MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202531131750 A) filed by Dr. Kshirod Sarmah; Dr. Debashis Dev Misra; Dr. Hitesh Choudhury; Ajit Das; and Dr. Monoj Kumar Muchahari on December 25, 2025, for System And Method For Automated Deep Learning Based Multi-Modal Generative Intelligence With Dynamic Prompt Optimization.
Inventors include Dr. Kshirod Sarmah; Dr. Debashis Dev Misra; Dr. Hitesh Choudhury; Ajit Das; and Dr. Monoj Kumar Muchahari.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: The present invention discloses a hardware-implemented multi-modal generative intelligence system is performs automated dynamic prompt optimization for deep learning generative models. A device housing contains at least one processor, a hardware accelerator, memory, multi-modal input devices (including microphone and camera), output devices (display and audio) and a network interface. Software modules executed on the processor encode user requests from speech, text or images, extract context features and generate an initial base prompt. A dynamic prompt optimization engine constructs and scores candidate prompts using stored interaction history and quality metrics, then selects an optimized prompt. A generative orchestration layer maps the optimized prompt to modality-specific prompts for text, image and audio models executed on the hardware accelerator or optionally on a remote server. Generated outputs are post-processed and rendered to the user. Feedback is logged and used to adapt future prompt optimization. The system reduces manual prompt engineering and delivers resource-aware, high-quality multi-modal generative outputs. (Accompanied Figure No. 1-3)
Disclaimer: Curated by HT Syndication.
