Amatrium adds Word export and redesigns AmatriumGPT interface

Amatrium on July 21, 2026, unveiled a redesigned AmatriumGPT interface for enterprise teams, adding one-click Word document export and new productivity features. The release also highlights citations, model isolation, zero data retention and multi-model support aimed at secure workplace use.

Why it matters: - Amatrium is positioning AmatriumGPT as a faster enterprise AI tool for teams that need reusable output, not just chat responses. - The new Word export feature is aimed at turning AI answers into working documents with less manual copy-and-paste. - The release emphasizes security controls and cited answers for companies that need traceable outputs.

What happened: - Amatrium on July 21, 2026, unveiled a redesigned user interface for AmatriumGPT. - The company added direct export of responses to a Word document. - The update is built to make enterprise AI faster, easier and more actionable.

The details: - AmatriumGPT is described as a knowledge hub for company collaboration. - The platform gives teams access to answers, simplifies content creation and delivers word-for-word translation across 100 or more languages. - AmatriumGPT can read handwriting and scanned documents. - The platform can interpret electrical, hydraulic and PCB drawings to pull information into bid packages and technical support workflows. - The software includes ChatGPT, Gemini and Claude. - The release adds Claude 5. - The interface includes a more user-friendly design. - Admins can track employee usage. - The Documents view now includes filters, usage charts and UI refinements. - Quick questions are included to help users get started faster. - A Resources/Help section includes how-to videos. - The default LLM remembers the last selected model. - Every response includes citations for verification. - The LLM only accesses information relevant to the prompt and does not scan entire documents. - The system uses model isolation, role-based access control and two-factor authentication. - Amatrium enforces zero data retention across supported models via OpenRouter's ZDR routing. - Requests are processed through zero-retention infrastructure including Azure, Google Vertex and Amazon Bedrock. - Customer data is not used for training or public learning. - The AmatriumGPT software uses four specialized agents: Orchestrator, Document Analyzer, Tool Executor and Validation Agent. - The agent workflow is designed to deliver clear, accurate and trustworthy answers. - The system combines source citations, calculations and verification into one organized response across long or multiple documents. - Amatrium directs readers to learn more.

Between the lines: - The redesign suggests Amatrium is competing on workflow efficiency as much as model quality. - The security and privacy language signals a pitch to enterprise buyers that are cautious about public AI tools. - The multi-agent setup is meant to reduce errors and make answers easier to verify.

What's next: - Amatrium is likely to market the new interface and Word export as productivity upgrades for enterprise teams. - The company appears to be leaning on secure deployment and cited output as key differentiators in future sales conversations.