New on Claude API: Structured outputs

Define your schema once. Get perfectly formatted responses every time. No more complex parsing logic or wasted tokens on retries and the best part: no impact on model performance

Available - Sonnet 4.5 and Opus 4.1 in public beta

Helm has officially released v4.0.0.

One of the biggest changes is the removal of Helm 2 migration code and old compatibility layers.

Helm v4 also improves security/maintainability by dropping unsupported APIs & removing older commands that were rarely used.
helm.sh/docs/overview/

Disrupting the first reported AI-orchestrated cyber espionage campaign A report describing an a highly sophisticated AI-led cyberattack

Anthropic confirmed the first documented case of a large-scale, AI-orchestrated cyber-espionage campaign, one where AI agents, not humans, executed 80–90% of the attack.
www.anthropic.com/news/disrupt...

Happy Friday 👋

He who knows not and knows not he knows not is a fool. "Shun him."

He who knows and knows not he knows is asleep, "Awakens him."

He who knows not and knows he not knows is hungry. "Feed him."

He who knows and knows he knows not is wise. "Follow him."

If you’re still sending raw JSON into your LLMs, you’re burning tokens, latency, and budget!

Try TOON (Token-Oriented Object Notation).

Clear like YAML, compact like CSV:

• 30–60% fewer tokens
• Up to 50% lower costs
• Shines for tabular data.

Read more:
medium.com/medialesson/...

What we captured: practical playbooks for reducing cloud costs, guardrails for secure supply chains, patterns for platform engineering, and repeatable automation standards that bridge R&D and enterprise ops.

Thanky'all for sharing hard-won lessons and shipping-ready guidance.

Highlights I loved:
• @portainerio: : smart container management that simplifies ops, ship faster with less cognitive load.
• @chainguard.dev: solid guardrails & supply-chain security for OSS + enterprise.
• InfrOS: forward-looking approaches to continuous governance, cost optimization...

Big thanks to Guy Brodetzki (InfrOS), Adrian Mouat (Chainguard), Neil Cresswell (@portainer.io), and Sarah Polan for joining our KubeCon Atlanta livestream. 🙏

Watch the highlights: www.youtube.com/watch?v=Mbpe...

JSON vs TOON — A new era of structured input? Why structure matters more than ever

One day in the afterNOON, you realize LLM prompts can be heavy on cost and tokens.

Meet TOON, a new file format sitting b/w JSON & CSV:

💸 Token-efficient: reduce token counts by 30–60%
🤿 LLM-friendly: built-in guardrails for safer outputs
🍱 Minimal syntax & more
medium.com/medialesson/...

Watch all the highlights on this link 👇

#KubeCon Atlanta Day 02 Live Coverage: youtube.com/watch?v=Db9x...

Day 03 Live Coverage starts here 11 am EST | 4 pm GMT: youtube.com/watch?v=Mbpe...

Thanks ALL, Rose leads efforts at "Code First Girls" to help underrepresented talent thrive in dev & engineering roles

Jon Brookes is creating Minimal Viable K8S, which'll be featured in future episodes #cloudnativefm

Gurdip Kalley Zivan from datafy.io to helps teams cut cloud block-storage costs

Just learned today, "Datafy helps teams cut cloud block-storage costs and remove EBS waste by auto right-sizing." Their platform combines EBS analytics, continuous monitoring, and an EBS Auto-Scaler that adjusts volume size and utilization. Thanks, Gurdip & Zivan.
www.youtube.com/watch?v=Db9x...

We are now LIVE on #CloudTherapist, join us and know the #KubeCon #Atlanta Vibes and LIVE AMA: www.youtube.com/watch?v=Db9x...

Falco Links Real-Time Detection with Forensic-Level Analysis in the Cloud Native Stack New integration connects Falco alerts to Stratoshark’s forensic tools, delivering Wireshark-style visibility into system call and audit log data Key Highlights ATLANTA—KUBECON + CLOUDNATIVECON NORTH…

🚨 Real-time detection meets forensic-level visibility!

Falco now integrates with Stratoshark, connecting alerts to forensic captures so teams can go from detection to deep investigation instantly.

👉 Learn more: hubs.la/Q03S69Cp0

CNCF Launches CNPE Certification The Certified Cloud Native Platform Engineer (CNPE) joins CNCF’s expanding certification portfolio as it meets growth training demand Key Highlights KUBECON + CLOUDNATIVECON NORTH AMERICA, ATLANTA…

CNCF launches the Certified Cloud Native Platform Engineer (CNPE), and it's currently in BETA., I'm lucky to be a part of the exam builder, validating real-world skills of architecting IDPs.

HERE YOU GO: "Certified Kubernetes Network Engineer (CKNE) is coming soon"!
www.cncf.io/announcement...

Implementing LMCache Plugin Framework & lmcache_frontend: Design Philosophy | LMCache Blog A flexible plugin system for enhanced observability and management Abstract In large-scale language model inference scenarios, efficient memory management and KV cache optimization are crucial. LMCach...

In large-scale LLM inference scenarios, efficient memory management & KV cache optimization are crucial LMCache, as a KV cache management system specifically designed for vLLM, requires more flexible extension mechanisms meet needs of monitoring/troubleshooting & more blog.lmcache.ai/en/2025/09/2...

CNCF | Dosu Dosu is an AI teammate that lives in your GitHub repo, helping you respond to issues, triage bugs, and build better documentation.

For those of you ✈️ #KubeCon #Atlanta or already there, the best way to find talks, project pavilion, contrib fest, co-located days details, where to find best ambassadors, ask ask.cncf.io, you'll get there #kubecon #cloudnativecon thanks @dosu_ai for building this!!

State of Cloud Native Development Cloud native is the foundation of modern software. The latest CNCF + SlashData State of Cloud Native Development Report reveals how developers are scaling, automating…

🔥 of the presses, 15.6 million developers are building with cloud native technologies. The CNCF State of Cloud Native Development Report, in partnership with SlashData, reveals emerging trends, from hybrid cloud's surge to growing AI developer adoption.

Get the report now: hubs.la/Q03SBh210

Whether you are prototyping on local hardware or deploying to a production K8s cluster, you need a unified API that abstracts infra complexity. That’s exactly what the Kubeflow Python SDK delivers.

Kubeflow SDK: Here we GO 👇
blog.kubeflow.org/sdk/intro/

For those of you who're travelling to ✈️ Atlanta for KubeCon,

Remember Network = Net Worth 🚀

Have a great event.

Still holds the record of ZERO in-person KubeCon

Does anyone hold a similar record? Say Hi 👋

Don't worry, you can join #cloudnativefm ✈️ to see the in-person VIBES.

#CloudTherapist & #CloudNativeFM hosting a Virtual Livestream Party KubeCon NA Atlanta.

The Livestream will continue for about 1.5 hours & tell us about what you have been up to lately, exciting news about new product releases, your company announcements.... Tag your friends, company - comment 👋

Canine - An open source alternative to Heroku Canine is an open source deployment platform that makes it easy to deploy and manage your applications.

In this episode of #cloudnativefm we speak with Neil Creswell, CEO of @portainer.io, about the art of simplifying container management for individual developers, platform teams, and large enterprise fleets.

Two new projects Neil teases: Canine.sh and KubeSolo.io

Watch: youtu.be/P1LK6-Hd56Q

Dropped another episode on #cloudnativefm

Digital Sovereignty. A topic no one is paying attention to is now covered with all the details from geopolitical, tariff hits, three-pillar view (Data, Operational, Technological), and a stepwise roadmap.

Watch now: youtu.be/4v1NU7__6iU

The new hot job in AI: forward-deployed engineers OpenAI, Anthropic and Cohere have stepped up hiring for specialists to help businesses adopt their AI models

When I grew up sometime I want want to be a "forward-deployed engineers"

"Artificial intelligence groups are on a hiring spree for a rare kind of software developer who can code and talk to customers, as they race to increase adoption of their cutting-edge technology"
www.ft.com/content/9100...

InfraGraph and the Return of Infrastructure Inventory — Why Cloud Needs a Single Source of Truth How a relationship-first knowledge graph could restore the system-of-record we lost moving from data centers to multi/hybrid cloud

We built the cloud to be simpler, more flexible, Yet somewhere along the way, we lost enterprise-grade inventory, a system of record that tells you what you own? where it runs? who owns it? and what policies apply?

InfraGraph promises bring it back 👇
medium.com/@saimsafder1...

YouTube video by Cloud Native Podcast #cloudnativewisdom12 HashiCorp Project Infragraph: “Google Maps” for Cloud Infrastructure?

In this ep of #CloudNativeWisdom, we dive deep into HashiCorp Project Infragraph: Google Maps for Cloud Infra?

A real-time infras graph maps &visualizes your entire cloud/infrastructure topology, helping teams see relationships, detect drift, & improve change management
youtube.com/shorts/0Tljp...

GitHub - project-dalec/dalec: 📦 Produce secure packages and containers with declarative configurations 📦 Produce secure packages and containers with declarative configurations - project-dalec/dalec

Dalec is a project aimed at providing a declarative format for building system packages and containers from those packages.

- No additional tools except docker
- Minimal image size
- Support for signed packages
- Build packages DEB-based, RPM-based & Windows containers.
github.com/project-dale...

Network = Net Worth 🚀

How to detect if we are in an AI bubble, short answer: get closer to the community.

This Saturday, I had the honor of participating as a keynote speaker for the Nexera AI event.

If done right, AI can reward creativity, systems thinking, and empathy as much as technical chops.

Practical LLM Security Advice from the NVIDIA AI Red Team | NVIDIA Technical Blog Over the last several years, the NVIDIA AI Red Team (AIRT) has evaluated numerous and diverse AI-enabled systems for potential vulnerabilities and security weaknesses before they reach production.

Insights from Nvidia Read Team about LLM attacks!
developer.nvidia.com/blog/practic...