The rise of the World Wide Web enabled developers to build tools and platforms on top of it. Similarly, the advent of large language models (LLMs) allows for creating new AI-driven tools, such as autonomous agents that interact with LLMs, execute tasks, and make decisions. However, verifying these decisions is crucial, and critical reasoning may be a solution, according to Yam Marcovitz, tech lead at Parlant.io and CEO of emcie.co.

Marcovitz likens LLM development to the evolution of programming languages, from punch cards to modern languages like Python. Early LLMs started with small transformer models, leading to systems like BERT and GPT-3. Now, instead of mere text auto-completion, models are evolving to enable better reasoning and complex instructions.

Parlant uses "attentive reasoning queries (ARQs)" to maintain consistency in AI responses, ensuring near-perfect accuracy. Their approach balances structure and flexibility, preventing models from operating entirely autonomously. Ultimately, Marcovitz argues that subjectivity in human interpretation extends to LLMs, making perfect objectivity unrealistic.

Learn more from The New Stack about the evolution of LLMs:

AI Alignment in Practice: What It Means and How to Get It

Agentic AI: The Next Frontier of AI Power

Make the Most of AI Agents: Tips and Tricks for Developers

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

Adam Jacob, CEO of System Initiative, discusses a shift in infrastructure automation—moving from writing code to building models that enable rapid simulations and collaboration. In The New Stack Makers, he compares this approach to Formula One racing, where teams use high-fidelity models to simulate race conditions, optimizing performance before hitting the track.

System Initiative applies this concept to enterprise automation, creating a model that understands how infrastructure components interact. This enables fast, multiplayer feedback loops, simplifying complex tasks while enhancing collaboration. Engineers can extend the system by writing small, reactive JavaScript functions that automate processes, such as transforming existing architectures into new ones. The platform visually represents these transformations, making automation more intuitive and efficient.

By leveraging models instead of traditional code-based infrastructure management, System Initiative enhances agility, reduces complexity, and improves DevOps collaboration. To explore how this ties into the concept of the digital twin, listen to the fullNew Stack Makers episode.

Learn more from The New Stack about System Initiative:

Beyond Infrastructure as Code: System Initiative Goes Live

How System Initiative Treats AWS Components as Digital Twins

System Initiative Code Now Open Source

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

Morgan McLean, co-founder of OpenTelemetry and senior director of product management at Splunk, has long tackled the challenges of observability in large-scale systems. In a conversation with Alex Williams onThe New Stack Makers, McLean reflected on his early frustrations debugging high-scale services and the need for better observability tools.

OpenTelemetry, formed in 2019 from OpenTracing and OpenCensus, has since become a key part of modern observability strategies. As a Cloud Native Computing Foundation (CNCF) incubating project, it’s the second most active open source project after Kubernetes, with over 1,200 developers contributing monthly. McLean highlighted OpenTelemetry’s role in solving scaling challenges, particularly in Kubernetes environments, by standardizing distributed tracing, application metrics, and data extraction.

Looking ahead, profiling is set to become the fourth major observability signal alongside logs, tracing, and metrics, with general availability expected in 2025. McLean emphasized ongoing improvements, including automation and ease of adoption, predicting even faster OpenTelemetry adoption as friction points are resolved.

Learn more from The New Stack about the latest trends in Open Telemetry:

What Is OpenTelemetry? The Ultimate Guide

Observability in 2025: OpenTelemetry and AI to Fill In Gaps

Honeycomb.io’s Austin Parker: OpenTelemetry In-Depth

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

Observability is expensive because traditional tools weren’t designed for the complexity and scale of modern cloud-native systems, explains Christine Yen, CEO of Honeycomb.io. Logging tools, while flexible, were optimized for manual, human-scale data reading. This approach struggles with the massive scale of today’s software, making logging slow and resource-intensive. Monitoring tools, with their dashboards and metrics, prioritized speed over flexibility, which doesn’t align with the dynamic nature of containerized microservices. Similarly, traditional APM tools relied on “magical” setups tailored for consistent application environments like Rails, but they falter in modern polyglot infrastructures with diverse frameworks.

Additionally, observability costs are rising due to evolving demands from DevOps, platform engineering, and site reliability engineering (SRE). Practices like service-level objectives (SLOs) emphasize end-user experience, pushing teams to track meaningful metrics. However, outdated observability tools often hinder this, forcing teams to cut back on crucial data. Yen highlights the potential of AI and innovations like OpenTelemetry to address these challenges.

Learn more from The New Stack about the latest trends in observability:

Honeycomb.io’s Austin Parker: OpenTelemetry In-Depth

Observability in 2025: OpenTelemetry and AI to Fill In Gaps

Observability and AI: New Connections at KubeCon

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

Generative AI is a data-driven story with significant infrastructure and operational implications, particularly around the rising demand for GPUs, which are better suited for AI workloads than CPUs. In an episode ofThe New Stack Makersrecorded at KubeCon + CloudNativeCon North America, Sudha Raghavan, SVP for Developer Platform at Oracle Cloud Infrastructure, discussed how AI’s rapid adoption has reshaped infrastructure needs.

The release of ChatGPT triggered a surge in GPU demand, with organizations requiring GPUs for tasks ranging from testing workloads to training large language models across massive GPU clusters. These workloads run continuously at peak power, posing challenges such as high hardware failure rates and energy consumption.

Oracle is addressing these issues by building GPU superclusters and enhancing Kubernetes functionality. Tools like Oracle’s Node Manager simplify interactions between Kubernetes and GPUs, providing tailored observability while maintaining Kubernetes’ user-friendly experience. Raghavan emphasized the importance of stateful job management and infrastructure innovations to meet the demands of modern AI workloads.

Learn more from The New Stack about how Oracle is addressing the GPU demand for AI workloads with its GPU superclusters and enhancing Kubernetes functionality:

Oracle Code Assist, Java-Optimized, Now in Beta

Oracle’s Code Assist: Fashionably Late to the GenAI Party

Oracle Unveils Java 23: Simplicity Meets Enterprise Power

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

The hardware industry is surging, driven by AI's demanding workloads, with Arm—a 35-year-old pioneer in processor IP—playing a pivotal role. In an episode ofThe New Stack Makersrecorded at KubeCon + CloudNativeCon North America, Pranay Bakre, principal solutions engineer at Arm, discussed how Arm is helping organizations migrate and run applications on its technology.

Bakre highlighted Arm’s partnership with hyperscalers like AWS, Google, Microsoft, and Oracle, showcasing processors such as AWS Graviton and Google Axion, built on Arm’s power-efficient, cost-effective Neoverse IP. This design ethos has spurred wide adoption, with 90-95% of CNCF projects supporting native Arm binaries.

Attendees at Arm’s booth frequently inquired about its plans to support AI workloads. Bakre noted the performance advantages of Arm-based infrastructure, delivering up to 60% workload improvements over legacy architectures. The episode also features a demo on migrating x86 applications to ARM64 in both cloud and containerized environments, emphasizing Arm’s readiness for the AI era.

Learn more from The New Stack about Arm:

Arm Eyes AI with Its Latest Neoverse Cores and Subsystem

Big Three in Cloud Prompts ARM to Rethink Software

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

Heroku has open-sourced its Twelve-Factor App methodology, initially created in 2011 to help developers build portable, resilient cloud applications. Heroku CTO Gail Frederick announced this shift at KubeCon + CloudNativeCon North America, explaining the move aims to involve the community in modernizing the framework. While the methodology inspired a generation of cloud developers, certain factors are now outdated, such as the focus on logs as event streams. Frederick highlighted the need for updates to address current practices like telemetry and metrics visualization, reflecting the rise of OpenTelemetry.

The updated Twelve-Factor methodology will expand to accommodate modern cloud-native realities, such as deploying interconnected systems of apps with diverse backing services. Planned enhancements include supporting documents, reference architectures, and code examples illustrating the principles in action. Success will be measured by its applicability to use cases involving edge computing, IoT, serverless, and distributed systems. Heroku views this open-source effort as an opportunity to redefine best practices for the next era of cloud development.

Learn more from The New Stack about Heroku:

How Heroku Is Positioned To Help Ops Engineers in the GenAI Era

The Data Stack Journey: Lessons from Architecting Stacks at Heroku and Mattermost

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.

Falco, an open-source runtime observability and security tool, was created by Sysdig founder Loris Degioanni to collect real-time system events directly from the kernel. Leveraging eBPF technology for improved safety and performance, Falco gathers data like pod names and namespaces, correlating them with customizable rules. Unlike static analysis tools, it operates in real-time, monitoring events as they occur. In this episode of The New Stack Makers, TNS Editor-in-Chief, Heather Joslyn spoke with Thomas Labarussias, Senior Developer Advocate at Sysdig, Leonardo Grasso, Open Source Tech Lead Manager at Sysdig and Luca Guerra, Sr. Open Source Engineer at Sysdig to get the latest update on Falco.

Graduating from the Cloud Native Computing Foundation (CNCF) in February 2023 after entering its sandbox six years prior, Falco’s maintainers have focused on technical maturity and broad usability. This includes simplifying installations across diverse environments, thanks in part to advancements from the Linux Foundation.

Looking ahead, the team is enhancing core functionalities, including more customizable rules and alert formats. A key innovation is Falco Talon, introduced in September 2023, which provides a no-code response engine to link alerts with real-time remediation actions. Talon addresses a longstanding gap in automating responses within the Falco ecosystem, advancing its capabilities for runtime security.

Learn more from The New Stack about Falco:

Falco Is a CNCF Graduate. Now What?

Falco Plugins Bring New Data Sources to Real-Time Security

eBPF Tools: An Overview of Falco, Inspektor Gadget, Hubble and Cilium

Join our community of newsletter subscribers to stay on top of the news and at the top of your game.