$ articles --latest
DevOps articles
New posts from the DevOps blogs we follow, plus articles the community submits.
latest · Platform Engineering · 45m ago · 7 min readScaling an autonomous agent from one team to fifty: A serverless-to-container decision frameworkA walkthrough of how an autonomous QA agent that files bugs from test failures would scale from one team to fifty, comparing when AWS Lambda remains the right choice against when a container-based platform fits better. It covers a decision rubric for workload characteristics, cross-team failure correlation, and bursty GPU-based bug classification.via The New Stack
Platform Engineering · 1h ago · 3 min readPerplexity’s AI agents keep working after you close the tabPerplexity's new Teammates lets multiple AI agents coordinate, resume earlier sessions, and keep working without new prompts. The article covers how shared project state is kept in Git-backed repositories, how agents are paused or forked, and why permission handling and stuck workflows remain unclear.via The New StackCI/CD · 23h ago · 7 min readHow we review 4x more AI code without reading every lineAviator describes how its nine engineers merged four times as many pull requests by having reviewers check intent, acceptance criteria, codified invariants, recorded decisions, and proof of behavior instead of reading every line of AI-generated code. The article covers what failed first, how the approach works, and the reported results, including faster merges and fewer reverts.via The New Stack
More articles
In the software factory, engineers become process engineersThe article argues that as AI agents take over writing code, engineers shift to designing the process that builds, tests, and ships software. It draws on manufacturing lessons such as Toyota's quality practices and Goldratt's theory of constraints to explain agent handoffs, built-in verification, and guardrails.CI/CD · via The New Stack · 1d ago9 minWhy AWS wants developers to stop babysitting their AI agentsAWS has launched Kiro bots, named long-running agents with their own persistent memory, scoped tools and permissions, and parallel sessions that they monitor on a developer's behalf. The article explains how the bots aim to reduce constant supervision of AI agents, and how AWS says humans remain accountable through approval controls.Cloud · via The New Stack · 1d ago7 minAI agents are getting more complex. Developers shouldn’t have to babysit them.Explains how harness engineering gives AI coding agents automated tests, feedback loops, and guardrails so they can catch and fix their own mistakes before human review. Covers Thoughtworks' Vanitha Kumar on self-correction, AI factories as a platform layer, security concerns around greater agent autonomy, and themes from Technology Radar Volume 35.Security · via The New Stack · 1d ago3 minYour Guide to the Developer Tools Track at AWS re:Invent 2026A guide to the Developer Tools track at AWS re:Invent 2026, covering must-attend sessions on CI/CD, infrastructure as code, full-stack development, and agentic development with Kiro. It also points to hands-on workshops, the Kiro Galaxy Run, the AWS Village, Builder Zone, and self-paced labs.Infrastructure as Code · via AWS DevOps Blog · 1d ago4 minCognition turned an open model into a coding beast. But it won’t release the weights.Cognition's SWE-2 Flash, a coding model built on Thinking Machines Lab's open-weight Inkling-Small, costs one-fifth as much as its larger sibling and runs about three times faster. The article covers its benchmark results, why Cognition is keeping its weights private citing cybersecurity risks, and what that means for open-model norms.Security · via The New Stack · 1d ago4 minRun Maestro tests on real Android devices using AWS Device FarmLearn how to run Maestro YAML test flows on real Android devices in AWS Device Farm using a custom test environment. The guide covers packaging flows, installing Java 21 and the Maestro CLI through a test spec, and collecting artifacts from the run.CI/CD · via AWS DevOps Blog · 1d ago8 min“Most apps never make it off localhost”: Cursor takes on Lovable and Replit with fully managed app deploymentCursor's new cloud agents can now build an app from a text prompt, preview it, and publish it to a live shareable URL with hosting, PostgreSQL databases, sign-in and automated security checks handled for the user. The article covers how the feature works, its focus on internal business tools, and concerns about whether automated checks catch complex access-control flaws.Cloud · via The New Stack · 2d ago7 minKubernetes, mark II? Why Google’s giving away gVisor, a code sandbox used by Anthropic and OpenAIGoogle is donating gVisor, Agent Substrate, and OpenRL to the Cloud Native Computing Foundation in hopes they become industry standards for AI agents, much as Kubernetes did for containers. The article covers why gVisor matters for sandboxing untrusted AI-generated code, how Agent Substrate and OpenRL fit in, and why vendor-neutral governance may speed adoption.Security · via The New Stack · 2d ago7 minDocker Sandboxes vs. Containers: Choosing the Right Isolation for AI AgentsExplains how Docker Sandboxes differ from containers for AI coding agents. Containers share the host kernel, while sandboxes are microVMs with their own kernel, filesystem, network, and Docker Engine. Covers when each fits, the setup costs, and how to choose between them.Security · via Docker Feed · 2d ago9 min
Sponsors · the tools that keep DevOpsChat freeBecome a sponsor
Oh DearMonitors website uptime, SSL certificates, DNS, cron jobs, broken links, performance and Lighthouse scores, with status pages.
UptimeRobotUptimeRobot monitors websites, endpoints, SSL certificates, ports, and cron jobs, and sends alerts by email, SMS, Slack, and more.
FathomFathom Analytics is a privacy-focused website analytics tool that shows traffic data without using cookies or tracking visitors.
MarblismMarblism provides AI employees that handle inbox, social media, SEO, lead generation, calls, and website building for businesses.