MUMBAI, India, May 29 -- Intellectual Property India has published a patent application (202611050415 A) filed by Ankush Kumar; Anshuman Gautam; Shashi Kant; and Dr. Sushant Jhingran, Greater Noida, Uttar Pradesh, on April 20, for 'adaptive decision support platform using open-source language models, retrieval-augmented knowledge processing, and multi-module autonomous agents.'
Inventor(s) include Ankush Kumar; Anshuman Gautam; Shashi Kant; and Dr. Sushant Jhingran.
The application for the patent was published on May 29, under issue no. 22/2026.
According to the abstract released by the Intellectual Property India: "An intelligent decision system based on computer implementation to produce strategic advice is revealed by utilizing open-source large language models organised in an agentic modular framework. The user objective in natural language input to the platform is automatically broken down into executable subtasks by a planner engine. A retrieval subsystem retrieves external and internal knowledge bases with the help of semantic embeddings, search by similarity in vectors, reranking, and reconstruction of contextual documents. Facts observed by a person are put into working memory through re-organization into interpretable facts. Plan quality, sufficiency of evidence and logical consistency is evaluated by an evaluator subsystem. In case of deficiencies, new tasks are produced, and extra retrieval or reasoning steps are performed. Subsequent actions, such as tool invocation, further analysis, resolution of contradictions or response completion are determined dynamically by a selector subsystem. A responding subsystem will subsequently produce an evidence-based output, specific to the user situation. The architecture minimizes generation of hallucinated responses by basing it on retrieved data and verified states of memory. It enhances explainability as it maintains plans, facts, source, and decision paths. It can be implemented with open systems models on either local, cloud or hybrid deployment thereby enhancing the privacy, cost management and customization. It can be used in technology strategy, enterprise intelligence, energy forecasting, transportation planning, and policy planning, as well as in healthcare operations and financial analysis. Examples of outputs are rankings, scenarios, forecasts, summary, and comparative reports, and also the executive recommendation. The revealed system is not the traditional chatbots and linear retrieval pipeline, but simultaneous autonomous planning, iterative validation, adaptive memory, and dynamic selection of actions organized into one system. Subsequently, organizations can acquire quicker, more dependable, and more adaptable decision intelligence to the intricate, information-voluntary setups of continual adjustment to evolving data streams."
Disclaimer: Curated by HT Syndication.
