TingIS: Real-time Risk Event Discovery from Noisy Customer Incidents at Enterprise Scale

2026-04-23 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Cloud Computing & IT Infrastructure · Depth: Expert, quick

Summary

TingIS is an end-to-end system for real-time risk event discovery from noisy customer incidents in large-scale cloud-native services. It addresses challenges like extreme noise, high throughput, and semantic complexity by employing a multi-stage event linking engine. This engine combines efficient indexing with Large Language Models (LLMs) to merge events and extract actionable incidents from diverse user descriptions. TingIS also features a cascaded routing mechanism for precise business attribution and a multi-dimensional noise reduction pipeline that integrates domain knowledge, statistical patterns, and behavioral filtering. Deployed in production, TingIS handles over 2,000 messages per minute and 300,000 messages daily, achieving a P90 alert latency of 3.5 minutes and a 95% discovery rate for high-priority incidents, outperforming baseline methods in routing accuracy, clustering quality, and Signal-to-Noise Ratio.

Key takeaway

For AI Architects or CTOs managing large-scale cloud services, TingIS demonstrates a robust approach to real-time incident discovery. Its integration of LLMs with efficient indexing and multi-dimensional noise reduction offers a blueprint for improving alert latency and discovery rates. Consider adopting similar multi-stage linking and cascaded routing mechanisms to enhance your incident management systems and reduce downtime.

Key insights

TingIS uses LLMs and multi-stage linking to extract actionable incidents from noisy, high-throughput customer data.

Principles

• Combine LLMs with efficient indexing for event merging.
• Integrate domain knowledge for noise reduction.
• Cascaded routing improves business attribution.

Method

TingIS employs a multi-stage event linking engine with LLMs for merging, cascaded routing for attribution, and a multi-dimensional noise reduction pipeline using domain knowledge, statistics, and behavioral filtering.

In practice

• Use LLMs for semantic event merging.
• Implement cascaded routing for incident attribution.
• Apply behavioral filtering for noise reduction.

Topics

• Real-time Risk Discovery
• Customer Incident Management
• Large Language Models
• Cloud-Native Services
• Noise Reduction

Best for: AI Architect, Research Scientist, CTO, AI Scientist, AI Engineer, MLOps Engineer

Related on AIssential

• EventADL: Open-Box Anomaly Detection and Localization Framework for Events in Cloud-Based Service Systems — EventADL provides open-box anomaly detection and root cause localization for cloud event data, leveraging semantic and frequency patterns.
• I Predicted the T20 World Cup 2026 Final With an Anomaly Detection Pipeline. — Anomaly detection system failures reveal more about production challenges than successes, especially regarding evidence thresholds and cold-start problems.
• Lazy or Efficient? Towards Accessible Eye-Tracking Event Detection Using LLMs — An LLM-driven pipeline automates eye-tracking event detection, making complex analysis accessible without specialized programming.
• When Alerts Mean Nothing: How AI Could Fix the Noise Problem in IT and Security — AI can mitigate alert fatigue by using behavioral analysis to distinguish abnormal events from normal activity.
• Real-Time vs Post-Call: Behind EdgeTier's Architecture 💡 — Prioritizing "near-time" post-call processing over real-time offers greater reliability and integration ease for contact center analytics.
• Your On-Call Queue Is Hiding Your Most Critical System Design Flaws. — Agentic AI can precisely classify high-volume incidents, distinguishing critical system regressions from noise.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.