The new study AI4Cyber continues the course already started in July, illustrating the evolution of the "Retrieval-Augmented Generation (RAG) for local and secure document retrieval". In continuity with what was described in the first publication, the development of the system had a twofold objective: the improvement of the interaction between the user and various documents and the introduction of new functionalities to extend the application scope of the system.
Among the main new features, the system now includes extended document management for text, visual and audio content, accessible through a completely revamped interactive dashboard. A module for the automatic processing and transcription of voice streams, a semantic description system for images using multimodal templates and the possibility of conditional summaries based on selected content have also been introduced.
One of the strengths of the system is the possibility of a AI assistant operating locally, capable of ensuring operational autonomy and full adherence to the fundamental principles of computer security, Confidentiality, Integrity, Availability (CIA). In an ecosystem where cloud solutions may expose data protection risks, local execution is a strategic choice, particularly suited to business contexts with high confidentiality requirements.
Overall, the system proves to be a reliable, flexible and secure platform, designed to provide intelligent, controlled and contextual access to information, in line with the operational needs of modern organisations.
If you wish to learn more, here is the link to our studio complete.
In addition, you can subscribe to the specific mailing list Cyber Studios by Tinexta Defence, to receive updates on upcoming research:
