MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202641073568 A) filed by Keshav Memorial Institute Of Technology on June 13, 2026, for Sashakt-The Child Empowerment App.
Inventors include Ms. Nidhi Srivastav; Dr. G Narender; Mr. M. Lalith Srinandan; Mr. A. Sudheep; Mr. M. Sree Charan Reddy; and Mr. Vardaan Bhatia.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: The Sashakt platform employs a multi-agent system for automated data analysis that transforms user-provided natural language queries into verified and executable analytical workflows. The system is designed to promote structured, transparent, and reproducible analysis while preventing unsafe or speculative execution. The architecture consists of three specialized agents with clearly separated responsibilities. The Planner Agent interprets user queries alongside dataset characteristics and decomposes them into structured analytical plans using predefined logic and context-aware reasoning. The Programmer Agent converts these validated plans into complete Python code, incorporating standardized preprocessing steps, defensive programming practices, explicit error handling, and consistent output formatting. Before execution, the Inspector Agent performs multi-stage validation, including syntactic correctness, semantic alignment with the analytical plan, safety and resource- usage checks, and verification of methodological soundness. Agent coordination is achieved through a structured message-passing protocol that preserves conversational context and supports iterative feedback and refinement cycles. This interaction model enforces a strict separation between planning, implementation, and validation stages, ensuring controlled analytical development. The system is supported by four core infrastructure modules. A Dataset Manager handles dataset ingestion, validation, integrity checks, and metadata extraction. An Execution Environment Manager provides secure and isolated execution using managed Kaggle kernel infrastructure with enforced computational and security constraints. A Session Manager maintains conversational state and execution history across concurrent sessions. An Inter-Agent Communication Protocol ensures reliable message delivery and constraint enforcement. Together, these components establish a controlled and verifiable analytical workflow that enhances reliability, reproducibility, and user trust.
Disclaimer: Curated by HT Syndication.
