What Do You Mean It’s Broken? When AI Meets the Harsh Reality of Te...
summary
AI-driven observability framework called Telco Observer combines live telemetry (logs, metrics) with static knowledge (standards, vendor docs) and a query optimization agent that picks expansion, composition or disambiguation strategies per query. Implementation runs across multiple Kubernetes clusters, integrates with MANO, and a proof of concept on free5G on k8s finds three UPF vendors and gives “pretty good” troubleshooting and data-retrieval results. Cites industry numbers: 97% adopting AI, top models reach 83% accuracy while many pilots fail, and the system trades latency for accuracy.
AI-written summary · about
chapters (12)
- 0:00AI in telecom networks
- 1:02Telco AI adoption challenges
- 3:33Observability in telecom
- 5:40Telco Observer framework
- 7:06From observability to AI assistant
- 9:33Why monitoring fails in telco
- 10:58Observability agent architecture
- 13:46Query optimization and reasoning
- 16:35Autonomous networks future
- 18:23Summary
- 19:04Q&A: AI hallucinations in telecom
- 20:29Q&A: AI scope and verification
related talks
28mDO WE STILL NEED TO "OBSERVE" THE FUTURE OF AI & OBSERVABILITY
38mAI as a Platform Engineer: Explaining Kubernetes Failures, Not Just Detecting Them
28mImplementing AI-Driven Observability in Healthcare
30mAI Agents for Incident Management
25mAI-Powered O11y in Microservices Architectures
20mO11y in the Evolution of AI-Era Networking Infrastructure
27mAI, Innovation, Standards & Scale for Large Telcos
44mAgentic Observability — The Path to AI Co-SRE