Heavy Networking HN812: Nokia EDA: AI Ops You Can Trust (Sponsored)
Jan 30, 2026
Stephen Butler, product line manager at Nokia for Event-Driven Automation, brings a short practical bio and tech focus. He discusses multi-vendor abstractions and combined config/state models. He explains schemas as an LLM decoder ring and workflow-driven troubleshooting. He outlines digital twin testing and modular apps that extend trustworthy AIOps.
AI Snips
Chapters
Transcript
Episode notes
Single Normalized Resource Model
- EDA uses multi-vendor abstractions that present a single, normalized resource view regardless of underlying OS.
- That unified spec+status model lets tools and LLMs reason about intent and operational state consistently.
Give LLMs Focused Context
- Provide LLMs with the right context by pre-selecting relevant resources and schemas before invoking reasoning.
- Use RAG and local preprocessing so the model sees only the structured inputs it needs.
Schemas As A Decoder Ring
- EDA exposes well-defined schemas (spec and status) for resources so models can parse fields without special training.
- Schemas act like a decoder ring, letting LLMs understand inputs, outputs, and metadata reliably.