Customer Journey Monitoring With AI Thresholds for API Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing monitoring and response systems in digital platforms struggle to provide real-time insights and predictive analytics, especially in microservices architectures, leading to challenges in detecting and mitigating critical issues, which result in high volumes of priority severity tickets, extended downtime, and negative impacts on customer satisfaction and business operations.
Innovation Solution
A system that utilizes an Isolation Forest algorithm and large language models (LLM) to determine threshold metric values, identify issues, and generate alerts, summaries, and proactive measures by analyzing historical metadata and real-time interactions, integrating ambient AI agents for unified monitoring across APIs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional monitoring systems are used in microservices architectures, then system complexity is reduced, but detection precision and response time deteriorate
Solution Approach 1:
The patent segments the monitoring system into specialized components: Isolation Forest algorithm for anomaly detection, LLM for natural language processing of alerts, and modular pipeline architecture. Each component handles specific tasks independently, improving detection precision without requiring complete system redesign.
Solution Approach 2:
The patent introduces an intermediary processing layer that sits between data collection and alert generation. This layer uses Isolation Forest for anomaly scoring and LLM for alert enrichment, acting as a mediator that enhances detection capabilities while managing complexity through standardized interfaces.
2Reliability
If real-time monitoring of all customer journeys is implemented, then reliability detection improves, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary actions by pre-training the Isolation Forest model on historical data and pre-configuring the LLM with domain knowledge before actual monitoring begins. This preparation enables faster real-time processing without compromising reliability detection.
Solution Approach 2:
The patent applies partial monitoring by focusing on critical customer journeys and key performance indicators rather than attempting to monitor every system parameter. The Isolation Forest algorithm processes only the most relevant features, reducing processing time while maintaining reliable anomaly detection.
3Reliability
If multiple monitoring tools are used across different teams, then monitoring coverage increases, but system integration complexity and data consistency worsen
Solution Approach 1:
The patent creates a universal monitoring framework that can ingest data from multiple sources and tools used by different teams. The system uses standardized data formats and the LLM's ability to process various input types, enabling multi-functional monitoring without requiring separate systems for each tool.
Solution Approach 2:
The patent merges data from multiple monitoring tools into a unified analysis pipeline. The Isolation Forest model processes features from different sources together, and the LLM consolidates alerts into coherent narratives, combining previously fragmented monitoring efforts into a single integrated system.
4Measurement precision
If manual threshold configuration is used for monitoring metrics, then system simplicity is maintained, but detection accuracy and adaptability deteriorate
Solution Approach 1:
The patent implements self-service by enabling the Isolation Forest algorithm to automatically learn optimal thresholds from historical data without manual configuration. The system adapts to changing patterns autonomously, improving detection accuracy while the automated nature keeps operational complexity manageable.
Solution Approach 2:
The patent dynamically changes monitoring parameters based on learned patterns. The Isolation Forest model adjusts threshold values and the LLM modifies alerting behavior based on historical performance data, allowing the system to adapt parameters automatically rather than relying on static manual configuration.
Data Source
AI summary
A method and a system for ensuring reliability of a customer journey with a platform are provided. The method includes: receiving first data that relates to an interaction of a customer with the platform; collecting historical metadata that relates to a plurality of reliability-based metrics for each application programming interface (API) that relates to the platform; determining, based on the collected metadata, a corresponding threshold metric value for each respective reliability-based metric from among the plurality of reliability-based metrics; comparing the first data to each corresponding threshold metric value; when the first data exceeds at least one corresponding threshold metric value: identifying, based on a result of the comparing, a corresponding issue; and generating, based on the identifying, an alert.