July 2026 Summaries

The Kotlin Benchmark, released by JetBrains, is a new evaluation framework designed to assess AI coding agents on Kotlin software engineering tasks, offering a more realistic measure of their performance in end-to-end tasks, from issue reading to solution validation. This benchmark aims to provide a credible, public means for developers to evaluate and compare different AI coding agents on tasks sourced from active open-source repositories, focusing on repository-level challenges and requiring agents to navigate project contexts to generate functional solutions. The initial iteration of the benchmark, which is based on the SWE-bench methodology, features 105 tasks, with the top-performing AI agent, Claude Code with Opus 4.7 xhigh, resolving 85.71% of them. JetBrains plans to expand the benchmark to include broader Kotlin ecosystem coverage, additional evaluation metrics, and more agent and model setups, thereby offering a continuous quality measurement pipeline for AI coding agents. The benchmark is built on the open-source Multi-SWE-bench infrastructure, and all datasets and test harnesses are publicly available, allowing for community engagement and feedback.

Shift-left static code analysis is a strategy in software development that emphasizes conducting testing, quality assurance, and security checks earlier in the software development lifecycle to identify and address issues as soon as possible. This approach involves automating testing processes, fostering continuous feedback, and encouraging developer ownership of code quality and security. By integrating tools like static code analysis into development workflows, teams can detect vulnerabilities and quality issues during the coding phase, which minimizes the need for context-switching and extensive rework later in the process. While implementing a shift-left strategy can present challenges such as false positives and integration friction, its benefits include faster feedback loops, reduced technical debt, improved code quality, and significant cost savings by catching defects early. This method does not eliminate traditional testing but enhances its effectiveness by ensuring that code has already undergone multiple reviews before reaching late-stage quality assurance.

The July 2026 edition of Java Annotated Monthly, curated by Irina Mariasova, offers a comprehensive roundup of the latest developments in the Java and Kotlin ecosystems, AI advancements, and notable technological trends. Highlights include Tom Cools' insights on optimizing complex problems with Timefold Solver 2.0, Felienne Hermans' keynote on cognitive programming from J-Spring, and Holly Cummins' revelations on benchmarking pitfalls. The edition also covers significant Java updates like the integration of JEP 401 into OpenJDK, tutorials on AI-enhanced Java applications, and the progress of Kotlin, now celebrating its 15th anniversary. Additionally, it explores AI's intersection with software development, offering numerous articles on AI-powered tools, frameworks, and security. Upcoming events such as WeAreDevelopers in Berlin and JConf Dominicana are spotlighted, alongside cultural discussions on platform engineering and data-driven conversations in tech. The issue concludes with updates from JetBrains and invites reader contributions for future editions.

TeamCity, a CI/CD platform by JetBrains, is evolving to address the growing integration of AI agents in software development, focusing on enhancing productivity, quality, and security. With the rise of AI-generated code, the demand for scalable, reliable, and secure CI/CD systems has intensified. TeamCity is enhancing its core features to support higher volumes of code changes and verification processes, while also developing new user scenarios to facilitate AI and human collaboration. The platform provides low-friction build configuration, faster investigation of build failures, efficient resource usage, and comprehensive visibility across the CI/CD system. JetBrains is also introducing AI governance capabilities through JetBrains Central, aiming to give organizations control, transparency, and policy enforcement over AI usage. The roadmap for TeamCity includes improving scalability, performance, and flexibility, along with expanding support for AI agents and integrations with other JetBrains products, to ensure secure and efficient software delivery in an increasingly AI-driven development landscape.

JetBrains is launching a new set of AI capabilities aimed at enhancing software development for teams and organizations by providing shared context, reusable workflows, and organization-level governance while maintaining cost control. This initiative addresses the fragmentation caused by individual developers using different AI tools, such as JetBrains IDEs, Claude Code, and Codex, by introducing a vendor-agnostic platform that connects these tools with centralized management through JetBrains Central. The rollout, starting in July, will include features like team automations, cloud agents, and JetBrains Context for improved repository intelligence, while transitioning from AI licenses to flexible AI credits for a more transparent pricing model. The goal is to facilitate scalable AI adoption that balances developer flexibility with organizational oversight, enabling engineering teams to operate efficiently without being confined to a single tool or workflow.

The research examined the reliability of coding benchmarks in assessing the capabilities of AI models, revealing a "meaning gap" between benchmark scores and actual performance across different coding tasks. Often, models show significant improvements on benchmarks they are specifically trained for, such as SWE-bench, but these gains do not necessarily translate to broader coding abilities or performance on other benchmarks like LiveCodeBench. The study suggests that current benchmarks mainly measure task-specific performance rather than general coding capability, leading to a distorted perception of a model's overall performance. The research highlights the need for more comprehensive benchmark suites and proposes solutions for better evaluation to bridge the meaning gap, urging the AI community to prioritize diverse and real-world evaluations over reliance on single benchmark scores.

Object detection is a crucial component of various applications, ranging from autonomous vehicles and security systems to medical imaging and retail analytics. In 2026, selecting the right object detection model involves evaluating different architectures, performance metrics, and use cases. Object detection requires identifying and localizing multiple objects within images or video frames, which is more complex than mere classification. Key performance metrics include mean average precision (mAP) and compute efficiency metrics like frames per second (FPS). The COCO dataset is a standard benchmark for evaluating object detection models. Models are built on CNN or transformer-based architectures, with single-stage detectors like YOLO gaining prominence for real-time applications. Leading models in 2026 include RF-DETR, renowned for its high accuracy and adaptability across domains, and YOLO12 and YOLO26, which offer strong performance on edge devices. The choice of model depends on specific requirements, balancing accuracy and real-time processing needs, with considerations for licensing and deployment scenarios. The landscape is evolving towards zero-shot detection frameworks and integrating sophisticated image embedders, expanding the potential for high-precision detection on resource-constrained hardware.

The recent update in Rider 2026.2 brings the powerful Natvis visualization tool to Linux and macOS, allowing developers to describe how their C++ types should appear in the debugger using simple XML, eliminating the need for complex data formatters previously required on these operating systems. This extension of functionality, previously exclusive to Windows, enhances the debugging experience by simplifying the representation of data structures like stacks, arrays, linked lists, and trees, making it easier to maintain and understand code. While Natvis offers broad compatibility with many existing C++ ecosystems, including game development engines like Godot and Unreal Engine, it currently relies on the standard LLDB expression evaluator on Linux and macOS, which may present challenges with more complex visualizations. However, developers can integrate Natvis with existing LLDB data formatters without conflict, allowing for a gradual adoption of this tool to improve project visibility and debugging efficiency.

In a study conducted by JetBrains testing the "Caveman" skill for AI agents, it was found that the claimed token savings of 65% were significantly overstated, with actual savings closer to 8.5% when the skill was forcibly activated. The Caveman skill, designed to make AI agents communicate more concisely by reducing token usage in non-essential text while preserving code and tool instructions, showed no detectable degradation in task quality across 82 paired tasks. Despite promising cost reductions, the anticipated savings were often negated by single-trial variances due to the nature of the tasks. The study concluded that while the Caveman skill can be fun and harmless to use, its real-world applications in coding tasks offer only minor savings, highlighting the importance of comprehensive testing over relying on initial, smaller-scale evaluations.

KotlinConf 2026 introduced the inaugural Golden Kodee Community Awards, celebrating individuals and communities that have significantly contributed to the Kotlin ecosystem through education, online presence, creativity, societal impact, and in-person engagement. The winners, including Matheus Leandro Ferreira, Jaewoong Eum, Nicole Terc, Eeva-Jonna Panula, and Yinlong Liu, exemplify the diverse strengths of the Kotlin community, with projects ranging from educational initiatives and content creation to fostering community collaboration and technological innovation. Their experiences highlight the importance of sharing knowledge and building connections within the community, emphasizing that contributions need not be grand to be impactful. The winners also expressed the personal and professional growth they have achieved through their involvement with Kotlin, illustrating the language's versatility and its supportive ecosystem. Their stories and advice encourage others to actively participate in and contribute to the Kotlin community, fostering a cycle of learning and growth.

IntelliJ IDEA 2026.1.4 has been released, offering several key fixes to enhance user experience and functionality. Users can update to this version through the IDE, the Toolbox App, or via snaps for Ubuntu, and it is also available for download from the official website. Notable improvements include the correct updating of the active Git branch, a resolution for the issue in Docker Compose files that affected PHP interpreter creation, and a fix preventing successful Gradle syncs from being marked as failed when using Gradle 9.5.0 on WSL. Additionally, an error causing Dev Container connections to fail with an "Unknown Docker endpoint schema" message has been addressed. Detailed release notes are available for a comprehensive overview of these updates, and users are encouraged to report any issues through the issue tracker.

Toolbox App 3.6 introduces significant improvements, including enhanced storage cleanup capabilities and better diagnostics for Windows installation issues. Users can now manage removable data more efficiently through the app's Settings, which provides a detailed breakdown of removable data like download caches and temporary leftovers, and can also set reminders for cleanup. The update also improves the Windows installer diagnostics by offering more precise information about issues such as file blockages and read-only errors, facilitating easier troubleshooting and log collection for support. Additionally, the CodeCanvas feature has been discontinued and removed from the app. Feedback from users is encouraged to further refine the app's functionality.

Kotlin support has been integrated into BlueJ, a significant development for one of the most widely used educational environments for teaching introductory object-oriented programming (OOP). This collaboration between JetBrains and the BlueJ team at King’s College London aims to offer students an environment that reduces syntactic distractions and boilerplate code, making it easier to focus on core programming concepts. BlueJ's visual tools, such as class diagrams and object benches, help students conceptualize programs as systems of interacting objects, and Kotlin's concise syntax further enhances this learning experience by minimizing verbosity. The integration retains compatibility with the Java ecosystem, ensuring familiar ground for educators, while offering features like null safety and simplified variable declarations to aid understanding. An onboarding guide and teaching materials have been prepared to facilitate educators' transition to using Kotlin in their curriculum, with feedback from the BlueJ community playing a critical role in refining the release.

Embedded software development faces challenges such as adhering to stringent safety and security standards, managing complex toolchains, and identifying defects early in the development cycle. JetBrains' CLion, in partnership with Parasoft C/C++test, offers an integrated solution to these challenges by enabling developers to conduct advanced static code analysis directly within their development environment. This integration supports compliance with industry standards like MISRA C/C++, AUTOSAR C++14, CERT C/C++, and CWE, and facilitates a seamless workflow by reducing context switching between development and verification tasks. Developers can utilize AI-assisted capabilities to gain insights into coding violations and receive suggestions for remediation, which is particularly beneficial when dealing with intricate standards. The system creates a continuous feedback loop by allowing Parasoft C/C++test analyses to be reused in CI/CD pipelines, ensuring consistency in verification throughout the software lifecycle. This integration is crucial for teams in industries such as automotive, medical devices, rail, aerospace, and industrial automation, where compliance and traceability are non-negotiable requirements.

TeamCity has released two updates for its On-Premises servers: version 2026.1.2, which addresses over ten issues, including build command cancellations, Perforce changelist problems, AWS credential expirations, and AWS S3 signature mismatches, and version 2025.11.6, which focuses solely on resolving security issues. Users are encouraged to install these updates to enhance performance, integration compatibility, build stability, and security. The 2026.1.2 update is compatible with other 2026.1.x releases, allowing for easy upgrades without backup concerns, and users can access updates via automatic updates, direct downloads, or Docker images. For support, users are directed to the TeamCity Forum or Issue Tracker.

JetBrains' engineering hiring process is a structured and comprehensive approach designed to evaluate not just technical skills but also the candidate's ability to think critically about problem-solving and understand the broader impact of their work. The process includes several stages: an initial CV review focusing on clarity and relevance, a recruiter call to assess alignment on skills and motivations, a team interview involving technical exercises and discussion of past projects, and a take-home task that reflects real-world challenges. The emphasis is on clear communication, decision-making, and understanding trade-offs. Successful candidates are those who demonstrate sound judgment, clean code, and the ability to explain their reasoning. The final stages involve task review, reference checks, and, if all aligns, an offer. JetBrains seeks engineers who are passionate about their craft and eager to contribute to meaningful projects, with the hiring process reflecting a commitment to building long-term, impactful teams.