{"podcast":{"title":"Heavy Networking","slug":"heavy-networking","podcast_index_feed_id":163010,"rss_url":"https://feeds.packetpushers.net/PacketPushersWeeklyPodcast","website_url":"https://packetpushers.net/","image_url":"https://static.feedpress.com/logo/packetpushersweeklypodcast-653bc236a0040.png","author":"Packet Pushers","episode_count":250,"summary":"Heavy Networking is an unabashedly nerdy dive into all things networking. Described by one listener as \"verbal white papers,\" the weekly episodes feature network engineers, industry experts, and vendors sharing useful information to keep your professional knowledge sharp and your career growing. Hosts Greg Ferro, Ethan Banks and Drew Conry-Murray cut through the marketing spin to explore what works—and what doesn't—in networking today, while keeping an eye on what's ahead for the industry. On air since 2010, Heavy Networking is the flagship show of the Packet Pushers podcast network.","last_synced_at":"2026-07-25T06:21:53.014209+00:00","page_url":"https://stenobird.com/podcast/heavy-networking"},"episode":{"title":"HN824: That’s Not a Job for an LLM: The Right Way to Apply AI to Network Operations (Sponsored)","slug":"hn824-that-s-not-a-job-for-an-llm-the-right-way-to-apply-ai-to-network-operations-sponsored","published_at":"2026-04-24T16:58:10+00:00","page_url":"https://stenobird.com/podcast/heavy-networking/hn824-that-s-not-a-job-for-an-llm-the-right-way-to-apply-ai-to-network-operations-sponsored","show_page_url":"https://stenobird.com/podcast/heavy-networking","url":"https://packetpushers.net/podcasts/heavy-networking/hn824-thats-not-a-job-for-an-llm-the-right-way-to-apply-ai-to-network-operations-sponsored/","audio_url":"https://feeds.packetpushers.net/link/12486/17324947/HN824.mp3","summary":"Moving beyond the hype, this discussion explores the practical application of AI and LLMs in network operations. The conversation focuses on using AI as a specialized tool for investigation and automation rather than a replacement for fundamental networking logic.","meta_description":"Learn how to effectively apply AI and LLMs to network operations without falling for the hype, featuring Kentic CEO Avi Freedman.","key_points":["Main idea: AI should be viewed as a set of specialized tools with specific use cases and inherent limitations rather than a universal solution","Practical takeaway: Use LLMs as an interface or 'advisor' to query existing machine learning insights and telemetry rather than feeding raw flow records directly into them","Failure mode: Relying on LLMs for deterministic network design tasks can lead to errors because they lack a complete, real-time worldview of complex network topologies","Practical takeaway: Agentic workflows can automate the 'drudgery' of investigation, allowing humans to focus on high-level decision-making and verification","Main idea: The future of network observability lies in composing specialized agents that can access APIs and internal tools to perform multi-step troubleshooting"],"chapters":[{"start_ms":60000,"title":"Defining AI in Networking","summary":"A discussion on moving past the polarized views of AI hype to treat it as a functional set of tools with specific capabilities and limitations."},{"start_ms":330000,"title":"The Evolution of AIOps","summary":"Reflecting on the history of intent-based networking and the transition from early automation to modern AI techniques."},{"start_ms":610000,"title":"The Limitations of LLM Reasoning","summary":"Comparing the current state of LLM 'reasoning' to early GPS technology—useful but prone to critical errors in complex environments."},{"start_ms":1160000,"title":"The Importance of Network Worldview","summary":"Why LLMs struggle with network troubleshooting due to a lack of understanding regarding hardware bugs, routing tables, and forwarding databases."},{"start_ms":1440000,"title":"Building Guardrails and Semantic Models","summary":"Exploring the potential for using semantic reasoning and guardrails to simulate sensicality in automated network tasks."},{"start_ms":1725000,"title":"Agentic Architectures","summary":"Discussing whether the future lies in specialized single-purpose agents or general-purpose orchestrators for network tasks."},{"start_ms":1995000,"title":"Human-in-the-loop Investigation","summary":"How AI can perform the initial heavy lifting of investigation and present findings for human validation and direction."}],"topics":["Network Operations","Artificial Intelligence","Large Language Models","AIOps","Network Observability","Automation","Telemetry","Agentic Workflows"],"duration_seconds":3671,"processing_state":"processed","actions":[{"name":"request_transcript","method":"POST","url":"https://stenobird.com/v1/public/podcasts/heavy-networking/episodes/hn824-that-s-not-a-job-for-an-llm-the-right-way-to-apply-ai-to-network-operations-sponsored/transcription-requests","description":"Idempotently request low-priority transcript generation for this episode."},{"name":"read_markdown","method":"GET","url":"https://stenobird.com/podcast/heavy-networking/hn824-that-s-not-a-job-for-an-llm-the-right-way-to-apply-ai-to-network-operations-sponsored.md","description":"Read the agent-friendly Markdown representation of this episode resource."}]}}