L'IA appliquée à la Supervision IT et l'Observabilité - EP 18

In this second part, Vincent and Denis Roussel (Senior Data Scientist at Centreon) move beyond theory and delve into real-world business use cases. Discover how AIOps is transforming observability through three key pillars: the development of new features, predictive analytics, and a complete overhaul of user interactions. Forget the hype surrounding LLM: for detecting network anomalies (Anomaly Detection), predicting hard drive capacity (Predictive Capacity), or identifying a failing server in a cluster (Outliers), traditional Machine Learning remains far more efficient and less expensive! We'll also discuss a major revolution: will AI spell the end of complex graphical user interfaces (UIs) in favor of natural language processing (NLP) interactions? Finally, Denis reveals the concrete projects the Centreon teams are currently working on (MCP servers, massive log summarization, support chatbots). 🎧 Listen to the podcast on your favorite platform: Spotify: https://open.spotify.com/show/24LgLR4... Deezer: https://www.deezer.com/fr/show/100132... Apple Podcasts: https://podcasts.apple.com/us/podcast... AntennaPod: https://antennapod.org/deeplink/subsc... 👉Free trial: https://www.centreon.com/free-trial/?... ⏱ Episode Chapters: 00:00 - Introduction: Applying AI to IT Professions 01:40 - The 3 Main Pillars of AI in Observability 04:15 - Beyond LLM: Anomaly Detection and Predictive Capacity 06:05 - Identifying Outlier Behaviors: The "Cluster Outlier" 07:30 - How Data Scientists Conduct Technology Watch 08:45 - The UX Revolution: The End of Graphical Interfaces in IT? 12:20 - The limitations of natural language for troubleshooting 14:50 - Inside Centreon's labs: Chatbots, MCP Servers, and Log Management 18:10 - Is AI being oversold by marketing (AI-Washing)? 19:30 - Should we put AI everywhere? (Security, privacy, and local models) 21:10 - Conclusion: Choosing the right tool for the right problem #AIOps #Observability #ITMonitoring #MachineLearning #LLM #DataScience #Centreon #ITOM #PredictiveCapability #TechPodcast