Hello!

GitHub's rolling out their new Copilot app, promising an agent-native desktop experience that's piquing everyone's interest. Meanwhile, Lenny's Newsletter dives into Claude Opus 4.8 (is it all it's cracked up to be?), and Red Hat's NPM channel faces a security headache with dozens of packages backdoored. It’s a busy week in tech.

At Pagepro, we’re all about smooth transitions, specializing in Next.js and Sanity migrations for enterprises that need their digital presence to just work.

Grab your coffee, settle in, and enjoy Frictionless.

React can feel sluggish. Animating data at 60 FPS is possible without third-party libraries. This slide deck shows you how — think Canvas, requestAnimationFrame, WebSockets, and React.memo.

1. Ditch libraries: Use native APIs like Canvas and requestAnimationFrame for smoother animations. You’ll streamline your code and reduce dependencies.
2. Optimize state management: Implement React.memo to prevent unnecessary re-renders and improve performance.

Ready to make React dance without the lag?

Why juggle two caches when you can streamline? This week, Dominik takes a deep dive into pairing TanStack Router with TanStack Query.

The router's cache works on a per-route basis, but Query's global cache shines when data is shared across routes.

He reveals how to wire the QueryClient into your router context and disable the router’s cache, letting one library handle the data layer.

Only 15% of engineers fully utilize their caching libraries.

Are you part of this minority?

Automation isn’t a cure-all. Radiology learned that the hard way — tech leaders should take note. Automating processes can speed things up, but without human oversight, you risk missing crucial errors.

Big Tech faces a similar conundrum. A staggering 90% of radiology errors stem from misinterpretations. This isn’t just about machines making mistakes. It's about missing the human touch in complex situations. CTOs, don't ditch your human checks in favor of shiny new AI. Your team’s expertise is still your strongest asset.

Guardrails aren’t optional — they’re essential.

Keep your processes human-centered to maintain quality and trust.

Small codebases are powerful. This isn't about skimping on features; it's about strategic restraint. Fewer lines of code mean fewer bugs, lower maintenance costs, and a happier team. Why? Because clean, concise code frees up time and mental energy for creative engineering.

Value isn't volume; it's impact.

How often do you encourage your team to solve a problem with elegance rather than overengineering?

Next team meeting, challenge them to delete more lines than they add. Think your engineers can rise to that?

Pulling the plug on a project is never easy, but sometimes it's the smartest move. You need to recognize the signs before sinking more resources into a dead-end. Two crucial indicators: a dwindling ROI and a demoralized team. Ignoring these can turn a minor setback into a catastrophe.

Cut your losses before they cut your team. It's not about failure — it's about prioritizing what truly matters.

Meet Varticas: your new AI coworker. It's not just another tool; it automates workflows and delegates tasks to AI agents. Imagine cutting down mundane tasks without lifting a finger. Varticas empowers you to execute real work with autonomous AI agents, which means more time for strategic decisions.

The real kicker?

It handles tasks end-to-end, freeing your team from repetitive chores. Are you ready to trust AI with your daily grind? Dive into the world of Varticas and see for yourself what's possible.

Anthropic's Claude Opus 4.8 promises sharper judgment and longer independence, sticking to the same price. But does it deliver? Early tests reveal it's a powerhouse for greenfield prototypes and fast execution. However, it fumbles at the last 10% of tasks and in navigating existing codebases.

Hallucination issues persist, echoing its predecessors' hiccups.

Two takeaways:

1) Use 4.8 for rapid prototyping where speed trumps perfection,
2) For data-heavy strategy, Opus 4.7 remains the go-to.

The real question: Is marginal improvement worth jumping ship, or does sticking with 4.7 make more strategic sense?

Drowning in context switches? GitHub's new Copilot app might be your lifeline. It's the control center you didn't know you needed, bringing all your agent-driven work into one cohesive view.

Forget juggling windows or tracing random commit trails; now it's all laid out in a single 'My Work' dashboard. This means less time hunting for context and more time doing what actually matters: productive coding.

Commits are soaring to 1.4 billion a month, showing the growing reliance on these agentic workflows.

It’s available in a technical preview for Copilot subscribers. Ready to streamline your chaotic agent affair?

Here's a plot twist: Red Hat's trusted NPM channel got hijacked to spread malicious worms. A threat actor took control of the @redhat-cloud-services namespace and snuck in over 30 booby-trapped packages. This mess has developers questioning the reliability of a name they've always trusted. Imagine a thief with your house keys. That's what these compromised credentials mean for Red Hat.

Security fumbles like this open the door to a cycle of vulnerability that's anything but trivial.

Actionable takeaway: Vet your supply chain dependencies and tighten access controls—your company's reputation is only as secure as its weakest link.

Thinking your systems are unassailable? Think again. Who's guarding your digital front door?

Do you have any comments about this newsletter issue or questions you want to ask? Drop me a message or book a meeting.