Operations | Monitoring | ITSM | DevOps | Cloud

Last year, Lumen shared its AIOps roadmap. This year: two updates. First, how Lumen's thinking on AI-era org design has shifted — including why the instinct to cut junior headcount is a trap, and what sustainable team structures look like instead. Second, what Lumen built as a result: a single AI core (AskGreg) extended into a customer-facing email agent (NORA), a chatbot, and an in-progress voice agent — all leveraging the same modular platform. Live video demos show the system diagnosing network problems from alarm data and driving real-time network decisions. The closing thesis ties it together.

AI in network operations has moved past the demo. In some shops it's already cutting detection time and clearing the alert noise that used to bury the engineering team. In others, a promising rollout has quietly stalled the moment it met production. That difference between the two rarely comes down to the model, it comes down to how the AI gets integrated into the network and most importantly, the team around it.

Time-series view of how consistently a reservation has been utilized and which resources are consuming it. Makes underutilization or coverage gaps visible at a glance instead of requiring manual investigation.

Alerts when a schedule has been disabled for longer than a set threshold (e.g., 7+ days). Solves the common failure mode where a team disables a shutdown schedule for maintenance and forgets to turn it back on, quietly costing money.

A dynamic alternative to static schedules. Define filters (subscription, resource group, tag) once, and matching VMs are automatically picked up, including new ones added later. Supports both metric-based triggers (e.g., CPU under 5% after 8pm) and time-based power on/off.

Sometimes there's a vulnerability in a package you can't remove. Legacy dependencies don't always come with an easy fix. The one thing you don't do is bury it in the backlog with no dates and no reasoning attached. This video covers what a defensible response actually looks like: Your ability to show your work on unfixed vulnerabilities is more important than simply identifying them.

Many times, it feels like you're maintaining multiple versions of the same dashboard (with a slight modification), *OR* simply spending more time writing queries rather than actually looking at the actual data? In this Campfire community call, we're taking a deep dive into two things that are reshaping how Grafana dashboards get built: Dynamic Dashboards and the Grafana AI Assistant and showing you how to combine them to go from a blank canvas to a reusable, production-ready dashboard in minutes.

You get an answer from the agent and no way to check how it got there. The route it took is recorded, and so is the reason it gave for taking it. AURA emits OpenTelemetry spans, and the Docker quickstart wires them straight into Phoenix. Four services come up together: AURA Web Server as the persistent agent harness, LibreChat as a browser interface for chatting with the agent, Phoenix to receive the spans, and MongoDB to store stateful data for LibreChat. The Compose file arrives pre-configured to point AURA at Phoenix and to enable content recording for the local demo.

The same agent that reads a Kubernetes cluster reads a Windows machine. One MCP server, one config entry, and the questions are the ones you already ask.

Learn how to delay OnPage notifications for a defined period so alerts can be queued and sent later without interrupting the on-call team in real time. OnPage’s Delay Notifications option temporarily holds notifications during a scheduled window and delivers the queued alerts afterward as low-priority notifications. This is useful when teams want to reduce interruptions during certain hours at night for use cases that allow them to do so, or planned activities while still preserving alerts for later review.