Kubernetes teams auto-deploy code but refuse AI CPU tweaks

A survey of 321 enterprise Kubernetes engineers finds 82% trust automated deployments, yet 71% demand human review before resource‑optimisation recommendations, and only 27% allow auto CPU or memory changes. AI inference workloads amplify the risk, turning over‑provisioning into expensive waste and rightsizing mistakes into hidden outages.

Nx launched Polygraph, a free‑access service that merges any number of repositories into a single synthetic monorepo, giving AI coding agents a unified view of code. By exposing cross‑repo dependencies and preserving agent interaction history, Polygraph lets agents continue work across repository boundaries, aiming to boost the modest speed‑up seen in large teams.

GitLab’s survey of 1,500 developers shows 80% adopted AI tools faster than they built policies, exposing reliability, security and cost risks. The gap demands "agentic infrastructure", machine‑scale execution, context, governance and orchestration, to keep AI‑driven code production under control.

The KEDA external scaler reads NVIDIA NVML metrics directly from a DaemonSet, eliminating the dcgm-exporter → Prometheus → KEDA pipeline. It trims metric lag to 2‑4 seconds and lets inference services like vLLM or Triton scale to zero without extra components. Simpler ops, faster reactions for GPU workloads on Kubernetes.

Announcing the public preview of AKS on bare metal lets Kubernetes run directly on edge hardware without a hypervisor, using a USB‑driven zero‑touch install that stays operational during connectivity loss. The same AKS tooling, RBAC and monitoring apply, giving teams a unified platform from cloud to on‑premises.

Tempo 3.0 makes the tracing backend default to a single‑copy, Kafka‑compatible storage layer, eliminating the previous three‑copy replication and slashing object‑storage expenses by roughly two‑thirds. It also graduates TraceQL metrics to GA, letting users derive real‑time metrics straight from traces while boosting query speed and reliability.

Microsoft just moved the Azure SDK for Rust from beta to stable, shipping six service crates, Core, Identity, Key Vault (Secrets, Keys, Certificates) and Storage (Blobs, Queues), with guaranteed semver stability and built‑in retries, tracing and async support. This gives Rust developers a production‑ready, idiomatic way to consume Azure services, shrinking the gap between high‑performance code and cloud APIs.

Cursor has acquired the open-source AI coding assistant Continue, merging its 34K-star codebase into the Cursor ecosystem. The deal hands the project to the community while giving Cursor a proven alternative to GitHub Copilot, tightening competition in AI-assisted developer tools. Expect faster feature rollouts and tighter integration with Cursor’s multi-model workflow.

Starting July 20, GitHub Code Quality leaves beta and becomes a paid service at $10 per committer per month, plus metered AI analysis and GitHub Actions minutes. The GA adds dashboards, coverage gating and API access, so teams can enforce maintainability and reliability in PRs. Expect a new line‑item on your bill and decide to enable or disable before the deadline.