Getting good at DevOps doesn't just mean you'll learn CICD, it means you'll learn: 1. Cloud 2. Containerization and Orchestration (like Kubernetes) 3. Scalability and performance optimization 4. True software engineering practices 5. Business requirements and needs 6. Cyber

Speaker badge obtained 😎 One of the coolest things about speaking this year was the session was so interactive. I specifically asked people to not wait until the end for questions. Interrupt me! Interactive sessions are the best sessions. Thanks so much for coming to my

MCP Servers give access to tools. These "tools" are a specific set of functionality (list repo branches, get Pods, create Services, etc.). The goal with these tools is to extend the knowledge and capability of an Agent. In this blog post, you'll learn how to create MCP Servers

Today’s the day! I’ll be giving a talk with my colleague James on using AI Agents to stress test a failing Kubernetes environment and have it fix the environment. Come check it out! https://t.co/NlXo4SVZwP #kubecon @CloudNativeFdn #agenticai

You typically see two types of Models: ✅ SaaS-based Models (GPT, Claude, Gemini, etc.) ✅ Local Models (Llama) The goal with local Models is to give you more control and governance on the data that an Agent has access to. In this blog post, you'll learn how to deploy a Llama

Sitting at the airport working on my #kubecon demo! Observability with Istio Ambient Mesh and Kiali for AI Agents #kagent @CloudNativeFdn

Talk is updated! See you all there. #kubecon #cloudnaticecon

Just got word that I’ll be a co-speaker at KubeCon 🔥 With my colleague, James, on Wednesday (talk below). If you’re around, come check out how we use AI to push a cluster over the edge… and then bring it back. @CloudNativeFdn #kubecon

It's KubeCon week 🚀 And there's going to be a lot of talk about MCP Servers. So it only makes sense to know what MCP is underneath the hood. I created this video to help everyone going to KubeCon this week have a clear understanding of: ✅ What MCP Servers are ✅ How

MCP Servers typically have a lengthy list of tools. 🤔 The question is - should Agents have access to EVERY tool? In my opinion, the answer is no. From a functionality and security perspective, no Agent needs "root access" to every single tool available. Make Agents specific

Multi-Model Failover 🔥 Three Models used: ✅ Claude 3.5 ✅ GPT 3.5 Turbo ✅ Claude Opus The goal here is to say "well, if one Model is down, having throttling issues, or slow responses, I still need my Agentic infrastructure to work..." and now it can 😎 #kubernetes

Creating an: 1. Agent 2. Agent Server 3. MCP connection to the Kubernetes MCP Server Using Agent Development Kit (ADK) from @GoogleAI

Some simple testing of the cloud-native architecture diagram MCP Server.

And here it is running in agentgateway

Sunday Funday. Built an MCP Server that should help with generating architecture diagrams. Let me know if you try it and if you run into any issues! https://t.co/PSnpvnnh04

If you haven't used the MCP Inspector tool yet, it's free and literally all you have to do is run: `npx @modelcontextprotocol/inspector` It's an MCP Client.

Keep your MCP Tool list to a MAX of 20 tools per Agent. After 20 (and honestly, even between 15-20), you'll begin to see hallucinations. This makes sense. Think about it - how much information can your brain hold onto before you begin to forget things?

Exited to be on the Program Committee for Agentics Day: MCP + Agents for KubeCon EU 😎 I'll be helping out with some other amazing reviewers for CFP submissions on all things 1. MCP 2. Agentic and more! If you haven't submitted a CFP already and you're planning to, the link

Just got word that I’ll be at KubeCon in a few weeks! If you want to come hang out with some fellow MCP, LLM, and Agent enthusiasts, feel free to stop by. I want to do some live coding sessions and building, so bring a use case for me and we’ll figure it out on the spot.

MCP isn't even a year (will be next month). And it's already the standard for connecting external system to Agents. That's how fast AI is moving.