Last updated: November 18, 2025
Code Red Newsletter #20
KubeCon North America has just wrapped, and honestly, it might have been the most energized one yet. The hallway track was in full force, the buzz never really dipped, and everywhere you turned people were deep in conversations about observability, OpenTelemetry, and all things cloud native.
At the Dash0 booth, things got a little wild - backpacks flying out faster than we could restock, Magic the Gathering cards changing hands like collector’s gold, and a steady line of engineers wanting to talk about Observability and Agent0. We also handed out hundreds of copies of OpenTelemetry for Dummies, which disappeared almost as fast as the coffee. It turns out nothing sparks a hallway conversation quite like a bright yellow book and a shared obsession with better telemetry.
In focus: Agentic AI in Observability
Every now and then, observability gets a plot twist. This time, it comes in the form of Agent0 - a cast of specialized agentic AIs woven directly into Dash0. Think of them as the supporting characters you always wanted in your debugging storyline: the ones who surface clues, connect the dots, and quietly do the heavy lifting so you can get back to building.
Here’s what it’s all about.
Introducing Agent0 - Dash0’s Agentic AI Platform for Observability
Agent0 brings a team of focused AI specialists into your observability workflow - the Seeker, the Oracle, the Threadweaver, the Pathfinder, and the Artist. Each one takes on a different piece of the puzzle, from triage to data quality to dashboards and queries, working together to turn raw telemetry into real insight. The result is observability that feels less like detective work and more like… having a support party.
An Observability Veteran on AI’s ‘Intoxicating’ Potential
In this guest post for The New Stack, Michele Mancioppi digs into the intoxicating potential of AI - and the temptation to hand over the keys too quickly. He writes about the blend of excitement and responsibility that comes with applying AI to debugging, why the human still needs to stay in the loop, and how agentic workflows can bring real context to telemetry instead of just more noise.
When Two Passions Collide: The Making of the Agent0 Magic the Gathering Cards
To celebrate Agent0’s launch, Michele leaned fully into Dash0’s collective nerdery and created a custom set of Magic the Gathering cards - one for each agent. The post walks through card types, colors, lore, power levels (wildly overpowered, as they should be), and the surprisingly complex process of making them feel “real.” They were an instant hit at KubeCon and disappeared long before lunch on Day 1.
Observability for Platform Engineers white paper
This whitepaper from PlatformEngineering.org extends the course Michele and I built. It reframes observability as an internal product: with self-service instrumentation, OpenTelemetry-native defaults, and guardrails so you don't blow the budget before lunch. The vision: observability that quietly does its job, without developers needing to fight it.
Choice cuts
Because KubeCon might be over, but the hot takes keep coming.
OpenTelemetry eBPF Instrumentation Marks the First Release
The first alpha of OpenTelemetry eBPF Instrumentation (OBI) has landed - a major milestone for zero-touch observability. It captures traces and metrics straight from the kernel, with no restarts, SDKs, or code changes required.
How many pillars of observability can you fit on the head of a pin?
Charity Majors returns with another sharp take, this time aimed directly at the beloved (and deeply misleading) “three pillars.” Her argument is simple: you don’t troubleshoot columns; you troubleshoot questions, context, and systems. Hard to disagree.
Kill The Three Pillars Manifesto
John Gallagher pushes the conversation even further: logs, metrics, and traces aren’t pillars - they’re data formats. Real observability comes from wide events, shared context, and signals that connect. A manifesto for those tired of diagrams that oversimplify the real work.
If KubeCon was any indication, the next phase of observability is already here - AI agents helping humans make sense of signals, kernel-level instrumentation reshaping what “auto” means, and the community pushing the boundaries of what’s possible. We even saw the first experimental release of the Zig-based OpenTelemetry injector we upstreamed with the community, marking a big step toward more portable, reliable auto-instrumentation.
See you in two weeks with more signals, stories, and maybe a few post-conference reflections from the world of open observability.
