Signals about AI expectations, JVM evolution, observability at scale and how modern software systems are changing.
🎯 Featured Article
📰 The Internet’s Most-Read Tech Publications Have Lost 58% of Their Google Traffic Since 2024
Major technology publications have collectively lost tens of millions of monthly search visits since 2024. The data highlights how AI summaries, social discovery and direct channels are reshaping how developers consume technical content.
⬢ Analyze search traffic collapse across major tech outlets like CNET, Wired and The Verge.
⬢ Highlight discovery shifts as AI assistants and alternative channels replace traditional SEO-driven readership.
⏩ TL;DR (Quick Recap)
⬢ Reset expectations around AI productivity, which appears closer to ~10% gains than the widely promised 10×.
⬢ Adopt stronger observability practices as AI-driven applications require tracing, reasoning visibility and evaluation data.
⬢ Follow JVM ecosystem evolution with Project Leyden, Kotlin AI tooling and performance improvements in JDK 25.
⬢ Revisit engineering fundamentals such as code review purpose, error handling strategies and architecture choices.
☕ JVM Corner
🧪 Introducing Tracy: The AI Observability Library for Kotlin
A Kotlin library designed to bring production-grade observability to AI-powered applications.
⬢ Trace model and tool calls to understand how LLM-driven workflows execute in production.
⬢ Collect evaluation and performance data for debugging, benchmarking and monitoring AI systems.
⚙️ SQG v0.10.0: Java Streams & List Type Support
SQG generates strongly typed database access code from annotated SQL queries.
⬢ Generate compile-time safe data access layers directly from SQL definitions.
🚀 How we integrated Project Leyden into Quarkus
The Quarkus team explores integrating Project Leyden to improve startup time and runtime performance.
⬢ Demonstrate practical integration strategies for modern JVM frameworks.
⚡ Java Performance Update: From JDK 21 to JDK 25
JDK 25 introduces numerous performance improvements across the JVM runtime and libraries.
⬢ Deliver faster execution for existing applications without code modifications.
🍃 Spring Updates
🏗️ Builder Pattern with Spring Boot
A practical guide to applying the Builder pattern in Spring Boot applications.
⬢ Improve readability and maintainability when constructing complex objects.
🤖 AI-powered applications with Kotlin and Spring AI
A hands-on tutorial demonstrating how to build AI applications using Kotlin and Spring AI.
⬢ Build practical AI-driven services within the JVM ecosystem.
🧠 Explainable AI Agents: Capture LLM Tool Call Reasoning with Spring AI
Explainable AI agents add visibility into how models choose tools during execution.
⬢ Capture reasoning context behind LLM tool selection decisions.
⬢ Improve debugging, observability and trust in AI-powered agents.
🔍 Extra Reads
AI Engineering & Developer Productivity
🤖 The 8 Levels of Agentic Engineering — A model describing how teams evolve from simple AI-assisted coding to fully agentic engineering workflows.
🧠 Your LLM Doesn’t Write Correct Code. It Writes Plausible Code. — Explains why developers must define acceptance criteria before relying on AI-generated code.
📊 AI productivity gains are 10%, not 10x — Real-world engineering teams report modest but measurable productivity improvements from AI tools.
🧩 The End of Coding? Wrong Question — Argues that AI changes developer workflows rather than eliminating programming.
Engineering Practices & Architecture
🧑💻 What Is Code Review For? — Reframes code review as a design collaboration tool rather than a bug-catching mechanism.
🤝 How to Do Code Reviews in the Agentic Era — Applies zero-trust contribution models to AI-generated code.
⚠️ The two kinds of error — Distinguishes between expected runtime errors and unexpected bugs.
Observability, Security & Infrastructure
📡 Running OpenTelemetry at Scale: Architecture Patterns for 100s of Services — Architectural patterns for scaling telemetry pipelines across large microservice environments.
📊 OTTL context inference comes to the Filter Processor — Adds context-aware filtering capabilities to OpenTelemetry collectors.
🔐 DPoP: What It Is, How It Works, and Why Bearer Tokens Aren’t Enough — Introduces proof-of-possession tokens as a stronger alternative to bearer authentication.
🕵️ I Audited the Privacy of Popular Free Dev Tools — The Results Are Terrifying — Demonstrates how widely used developer tools may leak sensitive data.
Platforms, Data & Industry
🗂️ How we made Notion available offline — Describes a system that tracks multiple reasons why pages remain cached for offline use.
📺 Scaling Global Storytelling: Modernizing Localization Analytics at Netflix — Explains how Netflix modernized analytics pipelines for global localization workflows.
💻 Big Data on the Cheapest MacBook — Benchmarks show entry-level hardware can still handle serious analytical workloads.
🧩 Returning To Rails in 2026 — A developer revisits Rails and reflects on building side projects for exploration and learning.
Design, Privacy & Ecosystem Trends
🌱 A Designer’s Guide To Eco-Friendly Interfaces — Sustainable UX promotes reducing unnecessary visual and computational complexity.
🔒 Pi-hole behind Tailscale — Combines Pi-hole and Tailscale to improve network-level privacy protection.
🧱 Is the Atlassian Ecosystem Starting to Crack? — Analyzes structural tensions within the Atlassian ecosystem.
Originally posted on marconak-matej.medium.com.