — We plan to offer API access and a headless mode for programmatic use of our AI capabilities without our standard UI.

Yes, flexibility is key. Our AI Assistant supports:

• Major cloud LLMs — Integration with leading models from providers like OpenAI (via its API or Azure OpenAI Service) and AWS Bedrock.‍
• Bring your own LLM — You can integrate your own self-hosted LLM, as long as it’s compatible with the OpenAI API specification. This gives you maximum control over your AI environment. We’ve seen good results with open source models like Llama.

Data privacy is a top priority.

• Control over data — Using a self-hosted LLM keeps your documents and data entirely within your infrastructure, which is crucial for strict compliance regimes like GDPR, especially concerning data residency for documents and their embeddings (AI-generated text representations).
• Cloud LLM policies — When using cloud LLMs via our integration, we use configurations designed to prevent your data from being used for training public models by those providers.
• No Nutrient model training on your data — We don’t train our core AI models on your specific customer documents. 
• ‍Secure processing — We adhere to robust security practices for data handling, aligning with industry standards.

Yes. For scanned documents, our system can perform optical character recognition (OCR) to extract text. This extracted text can then be processed by AI Assistant for Q&A and summarization, or to suggest redactions. While directly redacting parts of an image typically uses area redaction tools, identifying text within images via OCR is the first step. We’re also researching more advanced AI models for better layout and image content understanding in the future.

We’re continuously innovating. Key roadmap items include:

• Enhanced multi-document capabilities — As detailed earlier, enabling sophisticated search, Q&A, and insights across document collections.
• AI-driven document commands — Expanding AI control beyond redaction to a wider range of document editing, formatting, and annotation tasks via natural language (e.g. “highlight all action items,” “rotate page 3”).
• Deeper contextual awareness — Enabling the AI to leverage broader application-specific context for more personalized and relevant assistance. ‍
• Headless AI and API access — For greater developer flexibility.

Our AI Assistant is valuable across many sectors where intelligent document processing can drive efficiency and insight:

• Legal tech — Assisting with eDiscovery, contract review, due diligence, evidence analysis, and PII redaction for SARs.
• Corporate governance — Streamlining the review of board books, compliance documents, and meeting minutes.
• Financial services — Analyzing reports, policies, and customer communications.‍
• Any industry with document-intensive workflows — Helping users quickly find information, understand complex documents, and prepare them for specific needs, ultimately saving time and reducing manual effort.

Pricing for AI Assistant is typically based on factors such as the chosen deployment model (cloud vs. self-hosted LLM), the scale of usage (e.g. number of users, document volume), and whether it’s an add-on to an existing Nutrient SDK license. We work with you to find a model that suits your needs. For a personalized quote and to discuss your specific requirements, contact our Sales team.