SAIM SAFDAR

@theScottrHall @motyar That's interesting. Which category is harder to pick? Need to check whether it passes the test "50-tool MCP server burns ~8K tokens before the first question."

@motyar @cloudnativeboy There is a third path: MCP to direct API. The MCP just passes the api definition to the agent and the agent directly executes the API. This way you use each protocol for what it does best UseJoshua.com

MCP vs CLI MCP promises standardized tool discovery for AI agents, but there's a hidden cost: a MCP server with 40–50 tools? You've burned tens of thousands of tokens before asking a single question. CLIs win when context efficiency matters 👇 youtube.com/watch?v=TuIZoF…

🚨 Kubernetes v1.36 “Haru” is here and it’s a major step forward for production stability and AI-native workloads. Highlights: • Mutating Admission Policies (GA) • User Namespaces (GA) • OCI VolumeSource (GA) • DRA for specialized AI hardware • Scale-to-Zero HPA (Alpha)

Reporting from AWS Summit London 2026 🚀 Happening now on #CloudNativeFM & #CloudTherapist, featuring cloud startups, practitioners, and vendors building in AI, DevOps platform engineering. Join us Live -> youtube.com/live/tsDN5RxzI…

We’re going LIVE from ExCeL London at AWS Summit 2026 🎥 youtube.com/watch?v=tsDN5R… Join us as we feature conversations with vendors, practitioners, cloud startups, and builders across DevOps, AI, Kubernetes, and Platform Engineering. Real insights. Real people. Real systems.

📊 The latest CNCF blog breaks down how to measure the ROI of developer tools using: ✅ DORA signals from CNCF projects ✅ Qualitative dev surveys ✅ Cost-based analysis for large teams Read more: bit.ly/4tbDcPf

My take: if we already manage identities, APIs, and infrastructure, we now need to manage agents the same way with a registry, policies, and governance from day one. That’s the only way to scale agentic AI safely in the enterprise.

Happening next week. 📍 𝗘𝘃𝗲𝗻𝘁 𝗗𝗲𝘁𝗮𝗶𝗹𝘀 📅 𝟮𝟭𝘀𝘁 𝗔𝗽𝗿𝗶𝗹 𝟮𝟬𝟮𝟲 ⏰ 𝟵:𝟬𝟬 𝗔𝗠 – 𝟯:𝟬𝟬 𝗣𝗠 📌 𝗜𝘀𝗹𝗮𝗺𝗮𝗯𝗮𝗱 𝗖𝗵𝗮𝗺𝗯𝗲𝗿 𝗼𝗳 𝗖𝗼𝗺𝗺𝗲𝗿𝗰𝗲 & 𝗜𝗻𝗱𝘂𝘀𝘁𝗿𝘆 🔗 𝗥𝗲𝗴𝗶𝘀𝘁𝗲𝗿 𝗵𝗲𝗿𝗲: meetup.com/aws-cloud-club…

Great conversation at the @dash0hq booth during KubeCon Amsterdam 2026 🚀 Dash0 is building an agentic observability platform natively on OpenTelemetry, with a strong focus on transparency, product clarity, and the future of AI-driven observability. Link 👇

VEXXHOST's platform Atmosphere runs OpenStack on top of K8s, giving teams a prod-grade private cloud with zero vendor lock-in and no egress surprises. For VMware migrations specifically, their MigrateKit tooling helps teams move without rebuilding from scratch.

During KubeCon Amsterdam 2026, we stopped by Chainguard booth Hannah shared how @chainguard_dev is helping teams secure the software supply chain w/zero-CVE container images, rebuilt from source every 24 hours, plus a minimal attack surface for containers & language libraries.

🚨 A high-vulnerability CVE (CVE-2026-4342) has been identified in ingress-nginx. This vulnerability enables configuration injection and potential code execution on all versions below v1.13.9, v1.14.5, and v1.15.1. Details: github.com/kubernetes/kub…

We are live on @cloudnativefm Join us for LIVE AMA: 1/ Anthropic Economic Index: measuring AI with primitives 2/ Cloud native skills in demand right now and into 2026 3/ Abstraction vs ownership. Join here: youtube.com/live/eVbF0CbEd…

Bookmark 📍 youtube.com/live/eVbF0CbEd…

From Lenovo’s CES AI direction to Anthropic’s new AI economic primitives, and from the ongoing debates around abstraction vs. ownership to runtime security, agent governance, K8s 1.35 in-place pod rescaling the future of cloud-native skills, Join us live today, Link 👇

We are Livestreaming @KubeCon_ Amsterdam 2026 Afterthoughts: Happening right now youtube.com/live/c4eDw70Rj…. From security and policy to storage, networking, developer experience, and AI-powered operations, this episode pulls together the most important takeaways from the floor.

Great convo at Cosmonic’s booth during KubeCon Amsterdam live coverage with @Hectaman on why WebAssembly matters so much for the future of cloud native workloads A strong look at how WebAssembly, AI, cloud native security are coming together Watch: youtube.com/shorts/xt3C06B…

As AI agents become more autonomous and non-deterministic, orgs need visibility and control over both sanctioned and shadow agents. @tigeraio approach focuses on discovering agents, auth/authZ ing them with human-granted or agent-to-agent permissions, & enforcing policies ETE.