Intellectual Property India Publishes Patent Application for System And Method For Detecting Organizational Buying Intent From Web Signals Using A Hybrid Multi-Agent Architecture

MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202621052617 A) filed by Anish Sachin Katariya; Vijaya Sardesai; Gayatri Suryakant Ghorpade; Deepali Mahesh Gohil; Shraddha Gendlal Vaidya; Prajakta Pawar; Ninad Kale; Manasi Phand; Amey Kharade; Aditya Phadke; and Kuldeep Vayadande on April 24, 2026, for System And Method For Detecting Organizational Buying Intent From Web Signals Using A Hybrid Multi-Agent Architecture.

Inventors include Anish Sachin Katariya; Vijaya Sardesai; Gayatri Suryakant Ghorpade; Deepali Mahesh Gohil; Shraddha Gendlal Vaidya; Prajakta Pawar; Ninad Kale; Manasi Phand; Amey Kharade; Aditya Phadke; and Kuldeep Vayadande.

The application for the patent was published on June 19, 2026, under issue no. 25/2026.

Abstract: Detection of buying intent is a critical problem in B2B customer relationship management, and the identification of organizations that exhibit buying intent has been shown to improve sales force prioritization. Previous works on the problem have proposed the use of deep learning, knowledge graph reasoning, and reinforcement learning for buying intent prediction. However, the practical usability of the proposed methods has been limited by their difficulty to deploy, their lack of interpretability, and their difficulty in adaptation. In the proposed work, the authors have proposed a proof-of-concept architecture for the integration of web signal discovery, knowledge graph storage, vector similarity search, and large language model classification within a multi-agent framework. The proposed work does not contribute to the development of a novel learning model, and the novel contribution lies in the integration of the proposed components into a novel architecture for the purpose of B2B sales intelligence. The proposed work has been evaluated on a small-scale exploratory study, and the authors have proposed the use of 12 free-text queries and 30 manually labeled web signals for the purpose of classification. The proposed work has been evaluated on the basis of the classification accuracy, the fit score, latency, and the per query cost. The proposed work has demonstrated the practical usability and economic viability of the proposed architecture for the purpose of buying intent discovery.

Disclaimer: Curated by HT Syndication.