#50 – Can AI Run Production? Inside ResolveAI with Spiros Xanthos

Tessl founder and CEO Guy Podjarny joins Dash0’s Mirko Novakovic to explain why AI coding is only the beginning of a much bigger shift: from developers writing code to teams building software factories that can learn and improve themselves. They dig into how Tessl uses skills, evals, and production feedback to make that possible, and why Guy believes hand-written production code is ultimately going away.

ResolveAI CEO Spiros Xanthos joins Dash0’s Mirko Novakovic to make the case that the AI SRE is moving beyond incident summaries and into the critical path of production.Spiros explains what it takes to trust an agent to investigate an alert and one decides whether a human needs to be paged or whether it can recommend or take mitigation action. They also get into why Resolve is training its own models and why observability tools will need to evolve for a world where agents do most of the querying — not humans.

Cribl co-founder and CEO Clint Sharp joins Dash0’s Mirko Novakovic for a wide-ranging conversation about where observability goes next. Clint explains why he thinks the future is flexible data layers rather than all-in-one platforms, plus how AI agents change the economics of telemetry and the way that data gets queried. He also reveals the top things he’s hearing from his most advanced customers and why enterprises still need purpose-built infrastructure instead of simply dumping everything into a general-purpose data warehouse. Mirko and Clint also debate querying across multiple systems, AI SRE and what happens when dashboards are no longer the primary interface for production data.

Dash0 is acquiring Polar Signals. In this special behind-the-scenes episode, Polar Signals founder and CEO Frederic Branczyk joins Mirko Novakovic to share how a chance dinner in Berlin turned into a plan to join forces, and why continuous profiling and Polar Signals’ Great Lakes database were the missing pieces Dash0 was already looking to build. They dig into what profiling reveals that traditional observability can’t, how AI can now make performance optimization accessible to every engineering team, and what the acquisition unlocks for Dash0 customers, from always-on CPU profiling to an entirely new approach to GPU performance.

Platform engineering leader Rachael Wonnacott joins Code RED co-host Kasper Borg Nissen to explain why great platforms are defined by the experience they create for developers. Drawing on years of building internal platforms in large enterprises, she explains how poorly designed abstractions can turn a golden path into a golden cage, why reducing developer cognitive load often means transferring that complexity to the platform team, and how observability helps both sides understand what is happening beneath the abstraction. They also explore context-driven development: using platforms to give AI agents the architecture, constraints, and production signals they need to make better decisions. Links mentioned in this episode:Patrick Debois — Context-Driven Development: https://tessl.io/blog/context-development-lifecycle-better-context-for-ai-coding-agents/Daniel Bryant — Golden Bricks (KubeCon NA 2024 talk): https://www.youtube.com/watch?v=qhfQfQmnNd4Rachael Wonnacott — The Conservation of Cognitive Load (FastFlow keynote): https://www.youtube.com/watch?v=H4CTkModyEk

In this special launch episode, Dash0 Director of Product Evgeni Wachnowezki joins Mirko Novakovic to introduce Beyond Observability, Dash0’s vision for closing the loop from AI-generated code to production. They discuss why AI has shifted the bottleneck from writing code to safely shipping and operating it, and how Dash0 is building the control plane for that new reality. They also walk us through their new releases: AI Coding Insights, Agent0 Automations, AutoMerge and SignalControl.

Abdel Sghiouar, Senior Cloud Developer Advocate at Google and KubeCon + CloudNativeCon co-chair, joins Kasper Borg Nissen to discuss what’s changing across the cloud-native ecosystem. They explore why AI and agentic workloads are becoming a dominant theme in Kubernetes, why platform engineering teams are under more pressure than ever, and why the industry is still inventing best practices in real time. Abdel also shares lessons from his career in developer advocacy and the basic formula for a great conference talk.Links mentioned in this episode:Kubernetes for agentic apps: A platform engineering perspective – https://platformengineering.org/blog/kubernetes-for-agentic-apps-a-platform-engineering-perspectiveThe Kubernetes Podcast: https://kubernetespodcast.com/More from Abdel: https://linktr.ee/boredabdel

Heinrich Hartmann joins Dash0’s Mirko Novakovic to share how Zalando runs their observability at an extraordinary scale. Heinrich explains why Zalando standardized on OpenTracing and OpenTelemetry across 3,500 microservices and 250 Kubernetes clusters, what it takes to process up to 60 million spans per second during Cyber Week, and how the team approaches sampling, cost control, and adaptive telemetry. They also discuss the future of observability in the age of AI and Heinrich’s new conference, SignalsConf Berlin, which will focus on reliability and AI.

Buoyant CEO and creator of Linkerd William Morgan joins Dash0’s Mirko Novakovic to explore what happens when you instrument traffic at the service mesh layer. Drawing on his experience at Twitter, William explains how Linkerd gives platform teams HTTP and gRPC-level visibility without touching application code, why retry budgets help prevent cascade failures, and how workload identity creates a stronger foundation for security and observability. They also discuss MCP traffic, AI agents in production, and how observability needs to be re-optimized for LLMs.

Jad Naous, founder and CEO of Grepr, joins Dash0’s Mirko Novakovic to tackle one of observability’s biggest problems: too much data and not enough signal. Drawing on his experience at AppDynamics and Apache Druid, Jad explains why current architectures are fundamentally broken, and how Grepr compresses, aggregates and extracts signal directly in the stream, reducing data volume by up to 90% before it ever hits tools like Datadog. They dive into real-time pattern detection across logs and traces, why rule-based pipelines fall short, and how high-signal data is the foundation for any future AI SRE system.