Embedded Document Agent Targets Regulated Workflows
A new agentic editing layer in Nutrient's document SDK lets enterprise applications automate multi-step document workflows - including data extraction, form filling, annotation, and redaction - under configurable approval policies, the company announced.
The updated AI Assistant is embedded in Nutrient's Web, iOS, Android, and React Native viewer SDKs. It uses a three-tier approval model: actions can run fully autonomously, require user confirmation, or be prohibited outright. Edits are staged for review before being permanently applied, a design the company says keeps humans in control of final output.
The announcement is aimed at document-intensive sectors including government, legal, financial services, and healthcare, where redaction accuracy and audit trails carry compliance obligations. Use cases cited by Nutrient include staging redactions of client names for legal review, removing personal health information before sharing with research teams, and identifying personal identifiers across public records requests prior to applying permanent removal.
Developers configure what Nutrient calls "skills" - reusable task definitions that encode domain-specific business logic - alongside policies governing which actions the agent can take without human intervention.
Organisations can connect the AI Assistant to the LLM provider of their choice, including OpenAI, Azure OpenAI, AWS Bedrock, Anthropic Claude, or self-hosted models. Nutrient says this arrangement keeps full control over where data is processed - a material consideration for government agencies and organisations subject to Australian and New Zealand data sovereignty requirements.
" Every enterprise we talk to is trying to solve the same problem: They need AI that works with their documents, not just reads them. But they also need it to run inside their own application, under their own rules,” said Jonathan Rhyne, co-founder and CEO of Nutrient.
"We’re giving developers and the enterprises they serve a document agent they can embed, configure, and trust - one that operates inside their applications with the security and control their organizations expect."
The editing agent is currently available in the Web Viewer SDK only. Mobile and hybrid platform support is listed as coming "later in 2026", with no specific date given.
https://www.nutrient.io/sdk/ai-assistant/
