From frontier models to structured concurrency, tool use, code quality and team design.
🎯 Featured Article
🧭 Our decision on Cursor following its acquisition by SpaceX OpenAI plans to wind down its model contract with Cursor after SpaceX’s acquisition.
- OpenAI proposes ending model access through Cursor on November 12, 2026.
- The decision centers on contractual compliance and future model access.
⏩ TL;DR (Quick Recap)
- Frontier AI models are expanding from coding into autonomous professional workflows.
- Agent reliability increasingly depends on tools, constraints, safeguards and review.
- Java and Spring continue evolving toward simpler concurrent and AI-native applications.
- Team structure, culture and repository architecture still shape engineering productivity.
☕ JVM Corner
🧵 JEP draft: Structured Concurrency A draft JEP proposes structured concurrency to simplify task coordination, cancellation, error handling and observability.
- Related concurrent tasks are treated as a single unit of work.
- Structured lifecycles improve cancellation and error propagation.
📚 Modern Java: An In-Depth Guide from Java 8 to Java 21 A practical tour of Java’s evolution from lambdas and streams in Java 8 to records and sealed classes in Java 21.
🧩 The Grails Plugin Has a New Home: Apache Grails — The IntelliJ IDEA Grails plugin has moved under Apache Grails ownership and stewardship.
🍃 Spring Updates
🤖 Using Anthropic’s Claude Models With Spring AI Guide to integrating Anthropic’s Claude models with Spring AI for multimodal, conversational applications.
- The tutorial builds a chatbot using Claude through Spring AI.
- The example covers textual and visual inputs plus multi-turn conversations.
🛠️ Spring AI Recipe: Efficient Tool Use ToolSearchTool lets agents discover relevant tools dynamically instead of sending every tool definition to the model.
- Dynamic tool discovery reduces unnecessary tool definitions in prompts.
- Selective tool exposure can make agent interactions more focused.
🏠 Spring AI Recipe: Running Against Local LLMs Spring AI can connect applications to local models through tools such as LM Studio, Docker Model Runner and Ollama.
- Local model providers can replace cloud-hosted LLMs in Spring AI.
- The approach keeps model execution closer to the application environment.
🔐 Broadcom pledges to lock down open source Python, Java libraries
Broadcom plans additional security measures around key open-source projects including Spring and RabbitMQ.
- The initiative targets software used across Broadcom’s Tanzu ecosystem.
- Spring and RabbitMQ are among the projects receiving increased security attention.
🔍 Extra Reads
AI Models & Agents
🧠 Claude: Fable 5.1 and Mythos 5.1 — Anthropic introduces two safeguard profiles for the same model across general and specialized research use.
🖥️ GPT-6 Astra — OpenAI’s new model targets complex coding, computer use, research and multi-step professional workflows.
⚙️ Agentic Code Quality — Reliable agent-generated code increasingly depends on the constraints and controls surrounding the agent.
🔎 Three sites made 215,128 “best software” pages for AI. Perplexity cites them — A study examines how machine-oriented content sites influence grounded AI recommendations.
Engineering Quality & Productivity
🧪 Vibe coding and quality — AI-assisted development still requires engineers to understand architecture, abstractions, events and data flow.
🌱 Good Culture is the Biggest Productivity Hack, Not AI — AI productivity gains depend on an engineering culture capable of using those tools effectively.
📈 20 Lessons From 20+ Years in Tech — Twenty lessons from a long technology career focus on leverage across career, growth and life.
👥 Three People Is a Team. Seven Is a Committee with Zero Decisions. — Communication paths grow rapidly with team size, making organizational structure a key product-development constraint.
Platforms & Developer Infrastructure
🌐 Git at any scale — Large-scale Git hosting exposes difficult challenges around repository design, reliability, storage and distribution.
⚡ htmx 4.0.0 has been released! — htmx 4 introduces explicit attribute inheritance, standardized events and a more.
🏗️ Building a simpler, faster Uber — Uber is simplifying organizational structures, reducing layers and cutting its workforce by about 10%.
Originally posted on marconak-matej.medium.com.