Intellectual Property India Publishes Patent Application for Ai Powered Api Monitoring And Anomaly Detection.

MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202621051806 A) filed by Geeta S. Navale; Dr. Rashmi Kale; Amruta V. Patil; Dr. Smita Bhagwat; Vaibhav Pujari; Sania Khan; Isha Sable; and Rahi Dubewar on April 23, 2026, for Ai Powered Api Monitoring And Anomaly Detection..

Inventors include Geeta S. Navale; Dr. Rashmi Kale; Amruta V. Patil; Dr. Smita Bhagwat; Vaibhav Pujari; Sania Khan; Isha Sable; and Rahi Dubewar.

The application for the patent was published on July 03, 2026, under issue no. 27/2026.

Abstract: Contemporary microservice architectures expose API monitoring systems to rapidly shifting workloads and complex cross-service dependencies that overwhelm conventional static-threshold and manual log-review techniques. This invention presents APIGuard AI, a self-adaptive, AI-driven framework that addresses the complete spectrum of API observability and anomaly resolution. The system deploys a weighted ensemble of three unsupervised machine learning models—Isolation Forest, Autoencoder, and K-Means—each capturing distinct anomaly signatures across request rate, latency, error rate, and throughput telemetry. An embedded Large Language Model root-cause analysis engine translates numerical detection outputs into concise, engineer-ready diagnostic narratives by correlating anomaly scores with structured logs and distributed traces. A continuous adaptive retraining loop governed by the ADWIN drift-detection algorithm preserves model accuracy as API behaviour evolves, while a semantic correlation module identifies coordinated failure propagation across service boundaries. Implemented with FastAPI, MongoDB, LangChain, LangGraph, and Groq- hosted Llama 3 on an OpenTelemetry-instrumented Docker microservice cluster, the framework achieved 94.3% detection accuracy over 2,847 evaluation hours, reduced mean time to detection by 67%, and decreased false-positive daily alert volume by over 78%. The invention provides a scalable, production-ready foundation for intelligent API health management in cloud-native and IoT environments.

Disclaimer: Curated by HT Syndication.